Professional Documents
Culture Documents
Electric Machine Analysis and Simulation Abstract | Full Text: PDF (272K)
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...RONICS%20ENGINEERING/22.Energy
%20Conversion.htm (1 of 3)17.06.2008 16:33:30
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...RONICS%20ENGINEERING/22.Energy
%20Conversion.htm (2 of 3)17.06.2008 16:33:30
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...RONICS%20ENGINEERING/22.Energy
%20Conversion.htm (3 of 3)17.06.2008 16:33:30
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3010.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Air Pollution Control
VA
❍ Advanced Product
Copyright © 1999 by John Wiley & Sons, Inc. All rights
Search
reserved.
❍ Search All Content
DOI: 10.1002/047134608X.W3010
❍ Acronym Finder
Article Online Posting Date: December 27, 1999
Abstract | Full Text: HTML PDF (185K)
Abstract
Effects of Emissions
Conclusions
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3010.htm17.06.2008 16:34:26
GLOBAL
ELECTRICITY GENERATION
AND THE
ENVIRONMENT
The Role
of Electricity in Supplying Clean Energy
1993 1985–1993
1995 2000 2005
Generation Growth
Estimatea Projection Projection
Region/Country TW · h (%)
TW · h TW · h TW · h
Fraction (%)
Energy
Hydro/
Country/Region (TW · h) Coal
Oil Gas Nuclear Others
S + O → SO
tion of emissions begins with the calculation of the amount of
2 2
fuel used based on the amount of electricity generated. One
method of calculation uses the average heat rate of the plant, The
molecular weight of SO2 is 32 # 2 # 16 # 64. Thus, com-
electricity generation in gigawatt-hours (GW #h), and the heat plete
combustion of one ton of sulfur produces 64/32 # 2 tons
content of the fuel; thus, of
SO2 . For one ton of coal having 3% sulfur content, the
2
The amount of fuel-bound N2 converted to NOx during com-
= AP + BP + C (MBtu) (2)
bustion varies between 15% and 20% of the nitrogen in the
heat content of fuel (MBtu/ton) coal.
It depends on the characteristics of the coal, the firing
N2 + 2O2 → 2NO2
1200
atmospheric phenomenon as well. The terrestrial biosphere
1088
and
the oceans serve as sinks and sources for carbon dioxide,
800 907 each
exchanging with the atmosphere some 100 gigatons of
) carbon
(GtC) per year.
h 600
W The
OECD estimated anthropogenic CO2 emissions from
k
/ 590 energy
use in 1988 at 6.3 GtC per year, growing at the rate
g 400
(
2 of
2.5% per year (6). About 1.7 GtC was contributed by power
O 1200
generation (10). Half of the CO2 emitted appears to remain in
C
the
atmosphere and causes the rising trend in air concentra-
200 tions
of the gas.
4
Figure 1. Average CO2 emissions in the US in 1990 from fossil-fired (N O),
ozone (O ), and the chlorofluorocarbons (CFCs). All of
2
3
power plants (8). these
have a greater warming effect than CO2 on a molecule-
for-
molecule basis. Calculations based on concentration, radi-
The amount of thermal NOx is dependent on combustion- ative
effectiveness, and lifetime, however, show that CO2 ac-
time–temperature factors, furnace combustion conditions (re-
counted for 66% of greenhouse gas contributions to global
ducing or oxidizing conditions), and the type of burner and
warming between 1880 and 1980 (11), and 55% between 1980
combustion air distribution. and
1990 (9).
Figures 1 and 2 show average CO , SO , and NO emissions
While models predict that the increase in greenhouse gas
2 2 x
slightly less than half a degree Celsius (#C) [one degree Fahr-
enheit
(#F)] can be attributed to them. The models differ in
EFFECTS OF EMISSIONS
the
assessment of the effects of emission control laws on the
cost
of operating electric utilities.
Sulfur
Dioxide
8 7.7
SO2
) 7 NOx Sulfur
dioxide is a colorless gas produced mainly from the
h 6
combustion of fossil fuels. It can react catalytically with such
W 5.4
k
oxidants as ozone, hydrogen peroxide, and organic free radi-
/ 5
g
(
s 4 4 cals
to produce sulfur trioxide (which hydrates quickly to sul-
n furic
acid), sulfuric acid, and sulfates:
o
i 3
s
s 2.1
i 1.9
m 2
OH + SO → HSO
E
2 3
1
0
HSO3 + O2 → HO2 + SO3
0
Coal Oil Gas
SO + H O → H SO
3 2 2 4
Emissions
1995
2000
the elderly, the young, and those with respiratory ailments. cially
such textiles as nylon), leather, paper, electrical con-
There are also effects on the vegetation, on structural materi- tacts,
paints, and medieval stained glass. It accelerates the
als, and on the atmosphere. These effects are described in the corrosion
rates of metals such as iron, steel, zinc, copper,
following paragraphs. The reader can refer to Elsom (13) for and
nickel.
a more thorough discussion of the effects of SO2 and NOx Another
effect of SO2 emissions is the reduction in visibil-
emissions. ity due
to light absorption and scattering by sulfates. This
The best known example of the health effects of SO2 hap- suspended
particulate matter also enhances condensation and
pened in London in 1952, when 4700 deaths occurred above freezing,
leading to the formation of cloud and fog, increased
the expected number on account of respiratory failure. The
precipitation, and reduced sunshine levels.
largest single contributor was bronchitis, although death from
diseases involving the impairment of respiratory functions
Nitrogen
Oxides
also rose. The number of deaths due to heart disease in-
creased, possibly due to the strain on the heart caused by Oxides of
nitrogen account for 30% of acid rain precipitation
those respiratory problems. Sulfates suspended in the emis- in the US
(next to sulfur compounds, which account for 65%).
sions are suspect for increased asthma attacks, aggravation Acid rain
(precipitation having a pH lower than 5.6) is known
of heart and lung disease, and a lowering of resistance to re- to result
in the depletion of fish stocks, a reduction in forest
spiratory diseases.
productivity, human health problems, increased material cor-
Low added levels of SO2, have been seen to enhance plant rosion and
erosion, and reduced visibility.
growth by the addition of soil nutrients. At higher levels, how- The
most important oxides of nitrogen are nitric oxide and
ever, SO2 causes bleaching of plant chlorophyll and lowers soil nitrogen
dioxide; the other oxides are not known to be biologi-
pH values, resulting in reduced growth and yield. Effects are cally
significant. Anthropogenic nitrogen oxides are produced
more pronounced on plants such as lichen that contain rela- during
combustion when the temperature is higher than
tively little chlorophyll. Such information has been used to about
1000#C. Its principal sources are the combustion of fos-
map sulfur dioxide levels by experimental transplantation of sil fuels
in stationary sources and in motor vehicles. Like sul-
lichens and observing their fate. fur
dioxide, nitrogen oxide emissions have effects on health,
Sulfur dioxide also leads to corrosion of building stone. It on
vegetation, and on the atmosphere. In addition, they are
converts the calcium carbonate in limestone, sandstone, believed
to act as catalysts in the depletion of the ozone layer.
roofing slate, and mortar into soluble calcium sulfate (gyp- Aside
from promoting photochemical pollution, nitrogen
sum). The material increases in volume, resulting in scaling, oxides
have health effects of their own. High concentrations
of NO
3 3
blistering, and disintegration. The loose material is then x
(600 #g/m ) to 900 #g/m ) have been found to result in
washed away by rain. Sulfur dioxide also affects fabrics (espe- increased
susceptibility to respiratory infections, increased
combustion.
ulation and policy signals from the government.
stream. This direct contact between the slurry and the SO2-
ted as SO2 . Since the constant churning facilitates the trans-
laden gas removes the SO2 from the flue gas through the fol-
fer of heat to the tubes, the temperature of combustion can be
3 4
2 3
MgSO + H O + SO → Mg(HSO )
lime (CaO) as reagent instead of limestone. Mixed with water,
2 2 2 3 2
this produces calcium hydroxide [Ca(OH)2] or slaked lime.
(primary absorption reaction)
The calcium hydroxide reacts with SO2 in the scrubber, ab-
MgSO3 + 1 O2 → MgSO4
sorbing it from the flue gas with the following chemical reac-
2
4 2 4 2
Ca(OH) + SO → CaSO + H O
Mg(HSO ) + M(OH) + 4H O → 2 (MgSO · 3H O) (s)
2 2 3 2
3 2 2 2 3 2
2 2
soda ash (Na CO ) and lime (CaO) are fed to the system, pro-
2 3
2 3
The reaction between the soda ash and the SO2 is given
As in the wet limestone FGD process and for similar reasons, by
the following chemical balance equation:
Na CO + 2H O + SO → Na SO + CO + 2H O
Wellman–Lord Process. This consists of two stages. The
2 3 2 2 2 3 2 2
x
crystals with a small amount of coke under controlled condi- takes
advantage of the above characteristics of NOx produc-
tions produce solid MgO and an offgas of 7% to 9% SO2 . The tion.
The popular technique is to design combustion chambers
regenerated MgO goes back to the absorbent slurry prepara- and
burners for staged combustion—one under a fuel-rich
tion area, while the SO2 is converted to sulfuric acid in a con-
mixture and the other under a fuel-lean mixture. The initial
ventional contact acid plant. stage
involves partially burning coal in reduced atmosphere,
The following chemical reactions occur in the magnesia
allowing it to retain part of the fuel-bound nitrogen in the
slurry process:
residual unburned fuel. The burned fuel from the initial burn-
Demand-Side Technologies
air between the fuel-rich mixture in the initial zone and the
secondary combustion zone. This results in a reduction in the
Improved generation efficiencies reduce emissions by reduc-
range of 40% in conversion of fuel-bound nitrogen to NOx . ing
coal burn per unit of energy produced. In addition, the
Fuel Reburning. Another method being studied uses fuel re-
utility can avail itself of demand-side alternatives that delay
burning for chemically reducing the NO formed in the com- the
requirement for additional investments.
bustion stage. It involves burning the fuel in the lower fur-
Three demand-side management (DSM) load profile objec-
nace under slightly air-rich conditions. Additional fuel is then
tives utilities may consider are peak clipping, valley filling,
added downstream of the first zone to reduce most of the NO and
load shifting (21). Peak clipping reduces peak demand
generated in the first stage and convert it to N2 under slightly
without affecting off-peak demand. This strategy is effective
substoichiometric conditions. Tertiary air is then added to the in
reducing operating costs and postponing the need for addi-
fuel-rich products to complete the combustion under slightly
tional capacity. Valley filling increases off-peak demand,
air-rich conditions.
which improves the efficiency of utilizing existing generating
Other Options. The following options for limiting NOx
capacity, thereby reducing costs of delivering power. Finally,
emissions are also available: as
the name implies, load shifting moves energy consumption
postponed for a few years can be fitted with the latest technol-
from the oxidation of carbon in the fuel. Coal has the highest
ogies for emissions control, which will pollute less than the
carbon content among the fuels and produces about 32%
tors can also be used to reduce the total electric energy con-
lowing:
sumption (23).
1. Use a sorbent called selexol to recover CO from the gas-
Among the appliances that have enjoyed efficiency up-
2
ified coal in IGCC.
grades are refrigerators. The efficiency improvement poten-
burden
more effectively, the US government is now taking
Command and Control. A command-and-control approach
advantage of economic-incentive approaches. These policy op-
to pollution abatement involves the setting of standards and tions
use market forces to find the most cost-effective manner
direct regulation of polluters. This conventional regulatory of
pollution control. Economic-incentive approaches can be
mechanism either specifies the technology that must be used grouped
into four major categories: (1) pollution charges, (2)
(technology-based) or sets a cap on the emission rate that all
subsidies, (3) deposit-refund systems, and (4) market cre-
sources must meet (a uniform performance standard). The ation.
command-and-control approach has included the setting of
Pollution Charges. Pollution charges take several forms. Ef-
ambient air and water quality standards, objectives, and tar- fluent
charges can be collected on discharges into the environ-
gets, and the imposition of emission and discharge limits and/ ment
based on the quantity and/or quality of the effluent.
ment
utilities include the creation of a category of exempt whole-
sale power producers. This change allows utilities to operate 2.
Limited technical assistance in the form of workshops,
projects.
domestic commitment to return US greenhouse gas emissions
The
second approach is through a set of industry initia-
to 1990 levels by the year 2000. DOE estimates that these
tives
that utilities are engaging in collectively. These initia-
accords will reduce GHG emissions by 47 million metric tons
tives
currently include five Edison Electric Institute (EEI) in-
of carbon equivalent from expected year 2000 levels.
dustry
initiatives and five being conducted by other utility
Underlying the climate challenge program are two key trade
associations and utilities.
framework instruments. The first of these instruments is the
The
following are EEI’s five major initiatives (31):
Climate Challenge Program Memorandum of Understanding
(MOU) which was signed on April 20, 1994. This established
1.
The National Earth Comfort Program, which is a con-
the fact that the parties involved had agreed on a voluntary
sortium of more than 150 electric utilities and allied or-
partnership to reduce greenhouse gas emissions, and set forth
pump market
DOE–utility agreements to follow. 2.
EnviroTechSM, which is an investment fund created to
The second instrument was the Model Climate Challenge
1.
Electric End-Use Efficiency Technology Initiative
One or more of the following specific commitments are ex- 2.
Tree Power
pected of utility participants:
3.
Partnerships in Joint Implementation Projects
4.
International Donated Equipment Initiative
1. Make a specified contribution to particular industry ini-
5.
Combined Purchasing Initiative
tiatives
1988
levels.
Promotion of a Best-Generating-Mix Effort that Centers on
Denmark expects to meet the goals outlined in the FCCC
Nuclear Power. This involves increasing the proportion through
legislation tailored to (1) improve end-use energy ef-
of nuclear power, new energy, and nonfossil energy ficiency
and conservation, (2) enhance efficiency in energy
sources in relation to others. It also includes promoting supply,
(3) increase use of environmentally benign energy
the introduction of liquefied natural gas (LNG) and sources,
and (4) promote research and development. The Dan-
other fossil fuels with a low carbon content. To this end, ish
Parliament also included a CO2 tax in its plan as partial
TEPCO targets an increase of two percentage points
incentive to decrease such emissions.
each in the share of nuclear, hydro, and LNG genera-
McGowan (35) suggests that the barrier of politics is the
tion for the year 2000 over that of the year 1994. most
important one to surmount in the case of renewables.
The
Danish energy policy program is an example of how poli-
Improvement of the Thermal Efficiency of Thermal Power
tics,
legislation in particular, can spur progress toward envi-
Generation. TEPCO successfully improved the thermal
13.
D. M. Elsom, Atmospheric Pollution: Causes, Effects and Control
CONCLUSIONS
Policies, New York: Basil Blackwell, 1987.
14.
W. Fulkerson, R. R. Judkins, and M. K. Sanghvi, Energy from
Air pollution from electricity generation can be addressed at
fossil fuels, in Readings from Scientific American: Energy for
all stages of the energy cycle—from resource planning
Planet Earth, New York: W. H. Freeman, 1991, pp. 83–94.
through operations. Pre- and postcombustion technologies are 15.
H. G. Stoll, Least-Cost Electric Utility Planning, New York: Wi-
sufficiently mature, and the industry is nearing transition to
ley, 1989.
highly efficient next-generation combustion technologies. 16.
Organization for Economic Cooperation and Development, Envi-
However, the electric utility industry remains dependent on
ronmental Effects of Electricity Generation, OECD, 1985, chap. 2,
fossil fuel for electricity generation.
pp. 42–53.
As the consumption of energy, electrical or otherwise, in- 17. R
H. Shannon, Handbook of Coal-Based Electric Power Genera-
creases, the growth of CO2 emissions continues irrespective of
tion, Park Ridge, NJ: Noyes Publications, 1982.
conventions, treaties, and debate in the public media. Even 18.
R. Clement and R. Kagel, Emissions from Combustion Processes:
after many of the industrialized countries committed to re-
Origin, Measurement and Control, Chelsea, MI: Lewis Publish-
duce their CO2 emissions under the FCCC, most of them
ers, 1990.
failed to achieve their own goals. There is scientific evidence 19.
The Atmospheric Research and Information Centre (ARIC), Tech-
that global warming is unavoidable. However, the degree of
nologies for Reducing Acidic Emissions from Power Stations,
this warming can be slowed down if prudent action is taken
http://www.doc.mmu.ac.uk/aric/techno.htm, June 24, 1997.
by all parties, but most importantly, by the ones who can af- 20.
US Congress, Office of Technology Assessment, Changing by De-
ford the most and have some historical obligations due to
grees: Steps to Reduce Greenhouse Gases, OTA-O-482, Washington,
their past emission practices.
DC: US Government Printing Office, February 1991.
It is true that a few developing countries will gradually 21.
T. W. Keelin and C. W. Gellings, Impact of Demand-Side Manage-
surpass some of the highest CO2 emitters of the late twentieth
ment on Future Customer Electricity Demand, EPRI EM-4815-SR,
century, and the level of atmospheric CO2 concentration af-
Palo Alto, CA: Electric Power Research Institute, October 1986.
fects all countries. It is, however, judicious for the industrial- 22.
Customer Systems Division, Technical Brief: Load Management
ized countries to take unilateral actions to reduce their CO2
and the Environment, Report RP2788, Electric Power Res. Inst.,
emissions, at least so that they can take the moral high
1991.
ground in the debate on who has been responsible for global 23.
M. D. Levine et al., Electricity end-use efficiency: Experience with
warming and who will make how much sacrifice to reverse
technologies, markets, and policies throughout the world, Energy
the trend.
Int. J., 20 (1): 37–61, 1995.
24.
Project 88—Round II, Incentives for Action: Designing Market-
8. Acid Rain Division, Energy Efficiency and Renewable Energy: Op- 31.
W. L. Fang, The US climate challenge program voluntary govern-
portunities for Title IV of the Clean Air Act, Washington DC: Envi-
ment-utility partnerships to mitigate greenhouse emissions,
ronmental Protection Agency, February 1994.
IEA/OECD Workshop on Voluntary Approaches for Mitigating
9. N. Sundararaman, Energy and global climate change issues, in
CO2 Emissions, Bonn, Germany, October 30–31, 1995, JI Online
Electricity and the Environment, Proc. Senior Expert Symp., Hel-
World Wide Web Server, http://www.ji.org/iuep/eabonn.shtml, Oc-
sinki, Finland, May 13–17, Vienna: Int. Atomic Energy Agency,
tober 1995.
1991, pp. 203–217. 32.
International Utility Efficiency Partnerships, Inc., The Interna-
ARNULFO DE CASTRO
SAIFUR RAHMAN
CONCHA REID
Virginia Polytechnic Institute and
State University
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W6702.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Cepstral Analysis of Speech
● Email this page Enter words or
Standard Article
Abstract
Liftering
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W6702.htm17.06.2008 16:34:48
The
most popular model of speech production views speech
arable.
CEPSTRAL ANALYSIS
The
inverse discrete Fourier transform (IDFT), c , of
log#X
(#)# is called the cepstrum of speech signal x(t) and is
represented as
2π −π (2)
for n = 0, 1, . . .,N
× 10
symmetrical against the origin of the frequency axis.
2
In the cepstrum, the cepstra of the excitation source and
1.5
the vocal tract resonance are additively combined since the
IDFT is a linear operation. Thus, as is illustrated in Fig. 1,
1
the cepstral analysis of a speech signal is a series of opera-
tions consisting of windowing, DFT, absolute operation, loga- e
0.5
d
u
t
l
0
p
(Fig. 2), a log spectrum [Fig. 3(b)], and the resulting cepstrum m
–2
rapidly varying component, corresponding to the excitation
0 10 20 30 40
source, is represented by the high-quefrency component. Note
Time (ms)
that a strong peak component is observed at a quefrency
A low-
order cepstral coefficient represents a slowly varying
Logarithmic transform
Spectrum of a vowel
40
IDFT
20
(a)
)
B
d
(
0
e
d
u
(b)
Liftering t
i
n
g
–20
a
M
(c)
(d)
–60
0 1 2 3 4 5 6
Frequency (kHz)
Spectral Fundamental
envelope frequency Figure
3. Spectrum of a vowel sampled at a frequency of 12 kHz:
(a) LPC
spectral envelope (the order of LPC analysis is 16); (b) FFT
Low quefrency region High quefrency region
spectrum; (c) FFT spectrum obtained by truncating cepstrum at que-
frency
# 48; (d) LPC spectrum obtained by truncating cepstrum at
Figure 1. Cepstral analysis.
quefrency # 12.
Cepstrum If
Eq. (3) is represented by
0.4
H (z ) = (7)
A (z )
0.2
A (z ) = 1 − # α z −n (8)
n= 1
0
= # (1 − q z −1) (9)
k
–0.2
k = 1
the
log-spectrum can be represented as
–0.4
n
–0.6
n= 1
0 5 10 15
p
Quefrency (ms)
Rn = # qn (11)
l
Figure 4. Cepstral coefficients.
l = 1
The
variable denoted as Rn is called the root-power sum. By
viable pitch extraction method. Such strong cepstrum peaks
equating equal powers of z #1 between Eq. (10) and Eq. (4), the
are indicators of voiced speech sounds and they are not ob-
simple relation
1
of the peak indicates the periodicity of the speech signal and
cn = Rn (12)
can be used for voiced-unvoiced decision purposes.
n
#
has basic properties similar to the cepstrum and can be de-
p p
# #
(( k ) − 1)!
i= 1 i k
rived in a computationally efficient manner.
cn = #p (−α ) i (13)
(k !) i
In the linear prediction model, the transfer function H(z)
i= 1 i i= 1
of the vocal tract is represented by an all-pole transfer func-
tion with p poles as where
the sum is taken over all combinations of ki to meet
the
condition
1
#
H (z ) = p −n (3)
1 − α z
p
#
n= 1 n
iki = n (14)
where # , n # 1, 2, . . ., p are LPC coefficients and z is the
i= 1
n
c as
n
n
n= 1 sures
have been proposed for speech processing. These are
used
for measuring the spectral difference between two spec-
All the poles of H(z) must be inside the unit circle. After sub- tra.
The spectral distance measures have been widely applied
stituting H(z) from Eq. (3) into Eq. (4), derivative operations in
template matching-based speech recognition.
for both sides of Eq. (4) eventually lead to the following simple
The Euclidean distance of two spectra S(f )(#) and S(g )(#) is
relationship between the cepstral coefficients cn and LPC coef- given
by
ficients #
n
c1 = −α1 (5)
DCEP = 1 # π (log(S(f ) (ω)) = log(S(g ) (ω)))2 dω (15)
2π
−π
#n −1 k
−αn − 1 − α c 1 < n ≤ p
∞
#
k = 1 n k n −k
c = (6)
(f ) (g ) 2
#
n p k
= (cn − cn ) (16)
− k = 1 1 − n αk cn −k p < n
n=−∞
Truncating
cepstrum cn at n #
, is equivalent to multiplying
where r is a balancing weight. Spectral distance between it by the
lifter weight
spectra can be used for representing spectral difference, how-
ever, the log spectral distance provides better speech recogni-
#1 |n | ≤ ν
The
spectrum corresponding to the truncated cepstrum is
LIFTERING
smoothed.
As is shown in Fig. 3, the LPC spectrum (a) has
sharper
peaks than that produced by the truncated LPC cep-
An operation to separate slowly varying spectral components
strum (d).
#
trum Q(#) is given by
d log(S (ω))
= ncn cos(nω) (25)
#
dω
∞
n=−∞
clidean
distance between two spectral slope functions (2,3).
In digital speech processing, a speech signal is obtained not
(f ) (g )
The
spectral slope distance between two spectra S and S
as a continuous function of time, but as a sampled time series.
is given
by
When the sampling frequency is f s , the bandwidth of a speech
signal is limited to
#
2
∞
If Q(#) and S(#) are represented in a logarithmic magni- =
# #nc(f ) − nc(g )#2 (27)
n n
tude scale, these spectra are represented by cosine expansions
n=−∞
∞
log(Q(ω)) = # bn cos(ωn) (20) Weighted
Cepstrum
n=−∞
To
measure distance in a
multidimensional space, variance-
∞
log(S (ω)) = # cn cos(ωn) (21) normalized
distance measures such as the Mahalanobis dis-
n=−∞ tance
provide good performance in general pattern recogni-
tion. A
weighted cepstrum is such a variance-normalized pa-
where bn and cn are cepstral coefficients. The lifter impulse rameter
if each cepstral coefficient is regarded as an
response can also be represented by a cosine expansion.
independent parameter (4,5).
∞
cn
#
c(W ) = (28)
H (ω) = ln cos(ωn) (22)
n σn
n=−∞
where #2
is the variance of the nth ceptral coefficient. In case
By using the orthogonality of cosine functions, the following n
of actual
speech cepstrum data, 1/#n is approximately propor-
equation is derived:
tional to
order n, so the weighted cepstrum is approximately
b = l c (23) given by
n n n
c(W ) = w c (29)
Lifter weights ln correspond to the transfer function of each
n n n
cosine component in the log spectrum.
#
n if n < ns
Applying a low-pass lifter yields the slowly varying cep-
wn = (30)
ns otherwise
strum component corresponding to the vocal tract resonance.
DFT computation of the low-pass liftered cepstrum produces
the so-called spectrum envelope [Fig. 3(c)] consisting of the If ns #
#, the weighted cepstrum is equal to the root-power
slowly varying spectrum component. This operation is called sum shown
in Eq. (11). The weighting function wn saturates
cepstral smoothing. Since the spectral envelope is sufficiently at n # ns
to suppress excess weighting of high-order cepstral
smooth, local spectral peaks characterized by the formant fre-
coefficients. The Euclidean distance between two weighted
quencies and bandwidths corresponding to the vocal tract res- spectra
can be defined as a spectral distance measure.
L
speech spectrum as
c# (i) = l =−L cn (i + l )lw (l ) (41)
n L 2
l =−L l w(l )
1
#
H (z ) = p −n (31)
1 + α z
n= 1 n where i
denotes frame number of the cepstral time series.
p
#
1 Delta-
cepstrum picks up the trend in the window #L
l
L
= n= 1 1 − (zn /z ) (32) of a
cepstral time series. The original delta-cepstrum is de-
rived
based on linear regression against the scatter of cep-
p
#
stral
values in the time axis.
= Hn (z ) (33) The
formulation of the delta-cepstrum in Eq. (41) can be
n= 1
j ω j φn (ω)
L − l
Hn (e ) = A n (ω)e (34)
if l ≥ 0
G
w(l ) = L (42)
if l < 0
derivative of the phase
L
p
TG (ω) = # −dφn (ω) (35) gives a
bandpass transfer function that is better in eliminat-
n dω
n= 1 ing
side-lobes than a uniform weighting expression derived
from the
original formulation of regression. The frequency of
Cepstral coefficients for a group delay spectrum are de- the
temporally varying component in the cepstral time series
rived as is
called the modulation frequency.
A
delta-cepstral distance can be defined by a Euclidean
g n = ncn (36)
distance
like the cepstral distance. Let DCEP be a cepstral dis-
tance
and D#CEP be the delta-cepstral distance, the combina-
The original group delay spectrum formulation excessively
tional
distance can be defined as
emphasizes high-order cepstral coefficients. Thus, the follow-
ing generalized weighting function is used for practical
g = w c (37) where r
is a balancing weight between the cepstral distance
n n n
2 and the
delta-cepstral distance. A typical value of r for auto-
n
wn = ns exp −2τ 2 (s ≥ 0) (38) matic
speech recognition is 0.05.
2
weights represented as
dn (i) = cn (i) + 8#cn (i) − 8# cn (i) (44)
3
b = w c (39)
l =−3 lcn (i + l )
n n n
#cn (i) = (45)
3 l2
l =−3
The lifter weight
3 (l2 − 4)cn (i + l )
l =−3
2
ν n
# cn (i) = (46)
#
wn = 1 + sin π (1 ≤ n ≤ ν) (40)
3 (l2 − 4)2
2 ν
l =−3
delta-
cepstrum at time i.
TEMPORAL ANALYSIS OF CEPSTRUM
RASTA
Delta-Cepstrum
RASTA is
another filter for cepstral time series and performs
The auditory system is sensitive to temporal changes in as a
bandpass filter (10). RASTA achieves a lower resonance
sound features. Research effort has been focused on simulat-
modulation-frequency by multiplying a temporal integration
L ∞
#
The numerator is a delta-cepstrum. The denominator con-
tributes to temporal integration. The terms z #2 and z #4 do not
log(Q(ω,t)) = log(S (ω − λ, t + i#t))M (λ, i#t) dλ (52)
affect the gain of the modulation-frequency transfer function.
i=0 −∞
N
and a highpass filter as
bn (t) = # ln (i)cn (t + i#t) (53)
i=0
#
L l −l
l = 1 β z
H (z ) = 1 − (49)
This lifter has two different aspects as an order-dependent
#
L βl
l = 1
temporal filter for cepstral time series, and an elapsed-time-
elapsed time.
FM Neuron Model
BIBLIOGRAPHY
change. A model has been proposed to simulate the function
of the FM neuron (12). The FM neuron model is formulated
as a time and frequency derivative of a spectral time series 1.
M. R. Schroeder, Direct (nonrecursive) relations between ceps-
973, 1987.
The cepstrum coefficients for the FM neuron output are given
by 4.
Y. Tohkura, A weighted cepstral distance measure for speech rec-
8.
S. Furui, Speaker-independent isolated word recognition using
The filter coefficients are identical among all cepstral orders
dynamic features of speech spectrum, IEEE Trans. Acoust. Speech
in the above cepstral analysis. A spectrotemporal spectral
Signal Process., ASSP-34 : 52–59, 1986.
representation is achieved by a two-dimensional matrix lifter 9.
S. Furui, Speaker-independent isolated word recognition based
based on the auditory masking effect (13). In an auditory sys- on
dynamics-emphasized cepstrum, Trans. IECE Jpn., E 69 (12):
tem the target sound is suppressed by a masker sound. The
1310–1317, 1986.
masked spectrum is given by the spectrum reduced by the 10.
H. Hermansky et al., Compensation for the effect of the commu-
masking level. The masking level is a function of the fre-
nication channel in auditory-like analysis of speech (RASTA-
quency difference between the masker and the signal and the
PLP), Proc. Eurospeech 91, pp, 1367–1370, 1991.
12. K. Aikawa and S. Furui, Spectral movement function and its ap-
plication to speech recognition, Proc. ICASSP88, pp. 223–226
15.S5.11, 1988.
YOH’ICHI TOHKURA
KIYOAKI AIKAWA
NTT Laboratories
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3012.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Compressed Air Energy Storage
● Email this page Enter words or
Standard Article
Abstract
Conclusion
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3012.htm17.06.2008 16:35:09
Energy conversion paths are either direct or indirect (Fig. 1). With
indirect technologies the line of energy
conversion is the following: the primary source is exothermally oxidized
(burnt), the chemical energy being
converted into thermal energy. By a thermal agent (motor fluid), the thermal energy
is used as such, or by using
a thermal machine (turbine, engine) it is converted into mechanical energy
actuating an energy-generating
working engine or generator. With direct technologies thermal
intermediation is eliminated along with its
thermodynamic imperfections, and electrical energy is directly gained, for
instance, from the chemical energy
of fuel. Wind energy falls into this category.
There are several competitive possibilities for converting primary energy
into electrical energy: hydraulic,
wind-driven, electrochemical, and nuclear (including fusion); solar energy may be
directly utilized on a photo-
voltaic line.
Of all primary energy forms suitable for sustainable development, wind energy
is the most advantageous:
it is practically inexhaustible, nonpolluting, available everywhere, and free of
charge, since no primary extrac-
tion is required. In converting the wind energy to other forms, the following
stages are to be noted: extraction,
conversion, storage, and consumption. For wind, energy storage is necessary both
because of the limitations
of wind (it is irregularly distributed in time and space, and it is dilute, that
is, it has low concentration per
unit area), and because of the variety of its consumers. Figure 2
presents a general diagram of extraction,
conversion, storage, and utilization of wind energy.
In the first stage (extracting energy from wind), mechanical energy is gained
as rotation, translation, or
oscillation, depending on the type of the extraction installation. This motion may
be used in the second stage
for conversion into another form of energy by liquid pumping, gas compression, or
electrical or heat generation.
The energy gained may be stored by using pneumatic or hydraulic systems, electric
batteries, heat, hydrogen,
flywheels, etc. This is the third stage. The fourth stage, that of consumption,
consists in supplying energy to
consumers in the form of heat, electricity, water, etc.
A comparison between different energy storage systems is presented in Fig. 3.
For compressed air, starting
with the data from the literature, which takes into consideration only the
mechanical component of the energy,
we have also taken into consideration the possibilities of utilizing the thermal
component.
COMPRESSED
AIR ENERGY STORAGE 3
capacity on basic load, the compressor is switched off and the generator is
switched to the wind turbine. The
burner is supplied with gas fuel and compressed air from the storage system.
On compression without heat loss (adiabatic compression), the air temperature
increases. If compressed
air is stored in a thermally well-insulated accumulator, we speak of adiabatic
storage. When compressed air
at this temperature is taken out and used to actuate a turbine, as in the
previously mentioned case, less heat
is required than if the storage were at ambient temperature. Therefore, adiabatic
storage is more economical
than isothermal.
Compressed Air Storage with Hydropneumatic Accumulators.
Description of the Accumulator. The hydropneumatic accumulator (Fig. 7)
is made up of a pneumocham-
ber (1) and a hydrochamber (2). Between the two chambers there is a 65 m to 70 m
vertical level difference,
corresponding to a 6.5 bar to 7.0 bar pressure, joined by a connecting incline (4).
Both chambers are equipped
with closing dams (5). At higher storage pressures, the level difference
increases. The pneumochamber is
connected to the compressor station (6) by a compressed air supply duct (7), and
to the main network by a
distribution line (8). Both ducts are dimensioned to suit the average consumption.
Water is supplied to the
hydrochamber by a supply duct (9).
A hydraulic connection (11) exists between the pneumochamber and the
hydrochamber, made up of metal
ducts, with mining workings forming a hydraulic arrester in the lower part (12).
All ducts are fitted with shutoff valves (14), permitting the hydropneumatic
accumulator to be isolated for
inspection and maintenance as necessary. Between the compressed air supply duct and
the distribution line
there is a bypass (15), which may be used to short-circuit the hydropneumatic
accumulator.
The pneumochamber must be carefully designed to avoid water or compressed air
losses. Proper sealing
is decisive for good operation and for the economic efficiency of the accumulator.
With fissured rocks or with
medium-hard and soft rocks, consolidation and sealing are effected by cement-milk
or synthetic-resin injections.
The resistance dam closing the pneumochamber (1) must be specially
constructed for durability and
sealing, as it is crossed by the compressed air supply and distribution
ducts, transducers for working and
maximum levels, and an access opening where personnel can enter for control and
maintenance. Optionally
the resistance dam may be fitted with a water-gauge glass, which, during
test periods, may provide data
COMPRESSED
AIR ENERGY STORAGE 7
where
V = pneumochamber volume (m3 )
3
Qmax = maximum compressed air consumption (m /min)
Q 3
st = actual flow of active compressors (m /min)
τ = time of maximum-consumption requirement (min)
• One may use the following empirical function of the active power of
the compressor station: for each
kilowatthour, 1.5 m3 of accumulation buffer reservoir volume is needed at 7
bar pressure (1).
where
p p = pressure in pneumochamber (bar)
p s = pressure in compressor station (bar)
#p l = sum of linear losses (bar)
#p ε = sum of local losses due to supports, bends, or cross-section
alteration (bar).
where
H = height between the pneumochamber and the hydrochamber (m)
2
p p = compressed air pressure in the pneumochamber (N/m )
2
p 0 = air pressure above water in the hydrochamber, (N/m )
γ = volumetric weight of water (N/m3 ).
COMPRESSED
AIR ENERGY STORAGE 9
For large storage capacities (≥ 1500 MWh) and economical operation, hydraulic
and pneumatic storage
have proved to be the most advantageous. Hydraulic storage is recommended for areas
with uneven topography.
In flat regions pneumatic storage is better.
Thermoelectric power stations with air reservoirs have the following
advantages:
• Improvement of daily and weekly load balance and a more rational use of
existing stations
• The possibility of using secondary energy resources at the stations
• Fuel savings compared to other peak stations
• The possibility of coupling systems for the use of renewable energy resources
to thermoelectric stations.
The most advantageous economic and functional conditions for compressed air
storage reservoir arrange-
ments are offered by underground saline rock formations. Air storage containers can
also be excavated in hard
rocks such as granite, gneiss, sandstone, or limestone. Since these are
expensive operations, air reservoirs
should be hydraulically compensated (water column) to minimize the volume necessary
for the reservoir. In
some cases, natural or existing artificial cavities (abandoned mine workings,
emptied petroleum or natural-gas
reservoirs) can be used as air reservoirs.
Various schemes for pneumatic energy storage at thermoelectric stations are
now compared with respect
to the storage cycle and choice of reservoir.
RT is relevant. It equals
which supplies the energy necessary for the compressor drive, the total
efficiency η
the total fuel energy (kJ) required in the basic station and in the
air-reservoir station to obtaining 1 kJ at
RT the loss coefficient ε can be derived,
expressing the total energy lost to the
the alternator terminals. From η
surroundings as heat. The expressions for the two indicators are
where
EGen = energy delivered at alternator terminals at the air-reservoir
station
EMot = energy supplied to compressor motor at the air-reservoir station by
the basic station
η =E /E = basic station efficiency versus motor terminals.
B Mot f1
Fig. 11. Nonadiabatic storage installation fitted with low-pressure and high-
pressure combustion chambers. (a) Schematic
diagram (see also Fig. 10): J, recuperator; K, high-pressure combustion chamber; L,
high-pressure turbine; M, low-pressure
combustion chamber; N, low-pressure turbine; O, intermediary cooler. (b) Entropy
diagram.
The efficiencies of the adiabatic and nonadiabatic schemes are shown in Table
2.
A few remarks should be added on the financial aspects of these
comparative values. The total costs
associated with an air-reservoir station can be analyzed as follows (for
ten years of turbine operation with
reservoir load):
COMPRESSED AIR
ENERGY STORAGE 13
The low values for the reservoir refer to reservoirs in saline rocks. If
several groups of machines were
connected to the same reservoir, its share in the overall cost would be even
smaller.
In conclusion one may say that the energy storage installations described in
this section are economic.
Advantages and Disadvantages of Compressed Air for Use in Energy Transfer and
Storage.
Compared to other energy storage systems, compressed air shows the following
advantages:
Low energy efficiency is its main disadvantage, both in industrial uses and in
traditional storage systems.
The losses within the process of compressed air generation, transport, storage, and
use are as follows: 18% loss
by friction in the compressor, 10% air losses due to compressor leakage, 18%
internal energy losses by cooling
in storage, and 34% losses in adiabatic expansion. These losses sum to 80%, so that
20% remains available.
Ignoring constructive improvements of the compressor (losses due to friction
and leakage diminished) and
focusing on recovering the thermal component in compressed air generation and on as
complete as possible
utilization of the adiabatic expansion (pneumatic expansion engines,
refrigeration plants with air), we can
state that the available energy may be increased by about 40%, so that the
utilization coefficient for pneumatic
energy should be competitive with that for electricity.
COMPRESSED
AIR ENERGY STORAGE 15
where
i = specific enthalpy
s = specific entropy and subscript 0 refers to the surroundings.
The exergy e is seen to consist of two components: (a) one that depends only
on the temperature of the
fluid, called the thermal component eT , and (b) one that depends only on its
pressure, called the mechanical
COMPRESSED AIR
ENERGY STORAGE 17
and thus
• The thermal component represents the absolute value of heat exergy exchanged by
the gas with the exterior
in an isobaric temperature change from T to T0 .
• The mechanical component represents the isothermal work exchanged by
the gas with the exterior at
temperature T0 during a pressure change from p to p 0 .
The p–v [Fig. 16(a)] and T–s [Fig. 16(b)] graphs clarify the significance of
the two exergy components.
Examining the expressions for the components eT and ep of exergy, we
see that the thermal fluid can be
brought from state 1 to equilibrium with the environment in the following stages:
and taking into consideration the graphical significance of the exergy e1, we have
Considering that
where k is the adiabatic exponent of the gas, and representing graphically the
dimensionless values
the curves shown in Fig. 17 are obtained. It is to be noticed that unlike the
component eT , which is always
positive [Fig. 17(a)] irrespective of the temperature range (T >T or T <T
), the component e is positive in the
0
0 p
range of pressures p >p 0 and negative in the range p <p 0 [Fig. 17(b)].
Exergetic Method of Analysis of Processes within Pneumatic Networks.
Exergetic analysis that in-
cludes the first and second laws of thermodynamics allows quantitative definition of
the qualitative term energy
degradation. Consequences of various thermodynamically irreversible phenomena can
thus be calculated, lead-
ing to correct understanding of thermodynamic losses within a system. Exergetic
analysis contributes to the
definition of a thermodynamic efficiency, expressing the degree of perfection
or thermodynamic quality of a
system. The concept of exergy stands at the basis of this type of analysis: a
nonconserved quantity with the
dimensions of energy but with unlimited transformation capacity. In the case where
the system is made up of a
flowing fluid, the exergy of the system is equal to the difference between the
initial value of the Gibbs function
(free enthalpy) of the system and the value of the Gibbs function of the system
when it cannot generate useful
work by interaction with thermal reservoir, that is, when the system is in thermal
and mechanical equilibrium
with the reservoir (3).
In the particular case of compressed air flow through the sections
of a pneumatic network, the main
causes of internal and external irreversibility are:
COMPRESSED AIR
ENERGY STORAGE 19
Let Mit be the compressed air mass flow into a section; M et the mass flow out
of the section; M mt the average
mass flow through the section; #M =M et −Mit the leakage; Tac the average
compressed air temperature; p m
the average compressed air pressure; i0 and s0 the specific enthalpy and
specific entropy of the environment,
respectively; im and sm the average specific enthalpy and specific entropy
of compressed air; lfr the specific
frictional work; and qext the heat transfer to the environment. Then we have for
the equation of exergy balance
of a duct section
Inserting the exergy expression Eq. (5) for a flowing fluid in Eq. (9), we have
Let us move the term representing the exergy of the fluid coming out of the
section to the left side of Eq.
(10) and suppose that losses due to leakage take place through adiabatic laminar
flow, kinetic energy variations
being negligible, so that i =i . Then the exergy losses for compressed air
flow through the considered duct
m 0
section will be given by the equation
From the terms giving the exergy loss due to irreversible transfer thermal
between the system and the
thermal source,
one may notice that if all values entering the expression are assumed constant,
then the exergy loss #Eq is
the same irrespective of the evolution of the system to T >T or T
<T —an explicable fact, since the return
ac 0 ac
0
of the system to thermal equilibrium with the thermal source, from temperatures
equally spaced above and
below the temperature of the source, implies the same loss of exergy. The terms on
the right side of Eq. (12)
represent exergy losses due to irreversible phenomena accompanying compressed air
flow along the sections
of the pneumatic mining network. The analysis of Eq. (12) may lead to a series of
technical and organizational
measures to reduce exergy losses to a technical and economic minimum.
The exergy losses due to leakage in pneumatic networks,
represent the largest share (65% to 75%) of the total losses. To reduce
them, the following measures are
advisable:
----------------------- Page 50-----------------------
COMPRESSED
AIR ENERGY STORAGE 21
• Providing an as high as possible pressure for the compressed air along the
section
• Reducing pressure losses due to friction (#p ) by increasing the duct diameter
• Reducing pressure losses resulting from partial obstruction of the duct (#p
obt ) by removing humidity from
the compressed air and cleaning the insides of the ducts when they are mounted
• Conveying compressed air along the section at the highest possible
temperature, thus cutting down exergy
losses corresponding to work done against friction. (not advisable for
pneumatic mining networks, as it
leads to considerable increase of other loss categories)
Taking into account the relationship in Eq. (12), the expression for the
exergetic efficiency for a pneumatic
network section becomes
V 1 in the intake duct at parameters p 1 and T1; it is separated from the rest of
the air by an imaginary piston
under pressure p 1. At the end of the filling stage at V =const, the cylinder
contains the mass of air ma >mc at
parameters p a =p 1 −#p a and Ta >Tc [Fig. 18(b)]. When the intake valve is
closed, the mass of air in the cylinder
becomes ma , occupying volume Va at parameters p a and Ta [Fig.
18(c)].
Let eqa be the exergy exchanged by the air in the filling process, ea the
exergy when filling is over, e1 the
exergy when admission starts, lta the mechanical work of the isochoric
filling stage, and πira the exergy loss
due to the internal irreversibility of the filling process. Then the following
exergetic balance equation of the
filling process is obtained:
where
Denoting
COMPRESSED AIR
ENERGY STORAGE 23
it is seen that
is the exergy loss due to laminar air flow in the cross section of the intake valve.
Therefore
Denoting by
the exergy losses induced only by finite pressure and temperature differences during
the filling process, we
conclude that
[Fig. 19(c)].
Let eqe be the exergy exchanged by the air in the evacuation process, e2
the exergy when evacuation is
over, ed the exergy when evacuation starts, lt2 the mechanical work of
evacuation, lTod the isothermal work
of expansion, and πire the exergy loss due to internal irreversibility of
evacuation. Then the exergetic balance
COMPRESSED AIR
ENERGY STORAGE 25
where
We denote
In the last relationship, the second term gives the exergy loss due to
laminar gas flow in the section of the
release valve:
Denoting by
Combining with the above balances the exergetic balances of other components
in the energy conversion
and storage scheme, a thermal–economic calculation of the performance of the
energy-conversion and -storage
installation may be made.
where
w =
flow velocity
τ =
time
x =
spatial coordinate
g =
gravitational acceleration
ρ =
density
p =
pressure
λ =
fluid-dynamic resistance ratio
D = duct internal diameter;
δ= boundary-layer thickness of flow section
d= D −2δ. The fluid load loss was approximated with the loss corresponding
to the steady flow
expressed by the Weissbach–Darcy relation.
2
On neglecting the terms due to kinetic energy variation [∂(ρw )/∂x]
and gravitational forces (ρgx ) in
comparison with the term due to friction [ρ(γ/d) w2/2], Eq. (34) becomes
Equation (35), together with the continuity equation in Eq. (33) and the
state equation ρ=ρ (p ), forms a
determinate system of equations.
ˆ
value 2c=(λw/2d)m , and we use the relationship suggested by I. A.
Ciarnıi (6), assuming a constant fluid-
dynamic resistance coefficient λ in the expression for c, so that 2c=(λ/3d)
(winitial +wend ). Then the equation
becomes
For long ducts at times τ>L/a the term ∂(ρw)/∂τ may be neglected, and from
Eqs. (36) and (37) the following
system is obtained:
The same technique is used to remove the pressure from Eqs. (38) and (39).
Differentiating Eq. (38) with
respect to time and Eq. (39) with respect to x , and equating them, we obtain
Solving Eqs. (40) and (41) allows the pressure and mass-flow variation in
nonstationary fluid flow through
a long duct to be determined.
For the mass-flow variation law a relationship having the same form is found:
COMPRESSED AIR
ENERGY STORAGE 29
the trunk line compensates for the extra flow rate required by the consumers. When
the flow required by the consumers exceeds the flow supplied by the compressor and a
second compressor must
be coupled to the network to provide the required supplementary flow. The
accumulated flow, compensated
flow, and additional flow required are determined by Eq. (43) from the
differences #M =M (0, τ)−M (L , τ)
corresponding to each operating regime of the trunk line.
Possibilities for Compressed Air Storage with the Help of a
Collector–Distributor Section.
Pneumatic equipment is supplied with compressed air from a pneumatic distribution
network at points rel-
atively close to one another. Nevertheless, the distribution network is
connected to two trunk lines coming
from different compressor stations, to provide better service to pneumatic
consumers. For large systems it is
advisable to transform the pneumatic network from a branched network into an
annular one, by developing an
annular collector–distributor made up of one or several loops resulting from
parallel connection of distribution
ducts.
The collector–distributor will be supplied at the two ends of the
arragement. Flexible supply is thus
assured, leveling the load curve, consumption peaks being taken over by the
collector–distributor using com-
pressed air accumulated during low consumption periods. Supplying consumers with
required flows at neces-
sary pressures by this means involves certain restrictions on the pressure
variations in the collector–distributor
due to uneven consumption.
At the branching points of the consumers’ supply ducts to the collector–
distributor, the pressure must
be higher than a value p 2min determined by the pressure required for
pneumatic equipment and the losses
accompanying compressed air flow along the distribution ducts. Collector–distributor
supply points will be at
a lower pressure than the value p 1max set at the buffer reservoir discharge valve
in the ecompressor station.
The compressed air required for pneumatic equipment loads must be supplied
over a 24 h cycle by meeting
the following conditions:
where
p 1(L , τ) = compressed air pressure variation at the branching points of
the collector–distributor from
trunk lines
p 2 = compressed air pressure variation at consumer–distributor consumption
points
where
3
Q =Q +#Q −Q = flow deficit during the time τ (N·m /min)
di ci i sci di
3
Qci = flow required by consumers during the time τdi (N·m /min)
3
Qsci = flow conveyed to compressor station during the time τdi (N·m /min)
τdi = life span of flow deficit Qdi (min)
p 1, p 2 , p N = absolute pressures of compressed air before and after
filling up the collector–distributor,
and pressure corresponding to the normal physical state (bars)
T1, T2 , TN = absolute temperatures of compressed air before and after
filling up the
collector–distributor, and absolute temperature corresponding to the normal
physical state (K)
3
where Q =Q −(#Q+Q ) is the available flow for filling (N·m /min).
u sc c
The flow and pressure variations in time in the collector–distributor supply
are determined by Eqs. (42)
and (43) when their variations at the consumption points are known.
Turning the collector–distributor into a pneumatic-energy “flywheel” provides
the following benefits in
the operation of pneumatic installations:
Conclusion
BIBLIOGRAPHY
READING LIST
´
P. Chesse J. F. Hetet J. P. Frayret Determination par simulation d’une zone
critique des champs compresseur situee entre
´
le pompage et la stabilite, Entropie , 201: 9–11, 1996.
D. Coiante Uno Sguardo al futuro dell’Energia, Studi e Ricerche, ENEA, Italy: 1996.
IOAN-IULIAN IRIMIE
SABINA IRIMIE
ILDIKO-CAMELIA TULBURE
University of Petrosani
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3001.htm
} { { } }
HOME
ABOUT US
CONTACT US
●
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
DC Machines
Abstract
Motor–Generator Sets
Solid-State Drives
Control
Armature Coils
Temperature Control
Time-Varying Loads
Limits of DC Machines
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3001.htm17.06.2008 16:35:29
DC MACHINES 63
within
slots along the circumference of the cylindrical rotor
iron
core near the surface; its coil ends are connected to the
64 DC MACHINES
unit in spite of the fact that the machine itself is rather ex-
pensive due to the mechanically complicated rotor structure.
In Fig. 1(a) a simplified cross section of a two-pole dc machine
Nevertheless, the decrease of costs for ac inverters, their im-
is depicted (number of pole pairs p # 1). This simple machine
proved switching performance due to improved power elec-
has only one armature coil lying in two opposite slots at diam-
tronic devices such as insulated gate bipolar transistors
eter d, with its ends connected to a simplified commutator
(IGBT), and the advent of improved motor control for ac in-
with two copper segments. The machine—with its rotor turn-
duction machines (so-called vector control) due to improved
ing with rotational speed n— acts as a dc generator at no-load
microcomputer technology led to a change from ac to dc
condition, as there is no current flow in the armature, as no
drives. The induction motor is more robust, having only a
load is connected to the terminals. The north and the south
squirrel cage in the rotor instead of the mechanical commuta-
pole air-gap magnetic field with peak flux density B induces
tor and the wound armature winding, thus aiming for higher
an ac voltage (emf) ui into the armature coil due to the move-
speed limits and resulting in higher dynamic performance
ment of the coil sides with air-gap velocity v (length l of each
due to the lower moment of inertia, at reduced maintenance
coil side corresponding to the iron stack length and number
costs (1). In the high-power region above 1 MW inverter-fed
of turns per coil Nc) [Fig. 1(b)]:
synchronous machines are also replacing large dc machines,
as they allow a more compact build due to the lacking commu-
v = dπn (1)
tator. Larger ratios output power versus motor volume are
achieved and higher mechanical speed is reached. For big dc
drives, for example, two coupled machines (tandem motors)
The rectified ac voltage u yields a dc voltage U , which for the
i i
N Blv
N
c
B
Main pole Armature l d/2
(“north”) coil
ui
Rotor
Slot n
Nc Blv
Brush
Armature
B
coil Brush
Commutator Ui
N
Main pole
(“south”)
(a)
(b)
l K
and rectification by the commutator U . (c)
o 2N Blv
i 0
V 1 2 c
The nonuniform air-gap flux distribution 0 T 2N
Blv T Time t TC = nK
causes a dc terminal voltage U with con- 2 c
i
0
siderable ripple. (d) The voltage ripple ui
0 Time t
gets smaller if more than one armature
coil is connected in series (K/2 # 1). (c)
(d)
DC MACHINES 65
U
U
Uf
Ia Ia
If
If
A A
F
C C
(a) (b)
U U
I = I
f a
If
A F A
F
U = 2N Blv (2)
dπ
i c
# = αe lB (6)
1
2p
f a = np = (3)
T
In
the same way the expression for the electromagnetic
2π
with both ends to K commutator segments, resulting in two
parallel paths of K/2 series connected coils between the
The terminal voltage U has to balance the emf and the volt-
plus and minus brush, when a two-pole machine is consid-
age drop due to the armature resistance Ra and the brush
ered. The induced ac voltages of adjacent armature coils
voltage drop U , which typically is about 2 V for plus and
b
are phase-shifted according to the displacement of one com-
minus brush, nearly independent of load current. Depending
mutator pitch, yielding a superimposed ac ripple on the on
machine operation mode (motor or generator) and thus po-
rectified voltage with commutator frequency f C [Fig. 1(d)].
larity of armature current (positive or negative) one gets two
This ripple of typically 1% to 2% is lower the higher the
expressions for the voltage equation (6,7).
ratio K/(2p ) is chosen (typically 30 to 40), yielding a nearly
Motor operation:
ideal dc voltage
U = U + I R + U (8)
i a a b
1
f C = Kn = (4)
T
C
Generator operation:
i a a b
66 DC MACHINES
Generator Characteristics
Figure 3. No-load characteristic of a dc generator for fixed rotational
speed n: no-load voltage U , residual voltage U , induced voltage U
An external torque (generated by a turbine, a diesel engine,
0 R i
tor [Fig. 4(a)] is given for fixed field current, yielding a no-
ration.
resistors in the armature circuit are depicted. The field resis-
output exists.
2. Shunt excitation [Fig. 2(b)]
From Fig. 3 one easily derives the load characteristic of
DC MACHINES 67
U0
U0 I R
U
a a
e
U
g
t
e
l
g Voltage drop
o
a
V
t
l due to armature
o
V I = const reaction
f
0
0
0 Armature current Ia
0 Armature current Ia Iam
(a)
(b)
Voltage drop
due to armature Ui (Ia = 0)
reaction
U (I > 0)
U i a
e
U0
g U
a I R + U
t a a b
U
l
o
e
V
g
UR
I N
0
0
0 Current Ia = If
0 Armature current Ia
(c)
(d)
voltage drop along the circuit resistance and emf are equal.
smaller shaft torque (typically 1% to 2%). The decrease of
The discovery of this so-called self-excitation (dynamoelectric
speed with increasing torque is rather small [Fig. 5(a)], yield-
principle ) by Werner von Siemens in 1866 was a revolution to ing
a nearly constant speed with an upper value n0 at the no-
electric power generation and was one of the main reasons for load
condition:
the breakthrough of electricity in industrial application. The
same effect is now used for self-excitation of large synchro-
U
n0 = (11)
nous power generators during startup after blackout, and also
k#
for self-excitation of synchronous and induction stand-alone
generators, for example, in wind turbine application or emer- Flux
drop at high torque (or at high armature current),
gency generating sets.
caused by uncompensated armature reaction, will increase
assumed), one gets from Eqs. (5), (7), and (8) the following
(5), (7), and (8), one derives the speed–torque characteristic
M
For simplification, the electromagnetic torque M is assumed
to be the same as the shaft torque. Losses in the active rotor Thus
series motors are recommended drives for traction, ei-
iron, eddy current losses in the rotor conductors due to the ther
in battery-supplied electric vehicles and dc railway loco-
coil ac current, and inevitable friction losses cause a slightly
motives or in ac railway locomotives as single-phase commu-
68 DC MACHINES
π
2 Ra
2
n
k
d
l
e
a n
e
n U = const
p U = const
o d
s
i
t e
l
a e
a
t p
n
o s
Speed rise due to o
R
i
armature reaction t
R
0
0
0
Torque M 0 Torque M
(a)
(b)
n0
a n
o d
t e
a e
t p
R
Figure 5. Speed–torque characteristics of dc motors
0
for fixed armature voltage U. (a) Separate and shunt
0 Torque M
excitation, fixed field current If . (b) Series excitation.
(c) Compound excitation.
(c)
0
Combining series and shunt excitation again yields a
3
rather strongly load-dependent speed characteristic [Fig.
U0 = 2UL (14)
5(c)], but with the advantage of avoiding overspeeding with
π
SOLID-STATE DRIVES
N 0
DC MACHINES 69
uL
Fourth quadrant—motor operation, converter 1 active:
By
changing the delay angle #, the armature voltage U can
be
varied continuously, thus making itpossible to run the
Uf
drive with any selected pair (n,M) under steady-state condi-
U, u
tions within the limits of rated torque MN and maximum
m N
decreases with 1/n, Eq. (7). The slope of speed decrease with
a
0
1. Both converters are active and produce the same aver-
0 Time t
age output voltage U, but only one of them provides the
1
6f
(b)
Rotational speed n
Figure 6. Converter-fed, separately excited dc machine. (a) Six-pulse
converter for the armature circuit and two-pulse converter for the
field circuit (first-quadrant operation). (b) Average voltage U is de-
nM
creased by increasing the delay angle #.
Φ 2
P = UIa (16)
i
Φ > Φ U = const
1 2
Po = 2π nM (17)
U1
U > U
2 1 Φ = const
With a firing angle # larger than 90# the voltage U becomes
negative and the direction of speed will reverse from clock-
U = 0
wise (cw) to counterclockwise (ccw) direction. As current
M N
and torque are still positive, the dc power becomes nega-
tive, too, thus becoming a power flow from the dc machine
–M N 0 Torque M
N
First quadrant—motor operation, converter 1 active:
M ~1/n
n > 0, M > 0, U > 0, Ia > 0, α1 < 90◦
–n
n > 0, M < 0, U > 0, Ia < 0, α2 > 90◦
m
nN
and with fixed armature voltage U and weakened flux
for nN #
n < 0, M < 0, U < 0, Ia < 0, α2 < 90◦ n #
n .
70 DC MACHINES
i
of current flow (first quadrant operation). By varying the
T a
duty cycle T1/ TP (9) from zero to unity [Fig. 8(b)], the aver-
I
age output voltage is varied between 0 and U (pulse-width
f
0
U0 D
modulation technique):
T
U, u
U = 1 U0 (19)
creased.
which corresponds to the dc current value I . In the case of a
u ~i
quired (5 ms to 10 ms elapsing time for the changeover be-
I a
tween the two converters).
i *
Smaller dc machines (for example, permanent magnet
n* + Speed a – Current α
excited machines) are often fed by transistor choppers, be-
– controller + controller
ing operated with a fixed switching frequency f # 1/T .
P P
i
a
Larger machines in dc railway drives often use gate turn-
off thyristor (GTO) choppers. With GTO choppers for larger
currents only low switching frequencies are possible, typi-
cally 200 Hz to 500 Hz, whereas with transistors (used for
smaller currents) higher switching frequencies of 1 kHz to
I u ~n
f T
DC MACHINES 71
masses and decreases with increasing magnetic flux, Eq. (24). the
period of the current ripple. This time is needed to calcu-
As T is much shorter than T , the settling time of the current late
the average current value in order to eliminate the influ-
a m
trol.
These drives are
now replaced by permanent magnet
La dia + Raia + kn# = u (20)
synchronous machines with chopper control, which is trig-
dt gered
by a rotor position sensor. With these machines the
dn k
pulse-width-modulated inverter along with the rotor position
J2π dt = 2π ia # − ml (21)
sensor may be regarded as the replacement of the mechanical
# #
k# 2 less
dc drives (7).
2π
The cascaded speed control is usually extended by an emf
controller (field-weakening controller). If the measured speed
Combining Eqs. (20) and (21) yields Eq. (22), where u(t) is the n
surpasses rated speed nN , this controller—by using the no-
manipulated variable of the control circuit, whereas ml(t) de- load
characteristic (Fig. 3) as a numerical table—generates a
notes the disturbance. The response characteristic of a dc ma- new
setpoint value If , which is lower than the rated value
I ,
for the underlying I -current controller. This controller
chine with constant main flux, for example, to an increase of fN
f
load torque Ml may be described by the time constants Ta and
generates the necessary firing angle #f for the controlled recti-
Tm . The response of the armature current to a step of arma- fier
bridge of the field-current circuit [Fig. 6(a)], thus applying
a
voltage U to the field winding, which drives the wanted
ture voltage U at t # 0 is derived from Eq. (20). With ia(t) #
f
I for t # 0 and ia(t) # I
i for t # 0 one gets field current If . The
electromagnetic time constant Tf of the
field
circuit
# #
#U # t #
i(t) = 1 − exp − (25)
L
Ra Ta
T = f (27)
f R
72 DC MACHINES
τ
yielding in general the following law for simplex lap armature
p
ture):
South pole North pole
2a = 2p (28)
l
Magnetic asymmetries of different main poles cause slight
#####
v
slot. As one parallel path of simplex windings is usually de-
c
able slot cross section and the current density in the slot cop-
Brush
per bars, all larger machines are designed with lap windings
– +
(13). Very big machines used to have duplex lap windings,
U
consisting of two independent lap windings, each having a
(a)
commutator pitch YQ # 2. Parallel connection of both lap
YQ = (29)
2p
most cases p brush pairs are used, just as in the case with
– +
b
of the same polarity are connected in parallel. Therefore the
Ts = (30)
v
armature winding consists of four parallel electric circuits,
C
DC MACHINES 73
C C A
2a
justment of BC is accomplished by the proper choice of NC and
a C
T
to twice the rated armature current.
s
the stator poles (Fig. 12). This winding is also loaded by the
i
armature current, thus exciting an air-gap field, which op-
c
cogging, it
is possible to extinguish the armature reaction
v
field with its negative influence on the machine (16). The aux-
c
b
iliary windings do not play an active part in electrical energy
Figure 11. Commutation of armature coil current i . (a) The coil cur-
rent ripple (17), these losses yield a slightly lower machine
c
rent has to reverse its polarity, while the coil is short-circuited by the
efficiency when compared with ac induction and synchronous
brush. (b) The commutating lap coil is short-circuited by the brush.
machines (18).
TEMPERATURE CONTROL
from
I /(2a) to #I /(2a) and vice versa, causes a reactance
a a
There exist many different cooling systems for electrical ma-
voltage ur due to the inevitable coil stray inductance Lc [Fig.
chines (19); dc machines are manufactured often with open-
11(b)]:
circuit cooling. The cooling air flow enters the dc machine
through the axial gaps in the stator between the main poles,
This voltage causes ignition of sparks at the brush’s trailing
leaving the machine through openings in the drive-end end
edge, causing wear of the brush and the commutator, thus
shield (20). For variable-speed drives an external fan to gen-
reducing the brush life drastically. The ideas of the compole
erate a constant, speed-independent air flow is necessary. A
winding is to excite an additional air-gap field BC in the neu-
fan mounted on the rotor shaft of the machine itself would
tral zone between two adjacent main poles, which opposes the
deliver an insufficient air flow at low motor rotational speed.
armature reaction field. Thus ur can be compensated by an
The different windings of the dc machine (armature, field,
additional voltage uC, which is induced in the commutating
compole, compensation) and the surface of the commutator
coil due to the movement of the armature with speed v in the
with the brushes are cooled intensively with the cool air en-
compole field BC :
tering on the commutator side, passing closely to the surface
74 DC MACHINES
Field coil
####
Main pole
##
Compensating
winding
Main airgap δ
Rotor tooth
c Rotor slot
####
###
###
Shaft
#ϑ ∝ P = RI 2 (35)
d a
TIME-
VARYING LOADS
One has to consider that the heating of the armature winding
is caused not only by load losses, Eq. (35), but also by main As
already stated, the resulting inductivity of the armature
flux-dependent losses in the active iron core. By careful de- L is
rather low, yielding a low electrical time constant T ,
a
a
sign of armature and compole winding one tries to obtain giving
the dc drive good dynamical performance, which is nec-
nearly the same steady-state temperature rise in both wind- essary
especially under time-varying load conditions. The rea-
ings. So it is possible to protect the armature winding with son for
the low inductance is that the compole winding (or
the compole sensor and yet utilize both windings to their tem- compole
and compensating winding) with inductance LC #
perature limit. This limit is given by the thermal limit of the N2 on
one hand and the armature winding with inductance
C
utilized insulation material of the electrical winding (21). For L # N2
on the other hand excite air-gap fields, which oppose
A A
AC A C
perature plus 15 K additional temperature rise in hot spots, low
resulting inductance La for the whole armature circuit
resulting in a maximum temperature of 160#C, for at least (Fig.
13).
20,000 h, de facto for a much longer period of time. The corre-
sponding temperature control is designed for insulation class
LA = LAσ + LAm (36)
F usually with an operating temperature of 105#C
40#C #
145#C. Tripping miniaturized bimetal temperature relays and
LC = LCσ + LCm (37)
semiconductor components such as negative temperature co-
efficient (NTC) and positive temperature coefficient (PTC) ele-
La = LA − MAC + LC − MAC (38)
# #
N
ature measurement thermocouples or platinum resistance
La = LAσ + LCσ + 1 − C LAm (39)
N
elements are utilized. Good thermal contact between sensor
A
DC MACHINES 75
Armature winding LA
forced open-circuit cooling will have an armature circuit
ple, 10 min cycle time, 6 min load, and 4 min no-load period).
netic field that opposes the armature reaction field, yielding a low
resulting inductance. The field winding is not magnetically coupled
LIMITS OF DC MACHINES
with these windings (besides saturation cross-coupling).
Torque Limit
face, Fig. 1(b)] and l [iron stack length of the core, Fig. 1(b)],
Change of Load Current Eq.
(40), which is derived from Eqs. (6) and (7). Therefore a
M = αe d lAB (40)
seconds (22). One limit is given by avoiding saturation of the
2
compole magnetic circuit when the linearity between arma-
ture current and compole field, Eq. (34), would be lost with
Modern dc machines are designed for high utilization to keep
large armature currents, yielding too weak a compole field the
ratio of investment costs to installed power as low as pos-
BC . A second limit is the brush current density, for brushes
sible. A high output torque per motor volume can be achieved
4 2 by
increased current load and by increased air-gap induction.
can sustain current densities of about 20 # 10 A/m only for
several seconds without being overheated. The rate of change The
latter is limited due to the saturation of the rotor teeth
of the armature current can be very large if the compole mag- to
about 1 T. High current load with limits of typically 50 kA/
netic paths consist completely of laminated iron parts (dia/ m
(11) is accomplished by optimized cooling and by designing
dt 600IN per second, typically), as eddy currents in the iron the
machines for a higher steady-state temperature rise
core are suppressed. Older machines with massive iron yokes
through use of improved materials.
suffer from a delay of the compole flux change due to the in-
trinsic magnetic field of the eddy currents in the massive iron
Speed Limit
parts, causing too weak a compole field under transient load
A
high output power can be obtained by an extended speed
conditions (16). Thus with these machines the rate of current
76 DC MACHINES
urm
tor segments us must stay below 18 V (uncompensated ma-
Ia ∝ (41)
chines) and 20 V (compensated machines) to avoid flashover
n
I ~1/n
Springer, 1996.
a
M N M ~1/n
8. B. K. Bose (ed.), Modern Power Electronics—Evolution, Technol-
¨
n # n the output power P has to be reduced to limit the reactance
(ed.), Hutte-Elektrische Energietechnik, Band 1: Maschinen, 29.
¨
f
voltage of commutation.
ed., Berlin: Springer, 1978.
DC TRANSFORMERS 77
ANDREAS J. F. BINDER
Darmstadt University of Technology
POINTS.
DCT. See HADAMARD TRANSFORMS .
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3002.htm
} { { } }
HOME
ABOUT US
CONTACT US
●
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Diesel-Electric Generators
Abstract
Generator Applications
Environmental Aspects
AC Generators
DIESEL-
ELECTRIC POWER STATIONS
Worldwide,
interest in today’s electricity supplies is focused
on
centralized systems. Whereas electricity can be supplied
almost
everywhere in industrialized countries (ICs), thanks
to
national or international integrated grids, the developing
countries
(DCs) are far behind. Only 23% of global electricity
consumption
is generated in DCs, where it supplies about
43% of
humanity. A further 35% of the population in DCs has
no access
to electricity. Although integrated grids are increas-
ing in
global importance, they are not feasible in many DCs
for
technical and economic reasons. Decentralized electricity
supply is
thus of great interest in such places as well as, for
different
reasons, in ICs.
frame
which supports the entire system; (4) the synchronous
4 CHOICE
OF A DIESEL GENERATING SET
1
The
choice of a diesel generating set revolves around several
3 Class
of Service
Dimensions
Emergency Duty Continuous Duty Emergency Duty
Continuous Duty
tion, while an excess of dust may require a special air filter The
efficiency of the generator in fuel conversion depends on
for the engine. In some cases, temperature and dust problems
several factors, but mainly on the size of the machine and on
may necessitate the use of a special ‘‘filter-container’’ includ- the
percentage of the load applied to the generator with re-
ing all the equipment and using particular cooling systems spect
to the rated power (the efficiency is zero at no load and
which prevent the external air from entering into the con-
increases almost linearly reaching the highest value at a load
tainer. close
to the rated power). The rated fuel consumption is de-
0 200 400 600 800 1000 200 400 600 800 2000 200 400
600 800 3000 200 400 600 800 4000 200
100%
98
96
94
92
90
88 1
0°
86 C
2
) 84 0°
% C
( 82
t 3
0°
n 80 C
e
i 78
c 4
i 0
f
°
f 76 C
e
o 74 5
c 0°
g 72 C
n
i
t 70
a
r 68
e
D
66
64
62
60
58
56
54
52
50%
0 200 400 600 800 1000 200 400 600 800 2000 200 400
600 800 3000 200 400 600 800 4000 200
780 740 720 700 680 660 640 620 600 580 560
540 520 500 480 460
fined per produced kilowatt-hour when the engine runs at damage the
engine by corrupting the lubricant or, as a result
rated power. It ranges from 130 g/kW-h to 150 g/kW-h, for of passing
through the exhaust gas, may damage the turbine
big units, to 250 g/kW-h to 300 g/kW-h for small generat- (for the
turbo-charged engines), thereby increasing mainte-
ing sets. The lower fuel consumptions are typical of big nance costs
and lengthening the time that the equipment is
turbo-charged diesel units, and the smallest consumption out-of-order.
rate (130 g/kW-h) is characteristic of huge diesel units
(several MW of rated power) operating with a two-stroke cycle
with turbo-charger and low rotational speeds (as low as 100
rpm).
The diesel generator should operate at a level of power as )
100
n
f o
close as possible to the rated power, in order to achieve a o i
t
higher efficiency. Other operating conditions cause higher % p
( m
75
specific fuel consumption both for low load and for overload. n u
o s
i n
t
Figure 4 shows a typical curve of fuel burned even at no load, p o
c
m l
50
which represents the worst condition with respect to effi- u e
s u
n f
ciency. Particular care must be taken not to oversize the die- o d
sel set because both diesel engines and synchronous genera- c e
25
l t
e a
tors are characterized by low efficiencies at low load, thus u r
F e
h
increasing the specific fuel consumption (g/kW-h) and conse- t
0
quently the cost of the produced kW-h. The specific fuel con-
0 25 50 75 100
sumption is not critical when the generating set is used for
Electric power output
an emergency service while it is very critical when the plant
(% of the rated power)
1
3
k 1.10
k 1.00
t
t
n
n
e
e
i
i 0.90
c
c
i
i
f 1.00
f
f
f
e
e
o
o 0.80
c
c
g 0.90
g
n
n 0.70
i
i
t
t
a
a
r 0.80
r
e 30 40 50 60
e 0.60
D
D
modern
generators for diesel sets have a static or brushless
based on k coefficients can be applied. Figures 5, 6, and 7,
excitation system. In brushless alternators the excitation
give the k , k , and k derating coefficients, respectively, refer- power is
fed by an auxiliary alternator, coaxial with the main
1 2 3
iary
machine is fed by the main generator. The auxiliary sta-
tionary
field is fed by a power supply controlled by an auto-
ELECTRIC PERFORMANCE AND CONSTRAINTS
matic
voltage regulator.
Frequency
electric power at constant speed, implements the power con-
trol. The control also regulates the current of the excitation The
voltage frequency of the electric power delivered by the
field of the synchronous generator in order to maintain a con- diesel-
generator is usually 50 or 60 Hz depending on the local
stant output voltage at both variable load and power factor.
standard. Such a frequency is determined by the rotational
The terminal voltage of synchronous generators can be con- speed
and by the number of pole pairs of the diesel generator;
trolled by altering the current in the field winding. An auto- its
value is maintained fixed by means of a speed regulator
matic voltage regulator (AVR) compares the terminal voltage when the
generator works alone. The frequency in this case
of the generator with a voltage reference, and the error is does not
depend on the amount of power (torque on the main
processed to drive a rotating or static exciter. A need for accu- axis)
delivered to the load. If more than one generator is par-
rate modeling arises from the fact that the excitation system alleled
on a common bus to carry a load, it is necessary to
forms a feedback loop around the generator and we must be allow
each generator to vary its frequency in order to share
careful not to introduce instability in the generator system.
correctly the power between all generators.
An IEEE Committee has developed models to represent exci-
tation systems (4,5).
Frequency and Voltage Regulations
The
speed regulator of the diesel engines can be mechanical
Excitation Systems
or
electronic. Regulators can work in droop mode or isochro-
Very old types of excitation systems contain auxiliary direct- nous
mode. Droop mode regulators allow a frequency regula-
current (dc) machines with lower power than the generator; tion of
about #2% between no-load to full load in steady-state
these machines feed the dc power to the rotor windings of the
conditions. Isochronous regulators guarantee a narrower reg-
synchronous machine by two copper rings fitted on the shaft. ulation
bounded by #0.5% for the same load variation. Volt-
The current flows through a system of brushes. Almost all the age
regulation in steady-state conditions allows a variation of
#1.5%
for loads ranging from 0% to 100% and power factor
between
0.8 and 1. In transient conditions the voltage drop is
t to
return to within #3% of the steady-state voltage ranges
n
e
1.00
i from a
fraction of a second to several seconds.
c
i
f
f
e
o 0.90 Power
Factor
c
g
n
i Rated
powers (kilovolt-amperes) of generating units corre-
t 0.80
a
r spond to
a power factor of 0.8. At this power factor the active
e
D 0.70 rated
power of the synchronous generator corresponds to the
0 1000 2000 3000 4000
rated
power (kilowatts) of the diesel engine. The limit of
Height above sea level (m) power
of the diesel engine is the active power delivered to
Figure 6. Derating coefficient for a synchronous generator installed the load
while the limit of the synchronous generator is the
at altitude higher than 1000 m above sea level. apparent
power, as the limit is imposed by the current circu-
Stator Rotor
Stator
Main armature windings Main field Rectifier Auxiliary
armature windings Auxiliary Detection and control
a windings
field
windings
Automatic
voltage
regulator
maintenance service.
must decrease. At a power factor equal to 1 the apparent
power of the generator will equal the active power delivered
at a power factor equal to 0.8. At low power factor values Foundation
Block and Noise Reduction
with respect to the rated value, the active power delivered is Internal
combustion engines, because they are reciprocating
reduced, while the excitation system is overcharged, which machinery,
are a typical source of vibrations. Thus we must
sets a limit owing to the thermal conditions of the field wind- isolate
the generating sets by noise-reducing mufflers. High
ing. If the diesel engine is undercharged (the active power is levels of
noise insulation are normally expensive, and a
less than the rated value), this yields too low a temperature trade-off
is generally accepted between cost and acceptable
in the combustion chamber where poor combustion may cause level of
noise, depending on the specific application and
the formation of liquid particles. These phenomena are partic- according
to the noise reduction standards. In case of
ularly dangerous in turbo-charged units, because the particles ground
installation, a heavy block of concrete having two
mixed with the exhaust gas may damage the turbine. So it is to three
times the weight of the generating set will ensure
preferable to provide an automatic power factor regulator in a reliable
foundation. The maximal horizontal dimensions of
order to avoid low-power-factor operating conditions. the
generating set may be used to establish the horizontal
dimensions
of the foundation block, and consequently the
Single-Phase Loads block
height can be easily calculated because the block
relative to phase a:
fans, (3) inlet and outlet air windows, (4) synchronous genera-
tor, and (5) pumps, compressor, and other auxiliary ma-
v
= −r i + p λ (1)
chinery.
a s a a
λ
= −(L i + L i + L i ) + L i + L i + L i (2)
a aa a ab b ac c af f akd kd akq kq
lowing in the fuel supply system: (1) A valve to stop fuel flow
Lakd = Lakd sin θ (7)
must be placed outside the generator room; (2) include ade-
quate pipeline junctions to prevent fuel leakage; (3) the fuel
Lakq = Lakq cos θ (8)
A B
vb ib
b
+
ikd
v i
c c
c rs rs
nkd
+
nkq
ns ns
rkd
rkq
nf
ns
ikq
rf
r
s
if
v i
a a
a
vf
+
+ –
d s d q d
V = V sin δ = r I + X I (23)
d t s d q q
v = −r i + p λ (11)
0 s 0 0
V = V cos δ = −r I −X I + E (24)
# # # #
q r s q d d f
v = r i + p λ (12)
f f f f
0 = r# i# + p λ# (13)
The torque equation, expressed in terms of the terminal volt-
kq kq kq
age, is
0 = r# i# + p f λ# (14)
kd kd kd
where the prime on the variables and parameters of rotor
Te = 3 P EfVt sin δ + 1 # 1 − 1 # Vt2 sin 2δ (25)
2 2 X 2 X X
d d q
quantities means that they have been referred to the stator
by properly multiplying by the turns ratio of the windings.
where # is the rotor angle defined as the electric angular dis-
The flux linkages are expressed by
placement of the rotor with respect to the peak value of the ter-
# kg
Xq are, respectively, the direct and quadrature reactances.
# #
The detailed model of synchronous generators, described
λ = −L i + L (−i + i + i ) (16)
d # d md d kd f
by Eqs. (9) to (22), contains two state variables relative to the
λ0 = −L# i0 (17)
mechanical part (position and speed of the rotor) and six state
# # # #
variables relative to the electromagnetic part. This model is
λ = L i + L (+i − i ) (18)
kq #kq kq mq kq q
used when there are large variations of the variables. In cer-
# # # # #
tain circumstances it would be unnecessary to adopt this
λ = L i + L (−i + i + i ) (19)
kd #kd kd md d f kd
leakage inductances.
mation the rate of change in the field flux linkage can be ig-
The electromagnetic torque can be calculated by
nored. Reduced models (8) can be used for electric transients
Wc
tain accurate initial conditions. Various methods for dealing
mep = (29)
V
with saturation exist, and the final choice is often governed
d
by the availability of reliable machine data.
The
capacity of an engine to convert the energy contained in
fuels
into mechanical energy can be evaluated by the fuel con-
DIESEL ENGINE MODELING
version efficiency, given by the ratio of the work per cycle di-
vided
by the energy of the fuel mass:
Diesel engine modeling is very complex due to the difficulties
in describing the engine internal processes by equations (2,3).
Wc P
Two basic types of models are used to predict performance
ηf = m QHV = pm QHV (30)
f f
and emissions of engines. Models based on equations derived
by the energy conservation principle are known as thermody-
where
QHV is the heating value of the fuel, mf is the mass of
namic-based models, while models based on equations de-
fuel,
and pm f is the rate of mass per unit time. The intake
scribing the fluid motion are named fluid-dynamic models.
system
efficiency can be evaluated by the volumetric effi-
Basic assumptions of the former models are knowledge of the
ciency
η η nV Q ρ
f v d HV a,i
m
#
Wc = p dV (26)
P = 2 a (32)
From knowledge of the work, the masses of fuel and air in- and
the torque
ducted, and the engine speed, the performance parameters
m
can be derived. The power P developed per cylinder by a four-
f
η η V Q ρ
stroke diesel engine is expressed by
TD = P = f v d HV a,i ma (33)
2πn 4π
Wcn
P = (27)
2
Finally, substituting Eqs. Eqs. (30) and (31) into Eq. (29), we
get
where n is the rotational speed (rev/s) of the crankshaft.
m
The above is gross power, which differs from the brake
mep = η η Q ρ f (34)
f v HV a,i
power by the power required for friction, for engine accessor-
ma
e
Speed Governor
Engine
reference
Ki K2 pm f
–s τ T – 1 ω
K + K
e 1 K
+ – p s 1 + sτ 2 1
mep T + Js
PI regulator Actuator
BIBLIOGRAPHY
ANTONIO LEOTTA
UBALDO NOCERA
Alternative Energies and Energy
Saving
ANGELO RACITI
University of Catania
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3003.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Diesel-Electric Power Stations
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3003.htm17.06.2008 16:36:15
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3013.htm
} { { } }
HOME
ABOUT US
●
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Electric Machine Analysis and Simulation
● Email this page Enter words or
Standard Article
Abstract
Modeling of Saturation
Other Machines
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3013.htm17.06.2008 16:36:33
d–q–0
REPRESENTATION OF THREE-PHASE QUANTITIES
The
highly coupled nature of induction and synchronous ma-
chines
had led to the use of artificial variables rather than
actual
(phase) variables for the purpose of simulation as well
as for
visualization. The essence of the nature of the transfor-
mation
of variables that is utilized can be understood by ref-
erence
to Fig. 1 which shows three-dimensional orthogonal
axes
labeled a, b, and c. Consider, for instance, the stator cur-
rents
of a three-phase induction machine which is, in general,
made
up of three independent variables. These currents
(phase
variables) can be visualized as being a single three-
ia + ib + ic = 0 (1)
This
plane, the so-called d–q plane, is also illustrated in Fig.
1.
Components of the current vector in the plane are called
the d–q
components, while the component in the axis normal
to the
plane (in the event that the currents do not sum to
zero)
is called the zero component. When the phase voltages
and
phase flux linkages also sum to zero, as is the case with
most
balanced three-phase loads (including even a salient
pole
synchronous machine), this same perspective can be ap-
plied
to these variables as well. The components of the phase
# 1 − 1 − 1#
# # # # 2 2 # # #
f f
q 2 # 3 √3 # a
# # # # # #
f = f (2)
# #
d 0 − b
# #
f f
0 # 1 1 1 # c
While an electrical machine exists for the bulk of its time in
√2 √2 √2
# 3 # reference
frame. In general, it is not necessary to define rotat-
# # 0 0 # #
f f
q # 2 # a ing axes
to rotate synchronously with one of the vectors but
# # # # # #
# #
f = 1 1 f (3) simply to
define a general rotating transformation which
# # # #
d b
# √ √ #
0 −
f 0 # 2 2 # f c transforms
the phase variables rotating axes on the d–q
0 0 0 plane,
# #
# #
# #
shows the location of the various axes when viewed from the
# #
f (θ )
3 3 f
# # # #
q
a
d–q plane. Note that the projection of the a-phase axis on the
2 # 2π 2π #
# #
# # # #
f (θ ) =
sin θ sin θ − sin θ + f
d–q plane is considered to be lined up with the q-axis (the a- d
3 # # b
# #
3 3
f
f
# #
phase axis corresponds to the magnetic axis of phase a in the 0
1 1 1 c
case of an electrical machine). The other axis on the plane is,
√2 √2 √2
by convention, located 90# clockwise with respect to the q-
(5)
axis. The third axis (necessarily normal to the d–q plane) is
chosen such that the sequence d, q, 0 forms a right hand set.
Other notation, using symbols #, # (Clarke’s components), is For
completeness, the zero component is also given. The
sometimes used to denote these same variables. Also, with scale
change of #2/3 has also been included. The angle # is
the transformation shown, the d-axis is located 90# counter- the angular
displacement of the vector on the d–q plane mea-
clockwise with respect to the q-axis. These two axes are some- sured with
respect to the projection of the a-axis on this
times interchanged so that the reader should exercise caution plane.
Since the same scale change has been made for all
when referring to the literature. three
components, the zero component uses somewhat uncon-
When balanced sinusoidal three-phase ac voltages are ap- ventional
scaling. Specifically, Fortesque’s scaling for this
plied to such a load, it can be shown that the phase voltage component
was selected as
vector traces out a circle on this d–q plane with radius #3/2
V where V is the amplitude of the phase voltage. The vector
1
pk pk
f = (f + f + f ) (6)
rotates with an angular velocity equal to the angular fre-
0 3 a b c
# #
voltage
variables have been scaled by #2/3.
# # # #
1 0 0
f q f a In vector
notation, Eq. (5) can be written as
# #
1 1
# # # √ √ # # #
f d = 0 − f b (4)
# 3 3#
f T f
f 0 0 0 0 f c
ff qd 0 = TTqd 0 (θ )ff abc (7)
# #
# 2π # # 2π #
# #
3 3
b-axis
2 # # 2π # # 2π # #
T
# #
TT (θ )
= sin θ sin θ − sin θ + (8)
qd 0
3 # 3 3 #
# #
# #
1 1 1
√2 √2 √2
0-axis
(Normal to q-axis a -axis The
transformation Tqd0(#) can, for convenience and for com-
paper) putational
advantage, be broken into two portions, one of
which takes
variables from physical phase quantities to non-
rotating d–
q–0 variables (stationary reference frame) and
then from
nonrotating to rotating d–q–0 variables (rotating
reference
frame). In this case, one can write
d-axis
c -axis
f T f R T f
Figure 2. Physical and d–q–0 axes when viewed on the d–q plane.
ff qd 0 = TTqd 0 (θ )ff abc = RR (θ )TTqd 0 (0)ff abc (9)
where
( – )ω ω λ′
ωλ L L ′ r qr
r qs ls lr r ′
s r
– + + –
# 2 − 1 − 1#
+ i i′ +
# #
3 3 3
ds dr ′
v
L v
ds
# #
1 1
m dr
# #
√ √
T 0 −
TT (0) = (10)
# #
qd 0 3 3
– –
# #
√ √ √
# 2 2 2 #
d -axis equivalent circuit
3 3 3
( – )ω ω λ′
ωλ L L ′ r dr
r ds ls lr r ′
and
s + – – + r
+ +
# #
i i′
cos θ − sin θ 0
v qs L qr v′
qs m qr
# #
R sin θ cos θ 0
RR (θ ) = (11)
– –
0 0 1
s r
+ +
−1 −1 −1
+ +
f T f T R f
ff abc = TTqd 0 (θ ) ff qd 0 = TTqd 0 (0) RR (θ ) ff qd 0 (12)
i0s No i′0r ′
v0s v 0r
coupling
where
– –
# #
1 sented
in a rotating reference frame.
cos θ sin θ √
# #
2
# # # #
(13)
# 2π 2π 1 #
# #
= cos θ − sin θ − √
# 3 3 #
# #
2
Except fornotational differences, the parameters in this
# # 2π # # 2π # 1 #
circuit are essentially the same as the conventional per phase
cos θ + 3 sin θ + 3 √
equivalent circuit. That is, r , L , r#, L #, and L correspond to
2
s ls r lr m
the
per phase stator resistance, stator leakage inductance, ro-
# 1 # tor
resistance, rotor leakage inductance, and magnetizing in-
1 0 √
# #
2
ductance, respectively [typically labeled as R , L , R , L , and
1 1 2 2
# √ #
# #
−1 3 T 1 3 1 L#
(or Lm) respectively]. In most cases, the impressed rotor
# #
T T − −
√
TTqd 0 (0) = TT (0) = (14)
2 # 2 2 2 #
voltages are identically zero (squirrel cage machine) and will
# √ #
# #
1 3 1 be
assumed henceforth herein as zero. The zero sequence cir-
−
√
2 2 2 cuits
are included for completeness but are seldom necessary
and
will now also be omitted from further consideration. The
# #
# #
convenience.
d–q–0 REPRESENTATION OF THREE- The
voltage generators in the circuit represent speed volt-
PHASE INDUCTION MACHINES ages
which appear due to the fact that the circuit is being
solved
in a rotating reference frame. This term is to be ex-
The benefits of visualizing three-phase variables as a vector pected
of any physical system represented in a rotating
represented in a nonphysical coordinate system become ap- frame.
(Recall from mechanics the # # r term representing
parent only when the system equations of a coupled three- the
relative velocity of a stationary point in rotating system.)
phase magnetic component such as a reactor, transformer, or The
speed voltages in the stator portion of the circuit are pro-
motor are represented in terms of these newly defined vari-
portional to the reference frame angular velocity #(# #
ables. The equivalent circuit of an induction motor repre- d#/dt)
since the circuits, themselves, are stationary. The
sented in a rotating reference frame is shown in Fig. 3. Here, speed
voltages in the rotor portion are proportional to # # #r
the second subscripts ‘‘s’’ and ‘‘r’’ are used to denote ‘‘stator’’ since
the rotor circuits, themselves, are rotating at an electri-
and ‘‘rotor’’ quantities, respectively. The enormous simplicity cal
angular velocity of # (# # d# /dt, where # is the angular
r r r r
T − T = # 2 # J dωr (30)
dλ
em load P dt
v = r i + qs + ωλ (16)
qs s qs dt ds
dλ
where Tload is the load torque, and J is the inertia in SI units.
ds
v = r i + − ωλ (17)
ds s ds dt qs
dλ#
SIMULATION OF INDUCTION MACHINE USING
# # # qr + (ω − ω #
v = 0 = r i + r )λ (18) FLUX
LINKAGES AS STATE VARIABLES
qr r qr dt dr
dλ#
# # # dr − (ω − ω #
Since the differential equations of the machine, Eqs. (16–19),
v = 0 = r i + r )λ (19)
dr r dr dt qr
contain mixed variables (i.e., flux linkages and currents), ei-
ther
of these two quantities could be eliminated from the dif-
where the flux linkages are defined by
ferential equations by means of the algebraic relations, Eqs.
(20–
25). The traditional approach to simulation is to consider
λqs = Lls iqs + λmq (20) the
flux linkage as the state variables and currents as depen-
# # #
λ − λ
λdr = Llr idr + λmd (23)
iqs = qs mq (31)
ls
and
λ − λ
ids = ds md (32)
ls
λmq = Lm (iqs + i# ) (24)
qr
λ# − λ
#
i# = qr mq (33)
λmd = Lm (ids + idr ) (25)
qr L#
lr
λ# − λ
Note that while not necessary to be defined explicitly, the mu-
i# = dr md (34)
dr L#
tual (air gap) flux components mq and md have been included
lr
to aid the simulation process.
1 # λ# #
λ
The torque produced by the machine can be identified as
λmd = ds + dr (35)
1 1 1 L L#
the power consumed by the voltage generators in Fig. 3 di-
L + L# + L ls lr
vided by the actual rotor speed. Multiplying these voltage
ls lr md
generators by their respective currents,
1 # λ# #
λmq = qs + qr (36)
1 1 1 L L#
# # # #
T = 3 P 1 [ω(λ i − λ i )
Lls + L# + Lmd ls lr
em ds qs qs ds
lr
2 2 ωr (26)
+ (ω − ωr )(λ# i# − λ# i# )]
dr qr qr dr
These results can be inserted into the differential equations.
# # # #
3 P Upon
solving for the time derivative terms and integrating,
= (λ# i# − λ# i# ) (27) the
result is
2 2 qr dr dr qr
R # rs #
ls
# # # #
# # #
3 P
r
T = (λ i − λ i ) (28)
# r # #
em ds qs qs ds
λ = (λ − λ ) − (ω − ω )λ dt (39)
2 2
qr L# mq qr r dr
lr
R # r# #
or
λ# = r (λ − λ# ) + (ω − ω )λ# dt (40)
dr L# md dr r qr
lr
# # # #
# #
3 P
T = L (i# i − i# i ) (29)
P 1
em m dr qs qr ds
ω = (T − T ) dt (41)
2 2
r 2 J e load
vqs 1/p λ qs
λ ds x (Eq. 37)
i
qs
(Eq. 31)
λ
mq
ω (Eq. 36)
i′
qr
(Eq. 33)
x
λ dr′ λ ′
1/p qr
(x – y )*z
z (Eq. 39)
y
λ
qs
T
i em
ω qs
r (Eq. 28)
λ ds
ω
1/p r
ids
(Eq. 41)
λ (Eq. 38)
qs x
ids
(Eq. 32)
λ md
ω (Eq. 35)
idr
′
(Eq. 34)
x
λ ′
qr 1/p λ dr′
(x – y )*z
z (Eq. 40)
y
even if the
load is a symmetrical three-phase induction ma-
a a
chine or even a salient pole synchronous machine.
+
The phase voltages can now be solved in terms of the
eag Z (p )
known source voltages as
–
– +
g s
2 1 1
– ebg b Z (p ) Z (p )
vas = 3 eag − 3 ebg − 3 ecg (48)
e
cg
1 2 1
c + c
vbs = − 3 eag + 3 ebg − 3 ecg (49)
1 1 2
Figure 5. Three-phase wye connection having source voltages deter-
vcs = − 3 eag − 3 ebg + 3 ecg (50)
# 2 − 1 − 1#
v e
as ag
# 1 2 1 #
# #
# # # #
v e Se
= v = = SeSe (51)
# #
v = Z (p )i
bs abcs − − bg abcg
# #
as as
3 3 3
v e
vbs = Z (p )ibs
cs # 1 − 1 2 # cg
3 3 3
vcs = Z (p )ics
denoted by Eq. (51) and Eq. (10) can be readily solved to form
# #
bs bs
2 1 1
− −
3 3 3
Hence, when the sum of the three load currents equals zero,
# #
T S # 1 1 #
TT (0)SS = (52)
the sum of the phase voltages also sum to zero. Equation (45)
qd 0 # #
0 −
becomes
# √3 √3 #
0 0 0
1
vsg = 3 (eag + ebg + ecg ) (47)
The last row of zeros show that the zero sequence component
of
voltage is zero. That is, the zero component of voltage is
While this result has been illustrated for simple passive im-
impressed across the open circuit between points s and g and
pedances, it can be shown that the same conclusion is true
not across the zero sequence circuit of the machine itself
i s , i s
qs ds
i , i , i
as bs cs
i i
qs, ds
–1 –1
Tqdo (0) R ( )θ
(Eq. 14) (Eq. 15)
External
s s
Induction
(Fig. 4)
sinθ cosθ
e , e ,
e v , v , v
ag bg
cg as bs cs
– +
1/p x 1/p x
(shown in Fig. 3). In the event that the three source voltages
Air gap line
ω ]
also sum to zero, we have, finally,
[slope = L m (unsat)
# #
ω L l
1 mag
1 0 0
Vl-n
# #
# #
1 1
T S
TT (0)SS = (53)
# #
0 −
√ √
qd 0
# 3 3 #
Vl-n, Vgap V
0 0 0
(V, rms) gap R 1 j Lω 1 j Lω 2
+ +
In general, the reference frame velocity can be selected to
Vl-n Vgap j Lω m R /S
be any explicit or implicit function of time. The speed of the
2
reference frame is typically chosen to best suit the problem
– –
under investigation. For example, if the simulation requires
I mag (A , rms)
modeling piecewise linear or nonlinear elements such as
semiconductor switches, then the reference frame must be
(a)
constrained to rotate either with the stator or the rotor de-
pending upon where the switches are located. When a simple
balanced three-phase sinusoidal operation is investigated, a
synchronous reference frame can be used and often adds in-
Air gap line ∆λ
sight into the problem being investigated. In motor control
m
45°
pass directly through the block without modification. In Fig.
6, the reference frame velocity # is shown as coming from the
= L I (V/s , pk )
λ
external system as would be the case if the synchronous volt-
m(unsat) m (unsat) mag
age rotating reference frame were used. Krause and Thomas
(b)
(7) give an excellent treatment of simulation techniques to be
employed when series connected semiconductor switches open
Figure 7. (a) No-load saturation curve. (b) Derived curve.
and close, producing temporary open circuit conditions in the
phases.
λ
ing stator resistance, the slope of a line drawn from the origin
#λmd = md (unsat) #λm (54)
to a point on the curve is proportional to the sum of the stator
λm (unsat)
leakage plus magnetizing reactance # (L # L ), or (# (L #
e 1 m e ls
λ
Lm) in d–q notation) where #e is the angular frequency of the
#λmq = mq (unsat) #λm (55)
λ
source voltages. If the leakage reactances of the machine have
m (unsat)
been measured by locked rotor test or calculated, the voltage
drop due to magnetizing current flow in the stator leakage
Saturation in the q-axis can be incorporated if Eqs. (20) and
inductance branch can be subtracted from the terminal volt-
(22) are modified to form
age to obtain the voltage at the air gap. The slope of the air
gap voltage versus magnetizing current is clearly the mag-
λqs = Lls iqs + λmq (sat) = Lls iqs + λmq (unsat) − #λmq (56)
netizing reactance # L . The slope of the linear portion line
e m
λ# = L# i# + λmq (sat) = L iqr + λmq (unsat) − #λmq (57)
(air gap line) yields the unsaturated value of #eLm(unsat) . If the
qr lr qr lr
λ
md(sat)
– ∆ λ md λmd (unsat)
m (unsat)
* 1 1
L +
m (L L ′ )
ls lr
*
L
λ ds m
Lls
+
+
(⋅)2
2
L * +
λ λ m (unsat)
λ ′ m
md (unsat)
dr ′
Llr
∆ λ m
+ λ
m (unsat) ∆ λ
+
*
λ
L
m (unsat)
λ m
qs Lls +
λmq (unsat)
(⋅)2
+
*
L
λ ′ m
qr L ′
1 1
lr
*
L +
m (L L ′ )
ls lr
λ
+
Figure 8. Block diagram for the proce- mq(sat)
∆ λ md λmd (unsat)
λmq (unsat) = + 1 + 1
is given in (10).
Lm Lls L# (58)
lr
# λ# # 1 # #
λ 1
qs qr
SATURATION MODEL WITH CURRENTS AS STATE VARIABLES
L + L# + L + L# #λmq
ls lr ls lr
ω λ L L
1
r ds mr 1 mr2
L∗ = (59)
m
1 1 1
+ – +
+ +
rs Lls
Lm Lls L#
Llr 1 Llr2 Llr3
lr
ds L
rr 1 rr2 rr3
–
Another very important phenomenon in squirrel cage induc-
d -axis
tion machine concerns the uneven distribution of currents in
ω λ L L
the rotor bars, termed deep bar effect. Because a filament of
r qs mr 1 mr2
current experiences a greater inductance at the bottom of the
+ + –
bar than on the top portion, the current tends to rise to the
rs Lls
top of the bar facing the air gap, resulting in greater torque
Llr 1 Llr2 Llr3
v
as well as higher losses at a given slip frequency. This phe-
qs Lm
nomenon is frequently used to improve the starting perfor-
rr 1 rr2 rr3
mance of a squirrel cage machine since the effect is greatest
–
under the starting condition due to the fact that the bar re-
actance is greatest at this point.
q-axis
λ − λ
glected prior to this time. In reality, prior to 1981, flux link-
iqs = qs mq (70)
ages rather than currents were used to model core saturation
Lls
primarily to avoid the difficulties addressed in these refer-
λ# − λ
ences. It has been demonstrated that the solution of the two
i# = qr mq (71)
qr L#
methods are identical (8). Because of the complexity of the
lqr
simulation (involving inversion of a 4 # 4 matrix every time
λ − λ
step), the method is not recommended.
ids = ds L md (72)
ls
λ# − λ
i# = dr md (73)
SIMULATION OF SYNCHRONOUS MACHINE
dr L#
ldr
λ# − λ
Wound field and permanent magnet synchronous machines
i# = f r md (74)
f r L#
can be modeled by use of the same d–q–0 transformation
lfr
# #
md md L L# L#
eled in a reference frame rotating with the asymmetry (i.e., a
ls ldr lfr
reference frame rotating with the rotor) in order to simplify
# iqs i# #
the coupled equation which exists in phase variable form.
λmq = L∗ + qr (76)
mq L L#
The d–q–0 differential equations depicting behavior of a
ls ldr
where
dλ
v = r i + qs + ω λ (60)
1
qs s qs dt r ds
L∗ = (77)
md 1 1 1
dλ
+ # + #
ds
L L L
v = r i + − ω λ (61)
md ldr lfr
ds s ds dt r qs
1
dλ#
L∗ = (78)
0 = r# i# + qr (62)
mq 1 1
qr qr dt
+
L L#
#
mq lqr
dλ
0 = r # i# + dr (63)
dr dr dt
ω λ L′ r′
shorted squirrel cage constructed in much the same manner
r ds lfr fr
as the squirrel cage of an induction machine (often labeled as
+ – + +
kd and kq). The last equation corresponds to the excited rotor
rs Lls
L′
rotor circuits have been referred to the stator by the appro-
v ldr v′
ds fr
priate turns ratio. Note that Eqs. (60–63) are identical in
Lmd
form to the induction motor equation except that the speed of
r′
dr
the reference frame # has been set equal to the speed of the
– –
rotor #r , and that the rotor has asymmetry (rqr # rdr ).
d -axis
The flux linkages are related to the currents by
ω λ
r ds
λqs = L iqs + Lmq (iqs + i# ) = L iqs + λ # (65)
ls qr ls mq
+ + –
# # # # # #
rs Lls
λqr = Llqr iqr + Lmq (iqr + iqs ) = Llqr iqr + λmq# (66)
L′
λ = L i + L (i + i# + i# ) = L i + λ (67)
v lqr
ds ls ds md ds dr f r ls ds md
qs
mq
λ# = L# i# + L (i + i# + i# ) = L# i# + λ (68)
dr ldr dr md ds dr f r ldr dr md
r′
qr
λ# = L# i# + L (i + i# + i# ) = L# i# + λ (69)
–
f r l f r f r md ds dr f r l f r f r md
q-axis
An equivalent circuit of this machine can be established from
Figure 10. d–q–0 equivalent circuit of a wound field synchronous
these equations as shown in Fig. 10. These equations can be
machine.
v
qs
λ
1/s qs
λ ds
x (Eq. 60)
qs
(Eq. 70)
mq
(Eq. 76)
i ′
qr
(Eq. 71)
ω
r
λ ′
1/s qr
(Eq. 62)
qs
i em
qs (Eq. 28)
λ ds 1/s ω r
λ
(Eq. 61)
qs
x
ids
(Eq. 72)
λ md
(Eq. 75)
i ′
dr
(Eq. 73)
λ ′
1/s dr
(Eq. 63)
i ′
fr
(Eq. 74)
λ ′
1/s fr
v ′
fr
(Eq. 64)
λ
md (sat) – ∆ λ md
* 1 1
L +
L * m (L L ′ )
ls l dr
λ ds m
Lls ∆ λ md
+ + λ md(unsat)
∆ λ m
+
*
L +
λ ′ m
dr λ
L ′
md(unsat)
ldr
*
L
λ ′ m
fr
L ′
Figure 12. Saturation model for a salient pole
lfr
Three books that treat the overall subject of analysis and sim-
winding to yield the equivalent of a balanced two-phase in-
r ω λ L L
5. P. C. Krause, T. A. Lipo, and D. P. Carroll, Applications of analog
s r ds ls lmag
L L′ L
6. C. H. Thomas, Discussion of ‘‘Analogue Computer Representation
md ldr mag
r′
1184, 1956.
dr
–
7. P. C. Krause and C. H. Thomas, Simulation of symmetrical in-
1038–1053, 1965.
T. A. LIPO
University of Wisconsin
----------------------- Page 104-----------------------
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3014.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Electrochemical Electrodes
Abstract
Summary
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3014.htm17.06.2008 16:36:51
ELECTROCHEMICAL ELECTRODES of
the fuel and oxidant gases is generally counter or cross flow
combined to produce
appreciable amounts of electricity and
H2O CO2 are
joined by interconnects. Because of the configuration of a
flat
plate cell, Fig. 2, the interconnect becomes a separator
tubular solid oxide fuel cell is a special case and has a slightly
lyte film that wets portions of the porous electrode and react
Separator plate
electrochemically on the electrode surface. The balance of fuel
### matrix
multaneously reducing the cost of the materials and system.
Composite
# cathode
Electrode Reactions
# substrate
###
In a fuel cell, the anodic and cathodic reactants are fed into
#
reaction at the anode yields electrons, shown in Table 1.
###
These electrons are transported through the external circuit
and reach the cathode where they are transferred to the ca-
# Separator plate
thodic reactants. The cycle is completed by the transport of
#
ions from one electrode to the other through the electrolyte.
ferent fuels and oxidants for the various fuel cell types are
2 2
4 2
2
and reaches the reaction zone where gas, the electrolyte, and
ders with a very high specific area (#100 m /g) or
carbon
the solid conducting structure meet. This is the ‘‘three-phase
(1000 m2/ g).
powders are light, have a large surface area, are suitable for
the surface structure of the electrodes and the current op-
Cell Reaction
Nernst Equation
H # ##O # #H O E #
E 0 # (RT/2F )ln[P /P ] # (RT/2F )ln[P1/2]
2 2 2
H H O O
2 2 2
H # ##O # CO (c) # E #
E 0 # (RT/2F )ln[P /P (P ) ]
2 2 2
H H O CO a
2 2 2
H O # CO (a)
# (RT/2F )ln[P1/2](P ) ]
2 2
O CO c
2 2
CO # ##O # CO E #
E 0 # (RT/2F )ln[P /P ] # (RT/2F )ln[P1/2]
2 2
CO CO O
2 2
CH # 2O # 2H O # CO E #
E 0 # (RT/8F )ln[P /P 2 P ]
4 2 2 2
CH H O CO
4 2 2
# (RT/8F )ln[P 2 ]
Transport Processes
ηohm = iR
5 5 2
ated at a pressure of 4.0 10 to 4.4 10 N/m and a tempera-
the PAFC is capable of operating at the high end of the acid
ture of about 90#C. The high performance at low temperature
temperature range (100 to 220#C). In addition, the use of con-
was achieved using Teflon-bonded electrodes with pure, high-
centrated acid (#100%) minimizes the water vapor pressure,
loading noble metal catalysts. The anode consisted of pure so
water management in the cell is not difficult. The matrix
noble metal black (80% platinum, 20% palladium) on a silver-
universally used to retain the acid is silicon carbide (10), and
plated nickel screen. The cathode had 90% gold with 10% the
electrocatalyst in both the anode and cathode is Pt.
platinum on a gold-plated nickel screen. Later versions, which
The porous electrodes used in this system are described
are proprietary, of this system had improved materials, differ-
extensively in the patent literature (11). These electrodes con-
ent operating parameters, less noble metal loading, and elec-
tain a mixture of the electrocatalyst supported on carbon
trolyte reservoirs. However, this system is still far from being
black and a polymeric binder such as PTFE (30 to 40 wt%).
cost-effective for everyday applications (2). The
polymer binds the carbon black particles together to form
More commonly, the systems are built with catalyzed car- an
integral porous structure. A carbon paper substrate is also
bon on nickel screen current collectors or porous nickel sub-
used and serves a dual role as a structural support for the
strates with a hydrophobic backing. Platinum is the key in-
electrocatalyst layer as well as the current collector. A typical
gredient in the catalyst. Because these systems operate at
carbon paper used in PAFCs has an initial porosity of about
such low temperatures, plastics can be used for the cell
90%. This wet-proof carbon paper should contain macropores
frames. of
3 to 50 #m diameter, with a median pore diameter of about
More recent advances involve the use of a porous gas diffu-
12.5 #m. It should also have micropores with a median pore
˚
sion layer with 60 wt% PTFE on carbon, a nickel wire mesh,
diameter of about 34 A for gas permeability. The composite
and an active layer of 12.24 wt% cobalt tetraphenylporphyrin
structure consisting of a carbon black/PTFE layer on carbon-
(CoTPP) and 12 wt% PTFE on carbon for the cathode. The
paper substrate forms a stable, three-phase interface in the
catalyst for the anode consists of 8.3 wt% platinum, 8.3 wt%
fuel cell, with H3PO4 electrolyte on the electrocatalyst side
palladium and 17 wt% PTFE on charcoal (6). Additional re- and
the reactant gas environment on the other side of the
search is being conducted to improve the catalyst and elec-
carbon paper (12).
trodes (7).
The individual cells are physically separated by a bipolar
fuel cell stack. This plate also contains the gas channels for
PHOSPHORIC ACID FUEL CELL ELECTRODES
introducing the reactant gases to the porous electrodes and
plated Tantalum. During the past two decades, the cost of the
(14).
materials and the requirement for Pt has decreased. A major
Despite the progress in the development of cell components
breakthrough was the development of carbon blacks and
over the past decade, there is still significant research under-
graphite for cell construction materials (8,9). At that time, it way
on alternative and component improvements that have
was shown that carbon black and graphite were sufficiently the
potential of cost reductions and performance improve-
stable to replace the more expensive gold-plated tantalum cell
ments. This includes improved catalysts, catalyst supports,
hardware. Because the carbon black had high surface areas, and
electrolyte management. Alternative support materials
a dramatic reduction in Pt loading could be obtained with lit-
with improved physicochemical properties are desired to in-
tle or no reduction in performance. Appleby (8) reports that
crease the flexibility and remove the current temperature and
‘‘without carbon, a reasonably inexpensive acid fuel cell would
electrical constraints on PAFCs. A large number of materials
be impossible, since no other material combines the necessary
offer potential (15).
properties of electronic conductivity, good corrosion resis-
tance, low density, surface properties (especially in high-area
form) and above all, low cost.’’ Table 3 shows the evolution of
MOLTEN CARBONATE FUEL CELL ELECTRODES
the cell component technology for these fuel cells.
Concentrated phosphoric acid is used for the electrolyte in
MCFC is an intermediate temperature fuel cell that operates
this fuel cell, which operates between 150 and 220#C. At lower
between 600 and 700#C. The chronology of component devel-
temperatures, phosphoric acid is a poor ionic conductor, and
opment is shown in Table 4. Since the mid-1970s, the materi-
CO poisoning of the Pt electrocatalyst in the anode becomes als
for the electrodes and electrolyte structure have remained
more severe. The relative stability of concentrated phosphoric
essentially unchanged. The primary developments have been
acid is high compared to other common acids; consequently, the
modification of the electrolyte and improvements in the
Component 1965
1975 Current
Anode PTFE-bonded Pt
PTFE-bonded Pt/C PTFE-bonded Pt/C
black
Cathode PTFE-bonded Pt
PTFE-bonded Pt/C PTFE-bonded Pt/C
black
2
The electrolyte in this fuel cell is usually a combination of
alkali (Li, Na, K) carbonates, which is retained in a ceramic
The dissolution of NiO has been correlated to the acid-base
matrix of LiAlO2 . The fuel cell operates where the alkali car-
properties of the electrolyte. The basicity of the electrolyte is
bonates form a highly conductive molten salt, with carbonate
directly proportional to the CO2 partial pressure (log PCO ). In
Component 1965
1975 Current
3–6 #m pore
50–60% porosity
0.5–1.5 mm thick
0.1–1 m /g
Cathode Ag2O or lithiated NiO
Lithiated NiO Lithiated NiO
7–15 #m pore
70–80% porosity
0.5–0.75 mm thick
0.5 m2/g
and #-LiAlO2
2 2
10–20 m /g 0.1–12 m /g
Electrolyte 52% Li, 48% Na
62% Li, 38% K 62% Li, 32% K
43.5% Li, 31.5%
60–65 wt% Li 50% Li, 50% Na
Na, 25% K
50 wt% Li
Fabrication process Paste
Hot press ‘‘tile’’ Tape-cast
Printing process
0.5 mm thick
metals (such as Ti, Co, Cu, and Al) have been incorporated for
each component. The air electrode provides the inner tube
in anode structures (16,18,26–28). Copper-based materials for
the cell package. This tube is generally extruded and sin-
have also been evaluated as a lower cost alternative
tered with one end closed. The anode is generally deposited,
(26,29,30). to
a thickness of about 150 #m, using a slurry process. It is
An acceptable cathode material must have good electrical
important that these electrodes permit only electronic conduc-
conductivity and structural strength. It must also have a low
tion and interdiffusion of ionic species at operating tempera-
dissolution rate in molten alkali carbonates to avoid precipi-
ture (1000#C). In addition, they must be resistant to thermal
tation of the metal in the electrolyte structure. Over 50 ma-
cycling effects. That is, they must have closely matched ther-
terials have been examined to identify suitable materials. mal
expansion coefficients and little or no grain growth.
Perovskite-type compounds (Ln M M
O where Ln is a The resistivities for the anode are
critical to the perfor-
1#x x 3
coated NiO have also been shown to have low solubility in the
#C). This enables the use of metallic interconnects, relaxes
electrolyte. However, it has a very high resistivity (about 300
some cell construction constraints, reduces internal electrical
#
cm). Unfortunately, this has a high material cost which
resistance, and potentially can reduce the cost of materials in
may prevent large scale implementation (31–33). the
balance of plant. Planar cells require gas-tight seals as
Tubular
Interconnection
Electrolyte
Air
electrode
Fuel
flow
Air flow
Fuel electrode
Anode
Cathode ### ###
Interconnect
Anode
### ###
Electrolyte
Cathode
###
### ###
###
### Fuel ###
### ###
###
Oxidant
###
Fuel ###
### Cathode
###
Anode
###
Oxidant
Electrolyte
###
Cathode
Component 1965
1975 Current
cermet
praesodymium oxide
12 10#6 cm/cm
#Ca
sinter
Electrolyte Yttria-stabilized ZrO2 Yttria-
stabilized ZrO2 8 mole% yttria-
stabilized ZrO2
EVD
EVD, dry press, calendar,
tape cast
Cell interconnect Pt Mn-
doped cobalt 10 mole% Mg-doped
ern the potential of the various cells to be used for different 22.
D. A. Shores and P. Singh, Proc. Symp. Molten Carbonate Fuel
applications.
Cell Technol., Pennington, NJ: The Electrochemical Society, 1984,
p. 271.
23.
K. Kinoshita, Critical Survey on Electrode Aging in Molten Carbon-
BIBLIOGRAPHY
ate Fuel Cells, ANL 79-55, Argonne National Laboratory, Illi-
nois, 1979.
397.
Practice, H. C. Maru, T. Katan and M. G. Klein (eds.). Pen-
nington, NJ: The Electrochemical Society, 1984, p. 163. 27.
D. S. Erickson, E. T. Ong, and R. Donado, Abstracts 1986 Fuel
p. 372.
an Unconditional AFC-Design, 1996 Fuel Cell Seminar Program
Abstracts, 1996, pp. 694–697. 30.
S. H. Lu and J. R. Selman, J. Electrochem. Soc., 131: 2827, 1984.
698–701. 32.
D. Shores and Y. Qu, Dissolution of Oxides in Molten Carbonates,
8. J. Appleby, Proc. Workshop Electrochem. Carbon, S. Sarangapani,
Proc. 3rd Int. Symp. Carbonate Fuel Cell Technol., 93 (3): Pen-
J. R. Akridge, and B. Schumm (eds.), Pennington, NJ: The Elec-
nington, NJ: The Electrochemical Society, 1993, pp. 356–367.
367–370.
Fuel Cells, BNL 51418, prepared for Brookhaven National Labo-
ratory, 1979. 34.
D. C. Fee, S. A. Zwick, and J. P. Ackerman, Proc. Conf. High
1970.
lysts for Phosphoric Acid Electrochemical Systems, FCR-7157F,
prepared by Int. Fuel Cells under Contract No. 9-X13-D6271-1 37.
B. C. H. Steele et al., Solid State Ionics, 38/30: 465, 1988.
375–393.
14. C. S. Kim et al., Evaluation of the Electrode Performance for
PAFC by Using Acid Absorption Acceleration and AC-Impedance 40.
N. Minh, et al, Monolithic Solid Oxide Fuel Cell Develop.: Recent
Measurement, 1996 Fuel Cell Seminar Program and Abstracts,
Technical Progress, Fuel Cell Seminar Program and Abstracts,
1996, pp. 254–255.
1992 Fuel Cell Seminar, 1992.
16. A. Pigeau et al., Proc. Symp. Porous Electrodes: Theory and Prac- 42.
N. M. Sammes and Y. Zhang, Effect of Microstructure on the High
tices, Pennington, NJ: The Electrochemical Society, 1984, p. 234.
Temperature Mechanical Properties of (CeO2)0.8(GdO1.5)0.2 Electro-
lytes, 1996 Fuel Cell Seminar Program and Abstracts, 1996, pp.
17. H. C. Maru, L. Paetsch, and A. Pigeaud, Proc. Symp. Porous Elec-
115–118.
trodes: Theory and Practices, Pennington, NJ: The Electrochemi-
cal Society, 1984, p. 20.
D. TRAUB HOOIE
18. R. J. Petri and T. G. Benjamin, Proc. 21st Intersociety Energy
US DOE, Federal Energy
Conversion Eng. Conf., 2, Amer. Chemical Society, Washington
Technology Center
D.C., 1986, p. 1156.
394 ELECTROENCEPHALOGRAPHY
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3015.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Profile
Electrostatic Precipitators
● Email this page Enter words or
Standard Article
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3015.htm17.06.2008 16:37:10
fects
not related to velocity. Later in this article, the equa-
Corona. An electric corona is the key feature of operation tions
needed to calculate the charge on particles and the
of an electrostatic precipitator because it supplies the ions
electric fields in precipitators are presented. These equations
that charge the particles for subsequent collection. The corona make it
possible to calculate with useful accuracy the electric
discharge (see ELECTROSTATIC DISCHARGE) is created by the ap- forces
on particles as they move through a precipitator and,
plication of high voltage to a wire or electrode with sharp
ultimately, the overall collection performance. Still, the mi-
1
J. Webster (ed.), Wiley Encyclopedia of Electrical and Electronics Engineering.
Copyright # 1999 John Wiley & Sons, Inc.
2 ELECTROSTATIC PRECIPITATORS
tion, after Walter Deutsch who first derived it (2, p 164). The
High
ωp = (3)
A
gration velocity describes some of the performance aspects of
precipitators quite well.
k
precipitator, and c is a collection rate parameter. This equa-
tion is more commonly written as:
where k is a parameter, typically 0.4 to 0.6. The parameter
#
is much larger than the parameter # for the same precipi-
# # ##
k
ωA
eff = 100 × 1 − exp − (2)
tator. Although the form of the Matts–Ohnfeldt equation may
V
seem strange, it matches the performance of operating precip-
Voltage and current values are given for the inlet and outlet sections. Voltage
usually decreases from inlet to outlet; current usually increases from inlet to
outlet.
ELECTROSTATIC PRECIPITATORS 3
itators in the sense that most large precipitators are not as are
washed with internal sprays, rather than being rapped,
responsive to collector area as the Deutsch equation would for
particle removal.
imply.
Discharge electrodes come in a wide variety. They can be
All these equations are only approximations (5), and mi-
classified as (1) weighted wire, with a weight attached to the
gration parameters are basically fitting parameters. They ig- bottom
of the wire to keep it straight, (2) rigid frame, with the
nore many other processes that occur in precipitators (3, p. wires
or other discharge elements suspended between frame
96), but do show that plate area and gas volume are critical
members, (3) rigid mast, with a central member supporting
in determining the collection efficiency.
discharge elements between cross-members, and (4) rigid dis-
charge
electrodes (single, rigid structures with discharge ele-
Mechanical Configuration ments
attached directly to the central member). Regardless of
4 ELECTROSTATIC PRECIPITATORS
Transformer-Rectifier of
99.5%, must be achieved. Yet, precipitators have long dem-
PRECIPITATOR OPERATION
Hopper The
electrostatic precipitator is best described in terms of its
major
subsystems and their interactions. First of all, it is an
Corona Electrode
electrical machine in which the generation of corona and pro-
Tensioning Weights vision
of strong electric fields are essential for collecting parti-
cles.
Second, it is a large mechanical device whose geometry
Figure 3. Wire-plate precipitator. The gas enters at the left face, has
evolved over the years for providing the best collection of
passes through multiple lanes, and exits at the right face (not visible).
particles in the smallest volume. Third, for particle collection,
The plates shown partially across the left face are perforated with
particles must be charged and moved to the collection sur-
many holes to provide uniform gas flow. Electrical power to three
faces
while large volumes of gas pass through the precipita-
independent sections is supplied at the top. Plate rappers periodically
dislodge the collected particles and allow them to fall into the hoppers tor.
These basic subsystems work together to provide the de-
below. (Illustration courtesy Research-Cottrell, Inc., Somerville, NJ sired
collection, but their operation can be upset by unusual
1997 by Research-Cottrell. All rights reserved.)
conditions, poorly maintained mechanical equipment, or by
changes
in the properties of the particles being collected.
The
understanding of precipitator operation has been
2. iron and steel industries for particulate control aided
immensely with the use of computer models. The num-
3. cement industry for particulate control in roasting ov- ber of
variables and their interactions that must be consid-
ens (see CEMENT AND BRICK INDUSTRY) ered
create a complex situation. Operations that are easily
4. sulfuric acid manufacturing for the collection of sulfuric
described with a few equations have to be calculated repeat-
acid mists edly
in order to accommodate changing conditions inside the
ELECTROSTATIC PRECIPITATORS 5
try, properties of the gas, and properties of the electrode sur- rona
is a function of wire radius (but not material) and gas
face to some extent. In all cases, the corona is a Townsend
density, as given by
electron avalanche sustained by secondary feedback (8).
An electron avalanche occurs when a free electron moves
6 # # δ #
gap.
For negative corona (either glow or tuft), the electrons
The magnitude of the electric field, E(z), for any value of z from
the avalanche are captured by electronegative molecules
between r and R is given by
(typically oxygen) to form negative ions that continue to move
away
from the wire. In industrial flue gases, more acidic mol-
E0r
ecules (sulfur dioxide, sulfuric acid) will rapidly collect elec-
E (z ) = (6)
z trons
or strip them from oxygen ions to form the primary cur-
rent
carrying ions in the gas.
The applied voltage can be made sufficiently high so that the
The corona itself extends into the gap only as far as the
electric field at the surface of the wire and for some distance
electric field is able to support the avalanche. However, co-
into the gap is larger than 3 MV/m. Under such conditions, a rona
serves as an almost unlimited source for ions that con-
free electron (from a passing cosmic ray) would create an ava- tinue
across the gap. Without additional stabilization, the
lanche.
current from a wire would increase without limit. The stabili-
If the wire is positive, the electrons move toward the wire
zation is provided by the accumulation of charged ions be-
and are neutralized on the wire surface. If the wire is nega- tween
the wire and the collection electrode, called the ionic
tive, the electrons move away from the wire into regions space
charge.
where the electric field is too weak to support the avalanche.
The ionic space charge produces an electric field that op-
Each avalanche would produce only a single burst of electrons poses
that applied field. This opposing field stabilizes the co-
and ions, if there were no mechanisms to liberate more free rona
and permits the voltage to be increased well above the
electrons in the high field region.
corona onset. With each increase of voltage, the corona injects
One mechanism for sustaining the corona is photoioniza- more
ions into the electrode gap, but the increased space
tion. When an electron and a positive ion recombine, the atom
charge prevents the number of ions from rising indefinitely.
emits ultraviolet radiation to remove the energy of recombi- The
difference between a corona device and other gas dis-
nation. This radiation can ionize a similar molecule some dis-
charge devices, such as neon lamps or thyratrons, is that the
tance away. Since there are a multitude of photons in the
electrode spacings and gas pressures are very different; ionic
initial avalanche, some electrons are likely to be liberated at space
charge can stabilize the corona, but not the discharge
the proper distance to begin a new avalanche. in
smaller devices.
6 ELECTROSTATIC PRECIPITATORS
2
and collection.
m
c Ionic
The corona onset voltage, Vc , in the wire-cylinder is calcu-
/ 60 Space charge =
A current
n µ 3
lated by direct integration of the electric field to obtain
( density 6.1 C/m
y
# #
t
R
i
s 40
V = E r ln (8)
n Equivalent
c c
e
r
d current density
n Maximum
For a single wire between plates, a similar relation holds:
e 20
r voltage
r Voltage
u
1 + cos #πr
1/2 C 0 offset
2b
#
V = E r ln (9)
c c πr
20 30 40 50 60
1 − cos
2b
Voltage (kV)
2
between plates, there is a mutual coupling effect that raises The
5 kV offset is equivalent to a current density of 20 nA/cm , which
the corona onset voltage for all the wires. When the wire-wire
reduces the amount of real ionic current that can flow at a given
voltage.
separation is greater than twice the wire-plate separation (a
common configuration), the coupling effect is only a few per-
cent of the single wire corona onset voltage.
(12). Eq. (11) and Eq. (12) are inferred from the insights of
Above the corona onset voltage, the relationship between
That is, the ions are so dense at the wire that mutual repul-
# #a + 1##
sion dominates all other effects (see SPACE CHARGE AND SPACE
V = V + E r a − 1 − ln
c c
2
CHARGE MEASUREMENTS), leading to a voltage dependence that
# j 0x # x #2 (10)
follows Sigmond’s saturation law (with a correction for the
a = 1 +
corona onset voltage) and an angular dependence that ap-
µ# E r
c
proaches a straight line expansion away from the corona wire.
Electric Field. The electric field with ionic current and with-
r
out
space charge in the wire-cylinder is given by
θ b
E (z ) = ##E0r #2 + jz #1 − r2 ##1/2 (14)
z µ# z
ELECTROSTATIC PRECIPITATORS 7
value, depending on the wave form. The electric field also fol-
lows the voltage and current density changes. The phase rela-
A similar expression works for the wire-plate geometry at the
tionship between voltage and current may be modified by the
central location:
capacitance of the system in addition to the ion transit time.
b µ# of
sparking is directly attributed to the high electric fields
sage over the same volume heats the gas, increases its con-
where the value of n is 2.
ductivity, and leads to the catastrophic spark. The streamers
All control systems for TR sets are designed to turn off the
Strictly speaking, the space charge will increase from charg-
does shut down the whole electrical section served by the set
effects are real, but small, for the amount of time particles
for
a few tenths of a second and represents a period of time
spend near any one electrode.
when little collection (except for the space charge field) takes
The interest in the values of field near the plate comes
#4 2
sumed while keeping the peak voltage high.
about 0.41 # 10 m /V # s at high concentrations (9, p. 24).
Generally, the average current density for a given average The
crossover from low- to high-concentration behavior occurs
voltage is not sensitive to the wave form. The current density- at
about 400 #L/L (16). Ion mobility is also a function of gas
voltage characteristic, Eq. (11) for example, is linear enough
density (a combined pressure and temperature effect). Conse-
over the normal operating range that the differences in wave
quently, it is possible for seemingly minor changes in gas
form do not produce different curves. Wave form does dramat-
composition and conditions to affect the operation of the pre-
ically affect the peak voltage and peak current density.
cipitator by a large amount.
8 ELECTROSTATIC PRECIPITATORS
sparking.
These spaces provide the opportunity for particle-
Electrode Design. Long electrodes are difficult to keep laden gas
to bypass the collection zones, a phenomenon called
aligned between the plates. Wire electrodes are often sneakage.
Internal baffles are often used to suppress the
weighted to keep them vertical and taut; the weights may be sneakage
flow or redirect it toward the active collection zone,
guided at the bottom to suppress swaying. With frame elec- but the
clearances must still be maintained.
trodes, wires are stretched across a rigid tubular frame to To
some extent, the sneakage flows are driven by the pres-
provide alignment; the frame itself must be carefully con- sure
drop along the length of the precipitator. The presence
structed to remain aligned. Rigid electrodes and mast elec- of plate
stiffeners produces a back pressure that is partially
trodes provide a strong tubular spine along which corona elec- relieved
by gas flows out of the top and bottom of the lane. A
trodes are placed at close-spaced intervals; these electrodes simple
model of the pressure drop (17) shows that increased
are often supported on alignment frames at both top and stiffener
protrusion into the lane increases the sneakage flow
bottom. and that
taller plates show proportionally less sneakage than
ELECTROSTATIC PRECIPITATORS 9
where n
is the number of elementary charge units e, # is the
Electric field, w:
0.1 1 10 100
w = aE #kT (20)
Particle diameter ( m)
µ
e
10 ELECTROSTATIC PRECIPITATORS
) Full charging
n
ρµt
( 1,000
τ = (21)
e
#
g
h 100
C
where # is the ion density, # is the ion mobility, and t is the
actual charging time. The quantity #t is often termed the ion
10 Field charging
exposure time.
1
With these definitions in hand, the charging rate for classi-
0.1 1 10
cal field charging is
Particle diameter ( µ m)
#
dν 3w ν 2
Figure 7. Computed particle charge (number of electrons) on parti-
dτ = F (ν, w) ≡ 4 1 − 3w , ν ≤ 3w cles
in a precipitator under normal operating conditions (E # 5 #
(22) 5
#5 2
≡ 0, ν > 3w 10
V/m, j # 1.2 # 10 A/m , time # 1.9 s). With the high electric
field,
sparking occurs in the precipitator but does not affect the
charge. The straight line computed by field charging alone shows the
where 3w is commonly called the unipolar saturation charge;
dependence on the square of the particle diameter and an increasing
it is the upper limit of charge that can be attained by the field
discrepancy from real charges below 3 #m.
charging mechanism. The field charging rate is named F (#,
w) for convenience in referring to it. The value 3 comes from
the assumption of conductive particles. (For most precipita- where
the area fraction, f (w), is described by the equation:
tors, this assumption is reasonable.)
The field charging model has a strong initial charging
1
rate, 3w/4 , that becomes zero at # # 3w. The larger the parti-
f (w) = (w + 0.475)0.575 , w ≥ 0.525 (26)
cle (radius a), or the stronger the field, (E), the greater the
= 1, w < 0.525
charging rate is. When # # 3w, the charge on the particle is
When
the particle charge reaches 3w, the field charging rate
2
ne = 12π#a E (23)
becomes zero, but the diffusive contribution continues to raise
the
particle potential above 3w, at a decreasing rate.
Large particles, therefore, acquire very large charges in the
Figure 7 shows some particle charges computed in model-
precipitator and can be easily collected. ing
a precipitator. Even though the electric field changes in
The classical diffusion charging rate in nondimensional no- time
(because of sparking) and in space (as the particles pass
tation is each
corona wire), the charge on each particle increases con-
ELECTROSTATIC PRECIPITATORS 11
100
electric fields near the wires and directly under them move
) the
particles toward the collecting plates. When the turbulent
s
/ eddy
velocities are lower than the particle drift velocity, the
m
c
particle can travel to the plate and be captured.
(
y
t
The collection of particles can be modeled as a series of
i
c 10
o
collection zones under the corona wires, followed by mixing
l
e
v
zones between the corona wires. The classical Deutsch rela-
t
f
i tion
for a given particle size can be written as
r
D
1
p (a) = exp(−v(a)/ωp ) (28)
0.1 1 10
Particle diameter ( µ m)
where p is the penetration for particle size a, v is the
drift
5
velocity, and #p is the specific precipitator velocity. Penetra-
Figure 8. Computed particle drift velocities in a field of 5 # 10 V/
m. The minimum in velocity typically occurs in the range 0.3 #m to tion
is 1 # eff / 100, the fraction remaining after collection;
0.5 #m. Although particle velocities greater than 100 cm/s are calcu-
emissions from the precipitator are proportional to the pene-
lated for larger particles, other factors related to the particle’s Reyn-
tration. However, a better collection model has been found
olds number need to be taken into account to compute the velocity to
be
accurately. Even so, such particles are collected very efficiently.
# #
v(a)
p (a) = 1 − (29)
N ω
away from the plate, even though the ions and particles con-
tinue towards the plate. This motion creates a large-scale
Deficiencies in Precipitator Operation
eddy that carries some of the particles back toward the center
There are aspects of precipitator operation that interfere with
of the lane. the
collection and disposal of particles. Some of these occur
with
every precipitator, while others are specific to certain
Local Particle Collection. Turbulence tends to homogenize
types of particles. Improper maintenance of the precipitator
the particle concentration across each lane, while the strong will
also produce problems of operation.
12 ELECTROSTATIC PRECIPITATORS
Back Corona. If the electric field in the dust layer from the
H (0.18vg )2 flow
of corona current becomes large enough, the field can
RR = (30)
L g #x
support corona generation in the interstices of the dust layer.
The
corona in the layer injects ions of opposite normal polar-
where H is the plate height, L is the plate length in the direc- ity
back into the interelectrode gap, hence the name back co-
tion of flow, vg is the average gas velocity, g is the acceleration
rona. The precipitator is intended to be unipolar (charge of
of gravity, and #x is the distance in the direction of flow be- one
sign only). When ions of both polarities are present, they
tween wires. The factor 0.18, obtained from fitting the model
charge particles in opposition, so that the net charge is much
to several measured cases, predicts gas velocity near the plate
reduced below the full unipolar charge.
as a function of the average velocity. The model gives rapping
It has been found that the critical field for back corona
5
reentrainment factors approximating real precipitator opera-
formation in coal flyash is about 5 # 10 V/m. This is much
6
tion. It shows that rapping losses increase dramatically with
smaller than the 3 # 10 V/m required for corona at the
gas velocity and points out the importance of good gas flow
high voltage electrode. Two factors assist the formation of
conditions.
back corona at lower fields in the dust. First, current flow
ELECTROSTATIC PRECIPITATORS 13
agent
designed to modify the resistivity of the particles for
Electrostatic precipitation is a mature technology, much like back
corona control. Ammonia is an agent that has been used
the automobile. Although we do not expect to find major tech- to
suppress back corona and improve the cohesivity of the
nological advances in precipitators, the variety of applications dust
cake; its back corona action is not certain but may result
and pressure of external regulations will bring about slow, from
the formation of a fine particle mode that increases the
continuous improvements in precipitator operations. Im- space
charge. Sodium conditioning to improve the resistivity
provements are often measured in terms of capital and op- of dust
cakes is another type of conditioning.
erating cost reductions or increased reliability because precip-
Although conditioning agents target specific precipitator
itator technology seems capable of controlling particle
problems, they do not always perform as intended. Some fly-
emissions to any desired level, given a large enough machine. ash
compositions are not responsive to sulfuric acid condition-
Some of the areas that could prove fruitful in these directions ing,
for example, or the precipitator operating temperature
are discussed below. may be
inappropriate. Understanding the methods by which
each
agent improves precipitator operation would guide the
Rapping Loss Reduction best
use of the agent and could lead to development of new
14 ELECTROSTATIC PRECIPITATORS
BIBLIOGRAPHY
113, 1992.
distribute the charging and collecting of particles over more
of the precipitator sections, reducing the impact of the high
4. S. Matts and P. Ohnfeldt, Efficient gas cleaning with SF electro-
¨
concentrations.
522, 1984.
decreasing, and the Cunningham slip factor is still close to
1. This size range, however, contains particles that are most
6. H. E. Rose and A. J. Wood, An Introduction to Electrostatic Precip-
pany, 1966.
lation. Improvement of particle collection in this size range
will require improvements to particle charging and mainte-
7. J. Katz, The Art of Electrostatic Precipitation, Munhall, PA: Pre-
ELECTROSTATIC PROCESSES 15
Reading List
The next best general reference is the book by Oglesby and Nichols
(9).
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3016.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Electrothermal Launchers
Abstract
Electrothermal Plasmas
Electrothermal Devices
Power Supplies
Plasma-Material Interaction
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3016.htm17.06.2008 16:37:43
40 ELECTROTHERMAL LAUNCHERS
ELECTROTHERMAL LAUNCHERS
ELECTROTHERMAL PLASMAS
barrel.
high enough to not only dissociate molecules into individual
atoms but to excite and ionize the gas atoms themselves to
form a high pressure plasma. The plasma turns out to be in
ELECTROTHERMAL DEVICES
a state of thermodynamic equilibrium (among plasma species)
and ohmically heated by an electrical circuit and hence the
Figure 2 shows a simplified schematic of an ET launcher,
name electrothermal plasma. Unfortunately, ablating plasmas
where the plasma source is attached to an expansion barrel
contain significant quantities of higher atomic weight species
that contains the mass to be accelerated. Examples of the
such as carbon and oxygen.
kind of projectiles accelerated range from very small to very
Anode
Ablating
wall material
Electrothermal
Cathode
plasma jet
Arc
Ablated
mass
ELECTROTHERMAL LAUNCHERS 41
Electrothermal source
Acceleration barrel
Payload
Figure 2.
Simplified schematic of an
Switch electrothermal
launcher showing an elec-
trothermal
plasma source attached to an
Electric power
expansion
barrel that contains the mass
supply
to be
accelerated.
# Discharge current
# # # #
# # and voltage measuring
# sensors
# ## ## #
# # # #
## ## ## #
## ## ## #
# #
# #
## ### ## #
Electrothermal
# #
# # # #
plasma source
# # #
# # # #
# # ##
# # #
Trigger #
# ## # # pulse
# #
# # # #
## ## # #
### #
### ## # #
# #
### #
### ## #
# ## # #
#
Spark-gap #
switch
#
High voltage
# ## # #
#
# ## # #
# #
power supply
# # #
## # # #
Isolation L# L
L R
transformer
## # # #
_
## ## ### #
#
# # #
#
#
### ## ### #
#
# ## # #
#
#
ET source # C C C
C
# ## # #
# #
#
#
Figure 3. Schematic of the electrothermal-chemical launcher concept Figure 4. A
simple pulse power system with pulse forming network
showing basic process of plasma injection, propellant’s burn and com- (PFN).
Capacitor modules are coupled to each other via inductors to
bustion, and motion of the projectile down the barrel. Basic compo-
form the desired pulse length of the discharge current. Coupling in-
nents are an electrothermal plasma injector, a combustion chamber
ductors are connected between the capacitors, and a charging resistor
that contains a propellant, and a barrel that contains a projectile. is
connected to the charging high voltage power supply.
42 ELECTROTHERMAL LAUNCHERS
C = 315 Fµ
will be shown in a following section. The temperature, in real-
(single capacitor, no PFN)
ity, will increase and decrease following closely the time his-
launch device
pulse length is longer and has a wider peak.
It is best to obtain current pulses with a flat top over a
longer period of time to provide a similar plasma pressure
time history into the source. The reason for flat-top pressures
Temperature
Pressure
is to achieve near-ideal electrothermal interior ballistics pro-
files. Ideal electrothermal interior ballistics profiles would
have a flat-top plasma pressure, an increasing velocity during
Velocity
the discharge cycle across the source, a slowly increasing tem-
perature (but kept as low as possible), and an increased elec-
tric power delivery to the source. The idealistic profiles are
shown in Fig. 6, where the pressure is maintained constant
over the discharge cycle, and the electric power is increasing
Discharge time
until the end of the flat-top pressure then decreases rapidly
ELECTROTHERMAL LAUNCHERS 43
into the cell from ablation of the wall and the rate at which
plasma temperatures and densities, where plasma radiates
∂t ∂z
the physics of electrothermal plasmas is illustrative. These
equations may be written in a global fashion to calculate the
where n is the number density of plasma particles (atoms/
time and spatial-averaged plasma parameters. A global, time-
m3), v is the plasma velocity (m/s), and is the time rate of
dependent set of equations would help in evaluating the time
change of the number density of ablated material from the
variation of the plasma parameters, as shown by many re-
cell wall (atoms/m3 s) and is given in (15) by:
searchers (see 8–10). However, a one-dimensional, time de-
pendent description yields a more accurate description that
2q##
shows the time and spatial variation of the plasma parame-
n˙a = H A R (2)
sub p
ters inside the plasma generator, and the plasma flow and
acceleration mechanism of the payload inside the launcher’s
s
then its travel along the axial direction into the barrel. A de-
scription of the set of equations for an electrothermal plasma
where f is the energy transmission factor through the vapor
injector that is operating on the principles of ablation-con-
shield, #s is the Stefan–Boltzmann constant (5.670 # 10#8
trolled arcs is given below for a simple capillary discharge
2 4
44 ELECTROTHERMAL LAUNCHERS
s
gradient. The third term is the velocity loss due to the in-
/
(
crease in the number density from ablated material (abla-
y
t
i
tion drag).
c 400
l
When setting the momentum equation for the barrel, one
e
t
has to include an additional term on the right hand side of
i
x
Eq. (4) to account for losses due to momentum transfer to
E 200
the payload (projectile). This additional term has to be added
Code results
Experiment
negatively to the right-hand side of the equation and is given
in (12) by:
0
0 2 4 6 8
ρproj ∂vproj
Input energy (kJ)
(5)
ρ ∂t Figure
7. A comparison between experimental and computer code
0.01
ms to 5 ms. As a result, critical components are damaged
where the first term on the right-hand side is the internal due to
surface erosion and thermal deformations. Surface ero-
energy due to thermal motion, and the second term is the sion
is one of the parameters that has an effect on the per-
internal energy due to ionization. The modified Saha–
formance, durability, efficiency, and lifetime of the launch
Boltzmann equation gives the relation between the effective
device. Minimum deformation and damage of the critical com-
ELECTROTHERMAL LAUNCHERS 45
# is about
60% less than that of aluminum. Molybdenum is
even
better and has about 75% less erosion than aluminum,
# # while
tungsten has no obvious erosion below 20 GW/m2, but
# surface
coloration and micro cracks may occur. Because re-
# # fractory
materials have less surface erosion, better thermal
Surface heat flux (q)
## resistance,
and better structural strength, launcher compo-
## nents may
be coated with layers of selected refractory materi-
als.
Graphite and carbon materials have no melting tempera-
ture when
direct sublimation takes place.
Time The
energy transmission factor through the vapor shield is
the ratio
between the actual heat flux at the ablating surface
Net erosion to the
incident heat flux from the plasma source, which is
about 10%
for most graphite grades at incident heat fluxes of
60 GW/m2
and greater. Insulators are important materials for
# # Many
polymer materials have
been considered as insulating
Ablative plasma shield components
in electrothermal launch devices, among these is
Lexan,
which is a polycarbonate polymer (C H O ). The va-
#
16 14 3
%
als to reduce surface erosion of rails and barrels; also high (
r
tensile insulating materials and specially prepared compos- o
Silicon carbide
t 50
c
ites may also be considered to eliminate the ablation of the a
f
sources (21–25). s
s
i
Under such short and intense high heat flux conditions m
s
the material surface suffers melting and vaporization, and n 30
a
r
Lexan
t
a plasma boundary layer will be formed at the ablating y
g
surface. Such vapor plasma absorbs a fraction of the incom- r
e 20
ing energy; thus less heat flux reaches the surface resulting n
Boron nitride
E
in less surface erosion. This is described as the vapor shield
mechanism, and is schematically illustrated in Fig. 8. The 10
46 ELECTROTHERMAL LAUNCHERS
Material
Erosion Thickness (#m)
Molybdenum on copper
18.8
Tantalum carbide on copper
4.3
Tantalum nitride on copper
3.8
Tantalum on copper
0.8
Tungsten #0 #0
#0 #0 #0
Tungsten-rhenium alloy (3% Re) #0 #0
#0 #0 #0 #0 #0
Lexan (polycarbonate, C H O ) 11.1 23.3
30.8 41 43.5
16 14 3
BIBLIOGRAPHY
11. J. D. Hurley, M. A. Bourham, and J. G. Gilligan, Numerical simu-
834–839, 1996.
Trans. Plasma Sci., 14: 423–434, 1986.
190–197, 1990.
5. D. Hewkin and E. Figura, Fundamental research and numerical
modeling of the internal ballistics of electrothermal chemical
16. Ya. B. Zeldovich and Yu P. Raizer, Physics of Shock Waves and
guns, IEEE Trans. Magn., 29: 561–566, 1993.
High Temperature Hydrodynamic Phenomena, Vol. 1, New York:
ELLIPTIC EQUATIONS,
PARALLEL OVER SUCCESSIVE RELAXATION ALGORITHM 47
Reading List
MOHAMED A. BOURHAM
JOHN G. GILLIGAN
North Carolina State University
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W1636.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Engineering Workstations
● Email this page Enter words or
Standard Article
Abstract
Architecture
Software Environments
Applications
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W1636.htm17.06.2008 16:38:04
ENGINEERING
WORKSTATIONS
An
engineering workstation is a complete computer system
used for
engineering applications. A typical engineering com-
putation
requires a moderate amount of computing power (es-
pecially
in the form of floating point operations), relatively
high-
quality graphics capabilites, and a large main memory.
ARCHITECTURE
some powerful workstations are used as departmental
servers as well.
Similar
to most other types of computers, a workstation
• Multitasking. Complex applications require the system
mainly
consists of one or more processors, main memory,
to support multitasking. The user may use the worksta-
mass
storage devices (disks), and other I/O devices such as
tion to do several tasks concurrently, and make full use
mice and
keyboards. Workstations are also equipped with
of the power of the workstation.
network
adaptors and such multimedia devices as CD drives,
• Networking. Workstations are networked for easy access
microphones, and speakers. Initially, most workstations used
to remote computing resources. Workstations can be con- 32-bit
traditional CISC (complex instruction-set computer) (3)
nected to local area networks or even wide area networks processors
such as the Motorola 680x0s. However, at present,
such as the Internet. The commonly used communication 64-bit RISC
(reduced instruction-set computer) (3) based pro-
protocol is transmission control protocol/Internet proto- cessors
running at more than 100 MHz are widely used in
col (TCP/IP) (1). modern
workstations. For example, the RISC processors Ul-
traSPARC
1, MIPS R10000, and PowerPC are used in Sun
In terms of computing power and physical size, worksta- UltraSPARC
stations, SGI O2 machines, and IBM RS/6000
tions can be classified into three types: notebook, desk-top,
workstations, respectively.
and server. Notebook workstations are rare because of their Two
levels of cache (Level 1 and 2) usually help bridge the
low performance/cost rates in comparison with notebook per- gap between
processors and memory. Level 1 cache, which is
sonal computers (PC). From a hardware configuration stand- on the
processor chip, and is often called on-chip cache, is
point, a workstation may be constructed using a single pro- about 32
k bytes. Level 2 cache, sometimes called external
cessor, or using multiple processors ranging from two to up to cache, is
256 kbytes to 1 Mbytes in size. The size of main
a few dozen. A bus-based symmetric architecture is adopted memory
varies from 32 to 128 Mbytes.
for a small number of processors, but when the number of All
components of a workstation are connected via several
processors increases, more complex structures such as hierar- buses as
shown in Fig. 1. The processor and memory are con-
chical buses may be used to enhance the data access band- nected by a
system bus, and all I/O devices share a common
width. I/O bus.
Simplicity and low cost are the major merits of such
The most common operating system for workstations is a
structure, which has been adopted by most workstation
UNIX. It is a time-shared multitasking system supporting
manufacturers.
standard languages such as C. A graphical user interface However,
with the increase of processor computing power
(GUI) is provided to facilitate the use of these workstations. and high
bandwidth requirement of multimedia devices,
Usually such an interface is based on the X Window system buses are
becoming the bottleneck, and their performance im-
(2), and is a WYSIWYG (what you see is what you get) inter- provement
always lags behind the need. Some new architec-
face. Two popular GUIs on workstations are Open Look and tures have
been proposed. For example, Sun developed the
Motif. ultra port
architecture (UPA) (Fig. 2), a crossbar-switched in-
Workstations are used by scientists and engineers to do terconnect
that enables multiple simultaneous data transfers.
computationally intensive or graphics intensive applications In its O2
products, SGI uses the unified memory architecture
such as research, software development, scientific computa- (Fig. 3).
The high-speed multiported main memory is at the
tion, and CAD/CAM. They may also be used for desk-top pub- heart of
the system. All data types required by various sub-
lishing, scientific and medical visualization, image analysis, systems
such as the monitor and video I/O are stored in mem-
and other applications. ory. Each
subsystem accesses the memory through high-speed
The idea of a workstation originated at Xerox during the ports
designed to ensure that no data starvation occurs on
1960s, but it was not until the development of microprocessor any port.
This architecture allows more efficient use of the
technology that workstations became widely available. Mod- entire
system memory and increases performance by reducing
ern workstations began to appear in the market in the 1980s. the buffer
copies among those components.
Apollo Computer, Inc. released its first Apollo/DOMAIN Most
multiprocessor workstations adopt the symmetric
workstations (DN100s) in 1981 with the operating system
multiprocessor architecture (SMP) because of its simplicity
Aegis SR1 installed on them. Andreas Bechtolsheim designed and low
cost. The architecture is very similar to that in Fig.
Processor
Memory
L1 cache
CPU Monitor
Memory
L2 cache
controller
System bus
Graphics
processor
compression display
I/O bus
Main
Video
Memory
PCI I/O
dress
space of the process. Each process has its own address
space
and a single thread of control. However, in some cases
it
is desirable to have multiple threads of control that share
Processor I/O a
single address space. The major difference between threads
cache subsystem
and
processes is related to the address space. Each process
has
its own address space, and the interaction between pro-
cesses.
calls
that may be blocked waiting for some events). In a sin-
gle-
processor machine, they may improve the system utiliza-
Figure 2. Sun Microsystems’ Ultra Port architecture. Rather than tion
and throughput. In a multiprocessor system, threads
using a bus to connect all components, a crossbar is used which allows may
run in parallel to support the exploitation of fine- or
multiple simultaneous data transfer among components.
medium-grain parallelism.
Communication of
Technology (MIT) in 1984. X Windows is an industry-stan-
dard
software system that allows programmers to develop
The communication protocol often used to exchange data be-
portable
graphical user interfaces. It is a collection of pro-
tween workstations is TCP/IP, which is also the protocol used
grams
and libraries that allow the development of applica-
by the Internet. A new IP protocol called IPv6 is being de-
tions
for graphics workstations in which the output can ap-
signed and standardized to solve problems encountered such
pear in
multiple windows on the screen. It also provides a
as the significant increase in the number of computers on net-
means for
a program to be run on a remote computer, but to
works. Layered protocols such as TCP/IP have high software
display
the results on the local workstation. It is character-
overhead when used to exchange short messages. There are
ized by:
efforts to address this. Usually, workstations are connected
with each other by Ethernet networks. Currently, there are
two bandwidth Ethernets, 10 Mbps (megabit per second) and •
Device Independence. X Windows allows programs to dis-
100 Mbps. Asynchronous transfer mode (ATM) (4) networks play
windows containing text and graphics on any hard-
are more expensive, but provide higher bandwidth ( 100
ware that supports the X Protocol without modifying, re-
Mbps).
compiling, or relinking the application.
•
Client-Server Model. In X Windows, application pro-
Programming Languages
grams are called clients. The clients talk (in X Protocol)
to a
program called the (display) server, which drives the
There are many programming languages available in work-
display and interprets keyboard commands and mouse
stations. The most commonly used ones are C, C##, and
clicks. The client and server may be on different ma-
FORTRAN 77. Designed in 1972 at AT&T, C is a high-level
chines, so sometimes X Windows is called a graphical
programming language used at first to write the UNIX op-
windowing system for computer networks.
erating system, and for system programming. Over the years,
•
Mechanism/Style. X Windows provides mechanisms to
the power and flexibility of C have made it a popular lan-
Tk
Perl Tcl
Motif OpenLook
Shell
Xlib/widgets
To other X servers
X server
Figure 4. A software environment on a
via networks
workstation. This diagram shows where Unix device drives
each software discussed is in the whole
software environment and the relation-
ship between each.
managers to arrange these widgets on the screen. The Tcl and system and
computation issues to be addressed for effectively
Tk source code is publically and freely available. There are utilizing
networks of workstations for parallel computing. In
many applications developed by academia and industry using practice,
a network of workstations is a heterogeneous and
Tcl/Tk. Figure 4 gives the relationship between all the soft-
nondedicated system. Heterogeneity represents computing
ware components discussed above. power
and architectural differences among workstations.
ENTREPRENEURING 119
BIBLIOGRAPHY
XING DU
University of Virginia
XIAODONG ZHANG
YONG YAN
College of William and Mary
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3017.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Fuel Cell Power Plants
Abstract
Networks
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3017.htm (1 of 2)17.06.2008 16:38:54
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3017.htm
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3017.htm (2 of 2)17.06.2008 16:38:54
fully
and efficiently leverage funding for the U.S. Stationary
Power
Fuel Cell Program.
FUEL
CELL TECHNOLOGY
Fuel
cells generate electricity and heat using an electrochemi-
cal
process similar to that of a battery. A fuel cell will continu-
ously
produce power as long as a fuel, such as natural gas,
and an
oxidant, air, are supplied to the system. Present early
market
systems are achieving over 40% lower heating value
(LHV)
cycle efficiency. The next-generation systems are ex-
pected
to achieve 55% and eventually 70% LHV cycle effi-
ciencies.
As
shown in Table 1, several different types of fuel cells
are
being developed for stationary power applications. The
platinum steel
(GRI) and the Electric Power Research Institute (EPRI) to
Oxidant (O or CO )
The DOE/FETC-sponsored PAFC development work at
2 2
IFC
was completed in 1992. ONSI
Corporation, located in
Cathode
South Windsor, Connecticut, has been actively involved in the
development and marketing of on-site PAFC systems and has
Electrolyte a 40
MW/year manufacturing facility. In their PAFC commer-
Anode
cialization, the ONSI Corporation, a subsidiary of IFC, is of-
units are in
operation in the United States and around the
• Electromagnetic process
world. An additional 22 units were added in the United States
• Direct conversion to electricity
through the Climate Change Fuel Cell Program. Operating
H + 1/2 O H O + Electricity
2 2 2
experience has been excellent, with availabilties of over 90%.
operates at 650#C. The MCFC, like other fuel cells and unlike
Natural on
coal or natural, refinery, or processed gas. MCFC stack
gas or
designs incorporate either internal or external fuel and oxi-
derived H2 dant
manifolding and either internal or external reforming.
coal gas Fuel Power Dc Power All
MCFC designs include flat cell components in the cell
processor section conditioner Ac
package (i.e., anode, matrix to hold carbonate, cathode, cur-
Steam power rent
collector, and separator plate).
2 2
within a porous ceramic matrix, commonly made of lithium
A/ft (160–270 mA/m ), at 0.60–0.80 V, with 50–85% fuel uti-
aluminate. An individual cell is approximately 6 mm thick.
lization (2–9).
The electrolyte is about 1 mm thick.
The goal of the U.S. MCFC program is to develop and com-
Figure 3 illustrates the structure of an MCFC stack. Con-
mercialize low-cost, packaged, simple, and modular fuel cell
ductive, bipolar separator plates connect the individual cells
systems. DOE is accelerating the drive for private sector com-
in a stack, both structurally and electrically. The bipolar sep-
mercialization of multifuel, MCFC power plants.
arator plate is made of stainless steel, and each plate physi-
The two MCFC developers have collected impressive stack
cally separates the fuel gas stream of one cell from the oxi-
test performance data under the 1990 program R&D an-
dant gas stream of the adjacent cell. One side of each
nouncement (PRDA). ERC is developing an externally mani-
separator plate channels a fuel stream so that it flows over a
folded, externally reforming MCFC and has constructed a 2
porous anode, while the flip side channels an oxidant stream to
17 MW/year MCFC manufacturing plant. ERC has con-
over a porous cathode. Each bipolar separator plate also col-
structed a 100 kW test facility in Danbury, Connecticut, and
lects current, thus, electrically connecting adjacent cells of a
has scaled up to a 6 ft2 2
Fuel out
Air out
Fuel
out
Fuel in
Oxidant
out
Oxidant
inlet
manifold
Fuel inlet
manifold
Fuel
Figure 3. MCFC stack structural de-
Oxidant
in
signs.
in
Fuel
Fuel
feed
exhaust
H2 CO2 H O
2
2 –
=
e
H2 + CO 3 CO + H O + 2
2 2 e
Anode
=
Molten
CO
Load
carbonate 3
electrolyte
Cathode 1 =
CO + O + 2 CO
2 2 2 e 3
2 –
Oxidant CO2 O2
Oxidant
feed
exhaust
Figure 4. Operation of an MCFC.
Flow
nents could make the SOFC an attractive power-plant alter-
Fuel electrode
native. However, packaging and cost reduction will be re-
quired to make the SOFC promise a reality.
Figure 5. Westinghouse SOFC design.
NETWORKS
shown in Fig. 6, has, according to IGT, the advantages of
more effective gas flow patterns, more compact design and
In
conventional fuel cell systems, multiple stacks have been
cell stacking, more efficient current and voltage transfer from
streams are ducted such that they are fed and recycled among
costs of the cell components (18). IGT replaces the ceramic
series can be nearly 10% greater than that of the stacks ar-
been attained (23). Allied Signal Aerospace Corporation is de-
ranged in parallel.
veloping the monolithic and flat planar designs and is now
plants is the
conceptual development of very high efficiency
Figure 6. IGT SOFC. fuel
cell gas turbine power plants (31–40). Studies have indi-
IRMCFC Exhaust
Stack
Fuel
and IRMCFC Exhaust
oxidant
Stack
flow
IRMCFC Exhaust
Stack
(a)
Fuel
and IRMCFC IRMCFC IRMCFC
Exhaust
oxidant
Stack Stack Stack
flow
Figure 7. (a) Parallel flow of reactant
Product Efficiency
Mature Price
Vendor Size (MW) (LHV) Fuel Cell
Gas Turbine Availability Target ($/kW)
Westinghouse 3 61 Pressurized
Heron 1999 Prototype 1200–800
5 69 Tubular SOFC
Heron
10 60 One 1.8 MW unit
Allison
Two 1.8 MW units
Multiple 1.8 MW units
Solar 1–2 58–63 Pressurized 4 :
1 PR 1999 Prototype #650
Planar SOFC
Turbo-expander
ERC 3.3 65 Direct MCFC
Steam bottoming gas 1250
3.8 68
turbine topping and
steam bottoming
Ztek 0.2 Pressurized 50–
100 kW 1998 Prototype 1000–1500
50 Planar SOFC
Allison 10–25 59–62 Unspecific
Allison 501 KB/KM or ATS engine in 425–450 for
new sources. In the short term, these restrictions may encour- about
exposure to electromagnetic fields from high-voltage
age the use of underutilized fuels, particularly natural gas, lines.
by electric power producers.
Smaller-scale distributed configuration power plants are
perfect
for commercial buildings, prisons, factories, hospitals,
25. N. Minh et al., Allied signal SOFC technology, Proc. Fuel Cell
METC-94/1006, US Department of Energy, 1994.
5. A. Leo, Status of the Santa Clara MCFC PDT, Proc. ’95 Contrac-
28. J. Wimer, M. C. Williams, and D. Archer, EMS device net-
tor’s Rev. Meet., DOE/METC 95/1020, 1995, pp. 165–171.
working, in Gas Purif. Sep., 22: 121–124, 1993.
6. H. Maru et al., ERC product improvement activities, Proc. ’95
29. M. C. Williams and J. Wimer, Mathematical modeling of MCFC
Contractor’s Rev. Meet., DOE/METC 95/1020, 1995, pp. 178–188.
cells/stacks and networks, Prepr. Pap., 206th Amer. Chem. Soc.
32. M. Williams and C. Zeh, Proc. Workshop Very High Effic. FC/GT
9. J. Scroppo et al., M-C Power’s PDI, Proc. ’95 Contractor’s Rev.
MARK C. WILLIAMS
US Department of Energy
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W5208.htm
} { { } }
HOME
●
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Fusion Reactors
Abstract
Nuclear Fusion
Inertial Confinement
Progress in Fusion Energy
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W5208.htm17.06.2008 16:39:11
process. Beginning about 1951, when one of the first fusion Thus, in
order to fuse most of the elements in the peri-
programs was begun at Princeton University by Lyman odic table,
one would need to cause them to approach each
Spitzer, Jr., many of the world’s developed nations, as well other at
energies of several to many MeV. This can be ac-
as some developing ones, have pursued research to even- complished
with particle accelerators for nuclear physics
tually produce a fundamentally different type of nuclear research
purposes, but is not practical for producing net
reactor. Since 1958, when research on peaceful uses of nu- power with
significant quantities of thermal reactants. As
clear energy was declassified, these many countries have a
consequence, only the lightest elements in the periodic
shared their research and openly collaborated, even during table have
sufficiently large probabilities (also called reac-
decades when some of the principal contributing nations tion cross
sections) of undergoing nuclear fusion for them
were political adversaries. This new type of reactor is a fu- to be
considered as fuels for fusion reactors.
sion reactor, and the very great progress which has been Fusion
reactions drive the core of the sun at a temper-
made in the decades since 1951 stands as a landmark to ature
thought to be roughly 10 million to 11 million ◦ C.
what may be the most universal and long-lived collabora- Fusion also
powers all the other stars, and thus supplies
tive effort in human history. most of the
universe’s observable energy. Our sun, in com-
Fusion reactors, like fission reactors, will use exother- mon with
other main sequence stars, obtains its energy
mic nuclear reactions to release energy. The fusion reac- through a
number of nuclear fusion reactions, beginning
tions themselves, however, are very different in character with the
fusion of 2 protons into deuterium. The deuteron
than are those in fission reactors, and they require entirely is
accompanied by a positron and a neutrino, and together
different conditions in order to proceed at an acceptably they carry
1.44 MeV of energy. The reaction probabilities
high rate for power plant applications. Due to these differ- (or cross
sections) for this and the succeeding solar nuclear
ences, fusion reactor designs will look very different from reactions
are much too small to be of any use to a commer-
those of fission reactors. Moreover, many of the problems cial fusion
reactor on the earth’s surface. The sun is able to
that have had to be surmounted in the pursuit of fusion produce
enormous amounts of energy with these reactions
power are of a fundamentally different nature than those only
because it is very large compared to the earth. The
encountered in making fission reactors practical. sun has a
diameter of about 897,000 mi and, because of its
Nuclear fusion is a reaction in which two atomic nu- size and
its gravity, has excellent energy confinement. The
clei merge to form a heavier element. If the reaction is time for
energy produced in the sun’s fusion driven core to
an exothermic one, then the fusion process will result in reach the
sun’s surface is estimated to be of the order of
the release of energy. This energy is carried as kinetic en- thousands
of years or more.
ergy by the reaction products, consisting of the product nu- Because
a practical fusion reactor for electricity produc-
cleus (which is positively charged) and another particle, tion needs
to be much smaller than the sun, it will have
such as a neutron (which is electrically neutral) or a pro- to rely
upon different nuclear fusion reactions with larger
ton (which is positively charged). Because all nuclei carry cross
sections, and it will also need to operate at temper-
positive electric charge, they repel each other, with the re- atures that
are at least 10 to 20 times those of the sun’s
sult that fusion reactions can proceed at significant rates core.
Temperatures in this range were achieved within a
only at very high temperatures which give the nuclei suf- mirror
confinement device in 1979 and within the Toka-
ficient energy to overcome their repulsion and approach mak Fusion
Test Reactor (TFTR) at Princeton University
each other close enough to merge. Alternatively, this same in the mid-
1980s, and have now also been obtained on the
end may be achievable by using somewhat lower temper- Joint
European Torus (JET) tokamak in England, and the
atures, but at high densities achieved by compressing the JT-60U
tokamak of Japan. Ion temperatures as high as
fusion fuel with high symmetrically applied pressures. 300 million
to 500 million ◦ C (roughly 30 to 50 times the
2 Reactors, Fusion
Reactors, Fusion 3
carbon. The actual weight of coal required would of course duce ill
effects. This is the result of two factors. The first is
be greater than this, since coal contains other elements be- that the
ionizing radiation released into the body by each
sides carbon. The fact that the mass of fuel which has to tritium
decay is much less than the decay energies of fission
be carried from the fuel concentration source to a fusion products
(which are typically at least hundreds of keV) or
power plant is more than a factor of a million smaller than the
energies in the decay chains of heavy elements, which
for competing fossil fuel plants is a significant advantage. can run to
over 10 MeV). The second factor is that most of
It means that moving the fuel for an entire fusion econ- any tritium
absorbed by the body would enter as water, and
omy would impose no requirements upon the transporta- water is
continuously excreted, with a biological half-life in
tion infrastructure, since the masses being moved would human
bodies of about 12 to 13 days. The absorption rate
be thousands of kilograms instead of billions of kilograms. for tritium
breathed into the lungs as gaseous molecular
The fact that fusion reactors will be able to produce a hydrogen is
very low, roughly 20,000 to 25,000 times less
gigawatt-year of electricity while using fuel which weighs than for
tritium in water molecules.
more than a million times less than that required for a Thus,
while tritium should always be treated with due
fossil fuel burning plant reflects the large difference be- care, its
possible health effects if mishandled do not appear
tween nuclear binding energies and electron binding ener- to be
significantly worse than those of many other chemi-
gies. When a reaction occurs between two parent particles, cals
routinely handled by an industrialized society. Indeed,
the extra kinetic energy carried by the daughter particles tritium is
already used in conjunction with phosphors to
comes from the change in the overall binding energy. This provide
light without the need of electricity in several ap-
kinetic energy is distributed among the daughters in ac- plications
such as school exit lights, some airfield land-
cordance with the laws of energy and momentum conser- ing
lights, and some modern illuminated watches. These
vation, with the result that the lighter daughter carries
applications incorporate significant amounts of tritium.
most of the kinetic energy. Tritium-
powered school exit lights typically use about 15 to
Burning a fossil fuel is a chemical reaction, which in- 25 curies
of tritium, and emergency runway landing lights
volves exchanges and rearrangements of the electrons out- use much
more. This compares with an amount on the or-
side the nuclei. The outer electrons of atoms typically have der of 100
to 150 curies injected into the Tokamak Fusion
binding energies of several to a few tens of electron volts. Test
Reactor at Princeton University for a high-power fu-
Thus, the changes in the net binding energy that occur in sion shot.
This device has operated on a routine daily basis
chemical reactions, such as burning fossil fuels, are typi- using
deuterium and tritium for experiments over a period
cally only a few electron volts. Nuclear binding energies, spanning
three and a half years without any significant
on the other hand, are typically many millions of electron incidents
involving tritium contamination of personnel or
volts. This has the consequence that the changes in binding the
environment. This demonstrates that tritium can be
energy involved in nuclear reactions are also much greater handled on
a large fusion system without unduly imped-
than for chemical reactions. For instance, in the D + T re- ing the
manner in which it is operated, although the total
action, 17.58 MeV of energy is released as kinetic energy quantities
of tritium being handled were orders of magni-
in inverse proportion to the masses of the daughters. Thus, tude lower
than would be required in a power plant. The
because a neutron weighs about one fourth as much as does neutron-
induced activation and damage of materials would
an 4 He nucleus, the neutron carries four times as much of be much
more challenging in a fusion power plant.
4 Reactors, Fusion
slightly
cheaper than with a natural Li blanket, because
the extra
4.8 MeV from the n + Li6 reaction is much larger
Tritium Production than the
amount of energy required to enrich the lithium.
The incident neutrons that induce these reactions arise The
United States has large reserves of easily ex-
from the D + T reactions, and perhaps also from other re- tractable
lithium, mostly in dry salt lakes and saline lakes,
actions initiated by the primary neutrons. The reaction in- which
could be produced at prices not greatly different from
volving 6 Li has the advantage that it is exothermic, adding the $20
per kg of today. This reserve is estimated to be
another 4.8 MeV of kinetic energy to the 17.58 MeV of ki- about 5
× 109 kg (1. With an electricity production effi-
netic energy released by the D + T reaction that produced ciency of
33%, this would yield about 3 × 1014 GJ of elec-
GJ. Thus, the price of breeding lithium could rise many-fold these
problems are more difficult for a reactor using the p
before it had a noticeable impact upon electricity costs. In + 11B
reaction.
Reactors, Fusion 5
Gravitational Fields
deposited new layers of material over much of the moon’s
surface, burying pre-existing surface layers. However, most
Gravitational fields produce suitably high restraining
of the solar system debris was swept out of the planetary forces
only for very large assemblages of mass, because the
orbits long ago through the direct interception of fragments
gravitational force is much weaker than the other known
by the planets and moons, and through deflections of the
fundamental forces of nature. Thus, while all the successful
orbits of fragments arising from close encounters with the fusion
reactors we can see, namely stars, use gravitational
gravitational fields of the much larger planets and moons. fields
for confinement, this is impractical for a commercial
The evidence of this early bombardment has eroded away
reactor by an enormous factor.
6 Reactors, Fusion
Dc Electric Fields
approximately given by the time required for the hot ions,
with a mean
velocity of about 108 cm/s, to traverse the ra-
If the reacting fusion fuel possessed a strong net electri-
dial
dimension of the fuel, which is much less than 1 cm.
cal charge, then it might be possible to confine it with dc
Thus, the
confinement time is in the sub-nanosecond range.
electric fields. However, in the types of fuel assemblages so
The fusion
output is proportional to the product of the den-
far used in fusion research, any net imbalance in charge
sity, the
ion temperature, and the confinement time. This
which developed in the fuel was much too weak to permit
is more
conveniently referred to as nτT, where n is density,
confinement solely through the forces that could be applied
τ is
confinement time, and T is the ion temperature. For re-
through dc electric fields. Significant electric fields do de-
alistic ion
temperatures of a few tens of keV, the very short
velop under some conditions inside the fuel using the mag-
confinement
time of the inertial approach requires com-
netic field confinement approach discussed later, and they
pression of
the fuel to very high densities, 40 to 400 times
may play a role in altering the quality of the confinement
the normal
density of the solid deuterium and tritium fuel
when they do occur.
(4), in
order to produce fusion energy releases relevant to
a reactor.
Radio Frequency Electromagnetic Fields
Reactors, Fusion 7
Z is
not unity arising from light contaminants in the deu-
vices, it is always the case that they are much greater than
greater than d ≈ λD , where the value of λD is given by:
a
Debye length.
8 Reactors, Fusion
by T. Stevenson.
field.
distribution carries twice the energy content of the parallel
Reactors, Fusion 9
10 Reactors, Fusion
will
rapidly escape the confinement device unless their ra-
tio
of perpendicular to parallel velocity is scattered back
into
the trapped region by a subsequent collision or colli-
lost.
plasma. Particles which undergo collisions such that their vp/vz
drops into the loss cone quickly escape along the field lines at the
For a plasma in which the temperatures of the electrons
magnetic mirror throat. and
the ions are of the same order, the electrons are the
more
collisional species, so they initially escape through
the
mirror throats more rapidly than the ions. This im-
These equations define the boundary between mirror
mediately gives rise to a net negative charge leaving the
trapped and untrapped particles in the space of paral-
mirror throats, and a net positive charge remaining in the
lel versus perpendicular velocities. Thus, charged particles
central cell of the mirror confinement device. This charge
with a velocity ratio vz (min B )/v which is sufficiently low
imbalance produces an electric potential which is positive
will be trapped. On the other hand, charged particles will in
the central cell, and thus an electric field which retards
promptly escape along the mirror field if they satisfy the the
loss of electrons along magnetic field lines through the
criterion:
mirror throats. This electric field causes lower energy elec-
are trapped, and ones with lower values of this ratio are
inated by these electron thermal losses. The equilibrium
lost. Since the perpendicular velocity corresponds to ro- rate
at which electrons and positive ions escape from the
tation in two dimensions, the boundary between trapped
mirror-confined plasma is governed by the rate at which
orbits and loss orbits in a three-dimensional velocity space the
less-collisional ions scatter into the loss cone, so ion
forms a cone. Thus, particles with orbits that have ratios
collisions set the rate of particle loss.
of parallel velocity to perpendicular velocity which are too
During the early decades of fusion research, devices
great for trapping are said to fall into the loss cone, depicted
incorporating magnetic mirrors as their primary confine-
in Fig. 3. ment
mechanism were extensively investigated. In their
A mirror-trapped plasma can never be isotropic in veloc-
simplest form these machines consisted of a central cell
ity space. This is because the transit time along a magnetic
permeated by a uniform solenoidal magnetic field, bounded
field line is very fast, so particles with velocity ratios that by a
stronger magnetic mirror field at each end. These mir-
fall within the loss cone will rapidly escape through the ror
machines were examples of “open” confinement devices
mirror throat (region of maximum B ). Thus, the average
because their magnetic field lines left the region of plasma
perpendicular energy in a mirror-confined plasma is larger
confinement through the throats of the mirrors. This ge-
than the parallel energy by more than the factor of 2 one
ometry had a practical appeal because it is mechanically
would expect just from the relative number of degrees of
simpler to build a linear device than one with some more
freedom involved.
complex shape. Moreover, mirrors are devices which do not
Note that neither the charge nor the mass of the par-
require a time varying field, and thus are intrinsically ca-
ticles enters directly into the criteria for trapping. In an
pable of continuous operation if the engineering systems
entirely collisionless plasma, the particles within the loss are
designed accordingly. In a practical mirror fusion re-
cone would be entirely lost within a transit time (the time
actor the magnetic field coils would be superconducting in
required for the slowest particles to travel the length of the
order to avoid ohmic dissipation.
region enclosed by the magnetic mirrors). After this very
Simple mirror confinement machines encountered dif-
rapid loss, the remainder of the plasma would remain con-
ficulties due to macroscopic instabilities which rippled
fined for as long as the magnetic field configuration was the
plasma surface and greatly augmented the loss of
maintained.
plasma particles and energy. These problems were ad-
In all systems of interest to fusion research, this static
dressed by confinement schemes which added additional
condition never arises because coulomb collisions between
current-carrying conductors to alter the magnetic field in
mirror-confined particles alter their ratio of perpendicular the
central cell so that it increased toward the edge of
to parallel velocity, creating a dynamic loss due to particles the
plasma. The intrinsically anisotropic velocity distri-
which scatter from trapped velocity ratios to untrapped ra-
bution residing within a mirror drives another smaller
tios. Once in the loss cone, these newly scattered particles
scale length microinstability, called the loss-cone instabil-
Reactors, Fusion 11
12 Reactors, Fusion
Reactors, Fusion 13
14 Reactors, Fusion
toroidal field coils and vertical field coils, there are coils ar- ing
from toroidicity. The excess energy leakage was called
rayed around the plasma to alter its cross-sectional shape,
anomolous diffusion, and was presumed to be driven pri-
and in some cases additional coils might be added to con- marily
by collective instabilities in which great numbers of
trol gross instabilities. More recently, the systems which ions
and electrons moved together to form waves and large-
deposit energy, momentum, and additional current into the scale
plasma deformations that transported energy much
plasma have also become important in improving confine- more
rapidly out of the plasma than could two-particle col-
ment.
lisions. The early years of fusion research were plagued by
Reactors, Fusion 15
16 Reactors, Fusion
There is, however, an important difference in how an portion of
the ions each pick up an electron from the gas
ordinary electrical conductor and a plasma behave when molecules.
After picking up an electron, a beam ion becomes
heated. In an ordinary conductor, such as a copper wire,
electrically neutral, and thus once again an ordinary atom.
the electrical resistance rises as the wire’s temperature in- Unlike an
ordinary atom, which at room temperature has
creases, which is to say that the scattering of the electrons an energy
of about one fortieth of a volt, these atoms in
increases. Thus, in an ordinary conductor, if the current what is now
a neutral beam have energies of 120,000 V. The
passing through it is kept constant by raising the driving remaining
ions in the beams are bent out of it with a mag-
voltage as the conductor’s temperature increases, then the net, and
then the purely neutral beam is able to pass unim-
current will become progressively more effective in heat- peded
across the tokamak’s magnetic fields to enter the
ing it until the wire eventually melts. In the sorts of plas- plasma.
Once inside the plasma, the neutral atoms of the
mas of relevance to fusion, the behavior is just the oppo- beam are
again ionized through collisions with the plasma
site. The frequency of scattering decreases as the tempera- particles
which detach the electrons from the beam atoms.
ture increases, so the electrical resistance drops. This has The
resulting 120,000 V ions, being electrically charged,
the consequence that, as the plasma gets hotter, more and are confined
by the magnetic field of the tokamak. They cir-
more current must be run through it to achieve smaller culate
along the magnetic field, colliding with the plasma
and smaller additional increases in the temperature. It particles.
Since the beam ions are much more energetic
would be extremely difficult to heat a plasma to thermonu- than the
plasma electrons and ions, they transfer energy
clear ignition temperatures in this way, and even if one to them,
and thereby heat the plasma. Eventually the beam
could handle the large current required, the poloidal field ions slow
down and become part of the bulk plasma. At this
it would produce might be greater than the optimum for point they
are said to have been thermalized because their
maintaining plasma stability and confinement. In light of energy is
similar to that of the bulk, or thermal ions.
these limitations, ohmic heating serves only as the initial In 1973,
when neutral beams were selected as the prin-
heating mechanism in tokamaks, typically raising the tem- cipal
heating technique for the Tokamak Fusion Test Reac-
perature to 10 million to 20 million ◦ C, modest by fusion tor, this
technology had reached the point of development
Reactors, Fusion 17
18 Reactors, Fusion
or
helium are examples of possible choices. This extracted
heat
will then be used to produce electricity, probably by
then be
removed from the blanket to produce electricity, ei-
ther
through steam turbines, or possibly by more efficient
tipliers
were added to the mix. The most commonly as-
sumed
multiplier is beryllium, which undergoes an (n,2n)
difficulty.
Figure 5. Example of tokamak plasma with top and bottom di-
vertors. The divertor coils produce open flux surfaces after the
Maintenance and Materials. Developing techniques to
separatrix field line, allowing the outer plasm to leave the main maintain
and repair fusion power plants is an area which
chamber to strike target plates. This reduces the influx of impu- still
requires substantial engineering development. It may
rities to the main plasma, allows the power density striking the be
somewhat simplified by the time commercial plants
target plates to be reduced through the expansion of flux tubes,
begin
operating some decades from now by advances in
and may allow further power density reductions through radia-
divertor, where it heats surfaces. Thus, all of the alpha par- tor, if
the fusion plant was built of HT-9, a ferritic steel. If
ticle energy is ultimately collected on surfaces. These sur- the
plant were instead constructed of the vanadium alloy,
faces will be cooled with some fluid or gas, of which lithium the one-
year radioactivity was depressed to a level about
Reactors, Fusion 19
105 less than for a fission breeder, and if silicon carbide was the sphere
has proven very difficult. In addition, with some
used for the fusion reactor, then the one-year radioactivity types of
drivers there may be preheating of the fuel due to
of the fusion reactor declined further to a value about 108 penetration
by driver ions, and laser-driven instabilities
3 may
degrade the symmetry of the compression. Ways of
less than the fission breeder. If D- He reactors ever proved
feasible through lunar mining, then the silicon carbide de-
ameliorating these problems are being studied.
sign would give a one-year radioactivity that would be more The
other implosion method is indirect drive. In this ap-
than 1010 less than for a fission breeder. proach the
fuel capsule is not struck by the primary driver
direct
drive is that quite uniform irradiation of the capsule
Parameter Regime with the
soft X rays can be achieved even with anisotropic
20 Reactors, Fusion
center, the fusion burn region will need to spread outward to pro-
An
inertial confinement fusion power plant would consist
of a
target chamber, the driver system of ion accelerators or
layer have included liquid metal flowing over the inner sur-
face
of the target chamber, as well as jets of liquid metal
Both of the ion drive techniques have the major advan- mak
funded by the US Department of Energy at Princeton
tage, relative to most of the laser approaches, that their
University. Similarly the energy leakage of the ions has
efficiency for conversion of electricity to ion energy can
dropped from being many times higher than that theoreti-
be quite high, 25% to 30%. With the heavy ion approach
cally predicted to values which in some cases are at or near
a target energy gain of a factor of 40 might be sufficient the
theoretically best confinement that could ever occur.
for a feasible plant, which is roughly a factor of 4 below
Temperatures in tokamaks have climbed from a few hun-
that required for a KrF laser driver. On the other hand, dred
electron volts to as much as 40 keV to 45 keV in TFTR
for either technology the final path of approach of the ion and
in JT-60U in Naka, Japan. Energy confinement times
beams toward the Hohlraum will be after the final focus- have
grown from a few milliseconds to a second or more on
ing elements of the ion optical system. Maintaining the
Japan’s JT-60U and Europe’s JET tokamaks. The power
necessary illumination intensity and uniformity will be dif- of
neutral beam systems heating the plasma has climbed
ficult as these high space charge density beams converge from
the tens of kilowatts prevalent a four decades ago to
toward even higher concentrations of space charge. Space 40
MW on TFTR and JT-60U. In recent years, neutral beam
charge effects, which arise because of the coulomb repul-
heating systems based upon negative hydrogen and deu-
sion between like charges, are strongly defocussing. In re-
terium ions have been deployed in Japan on the JT-60U
cent years, the vapor pressure in the most common target
tokamak in Naka and the Large Helical Device at Tokai.
chamber concepts is much higher than in earlier scenarios.
Negative ion based beam systems can maintain good neu-
This may result in sufficient plasma being produced along
tralization efficiency at the much higher megavolt beam
the beam trajectory to partially or totally cancel the ef-
energies which will be needed for magnetically confined fu-
fect of the beam space charge forces, and thus simplify the sion
reactors. The JT-60U negative ion based neutral beam
problem of obtaining small beam diameters at the target. has
operated at over 400 kilovolts.
Reactors, Fusion 21
BIBLIOGRAPHY
1986.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3019.htm
} { { } }
HOME
ABOUT US
●
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Gas Turbine Power Stations
Abstract
Advanced Cycles
Summary
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3019.htm17.06.2008 16:39:44
----------------------- Page 184-----------------------
GAS
TURBINE POWER STATIONS
3
2.
High specific power (kWe/kg/s; kWe/m )
3.
Flexibility, availability, and reliability
4.
Short build-up time (preassembled module)
5.
Short start-up time
6.
Remote control (i.e., pipelines)
7.
Water cooling not necessary
8.
Wide power range commercially available (100 kWe to
300 MWe)
9.
Very high efficiency in the cogeneration configuration
10.
Waste gas temperature well suited for combined cycle
emissions
while,
as a negative factor, we can consider the need for clean
fuel
(kerosene, distillate oil, natural gas).
500
mance.
1930 1940 1950 1960 1970 1980 1990 2000 2010 2020 2030
Year
THERMODYNAMIC ANALYSIS OF
inlet air mass flow rate) are represented versus cycle pressure
system, if present
Closed Brayton Cycle (CBC)
3. Thermal losses in the gas turbine hot section
Figure 1. Gas turbine cycle lay-out. C: compressor; T: turbine; CC:
4. Chemical losses correlated to incomplete fuel combus-
tion
combustion chamber; H: heater; GC: gas cooler; W : compressor work;
c
0.7
700
0.6
600
T3 = 1300 C° )
0.5
500 g
Perfect gas k
y η
(
c
T = 1000 C° k
n
3 r
e 0.4
400 o
i
w
c
i
f
c
f
i
f
E
i
e
0.3
300 p
T3 = 1300 C°
0.2
Specific work 200
β
fluence is evident.
Pressure ratio
w
and blade roots, and a substantial quantity of coolant to the
wu = wt − wc = wis ,t · ηis ,t − is ,c
nozzles and the rotor blades. Indeed, to obtain a very high
ηis ,c
gas temperature coupled with a satisfactory blade material
temperature, cooling must be utilized (combustor gas exit and
taking into account that compressor and turbine adia-
temperature 1100–1300#C, material temperature 800– batic
work are quite similar, the effect of turbomachinery in-
950#C).
efficiency is very clear. If the efficiency versus pressure ratio
p3 p4
Cooling
mass flow
Nonisentropic
Heat loss 4 Outlet pressure
compression
loss
Mass loss
4is
Nonisentropic
2
expansion
2is
T
p2
p0 = pamb.
Exhaust gas
p1
1
s
ies are shown.
0.7
T = 1000 °C β = 40
0.6 T = 1300 °C η p = 1
β = 40
30
22
η = 0.9
0.5 p β = 40
16
12
y 0.4 8
c
n
e
i β = 40
c
i
f
f 0.3
E
1612 8
22
0.2
β = 4 β = 4
η = 0.8
0.1 β = 30 p
more
clearly in Fig. 7, where the existing gas turbine pressure
EXISTING GAS TURBINES ratio
is plotted versus power. Aeroderivative systems, taking
also
into account the very high T3 values utilized in aeroen-
The calculation of the thermodynamic cycle of a real OBC gas
gines, show very high pressure ratio values-# # 20 (maximum
turbine, as already stated, is a very complicated matter. Just
efficiency design—see Fig. 5—minimum fuel consumption),
to mention some of the phenomena to be accounted for, it while
lower values for the pressure ratio are utilized for in-
must be considered that: air flows bled from various points of
dustrial gas turbines (maximum specific work-minimum
the compressor are reinjected in the main stream after cooling plant
cost; high outlet turbine temperature T4 value-combined
turbine nozzles, rotor blades, and disks; the compression and cycle
application), considering also the reduced T3 values uti-
expansion process in turbomachinery is neither adiabatic nor lized
in heavy-duty gas turbines.
at constant polytropic efficiency; leakages, pressure, and heat
For small gas turbines (P # 10 MW) # is between 5 and
losses occur in various parts of the machine. Accurate evalua- 15,
due to the impossibility, at commercial costs, of realizing
tion of such complex phenomena can be completely performed
small, high pressure and high efficiency turbomachinery. Fi-
0.45
GE Westinghouse
Allison
0.40
Hitachi
Zosen
0.35
Mashproekt
Solar Turbines
ABB
y 0.30
c
n
e
Heavy-duty
i Pratt & Whitney
c
Mitsubishi HI
i
f
European GT
f 0.25
E
Turbomeca
0.20
Small size
0.15
Kawasaki
Figure 6. Efficiency vs. power of existing
gas turbines (ISO conditions). Three dif- 0.10
0.1 1
10 100 1000
ferent groups are evident: small size, aer-
oderivative, and heavy-duty gas turbines.
Power (MW)
nally, it is worth noting that only two large gas turbines pres- heavy-
duty values are higher than aeroderivatives, but this
ent a very high pressure ratio (# # 30): this value is corre-
difference is not so evident as in the case of the previous data
lated to the nonsimple OBC configuration (reheating cycle),
analyzed (Figs. 7 and 8). Large industrial gas turbines (P #
analyzed in the final part of this paper. 50 MW)
show TOT values higher than 500–520#C, which are
Figure 8 shows a simultaneous increase in specific work
particularly well-suited for a combined cycle solution.
and efficiency of the existing gas turbines. At fixed specific To
complete the analysis of existing gas turbines Fig. 10
work, aeroderivative gas turbines show higher efficiency, shows
the average prices listed only for equipment (not turn-
while at fixed efficiency industrial gas turbines show higher key).
Prices per electrical kilowatt generated shown include a
specific work values (see also Fig. 5). basic
gas turbine generating package with a single-fuel gas
Another thermodynamic feature of existing gas turbines is
turbine, an air cooled electric generator, standard starting
reported in Fig. 9, where turbine outlet temperature (TOT) is system,
skid enclosure, standard inlet and exhaust ducts, ex-
shown. This parameter is particularly important when a gas haust
silencer, and standard controls. Not included are any
turbine is utilized as a topping system in combined cycle con-
financial or debt service costs, contingency and insurance
figurations (the bottoming system is a Rankine cycle heated costs,
legal and environmental permitting costs, and so on.
with the heat recovered from the gas turbine exhaust gas). The
price per kilowatt drops as the unit size (output) of the
All the TOT’s data are in the range 400–600#C, and the gas
turbine increases as a function of the economy of scale.
40
Westinghouse EL Rolls-Royce
Aeroderivative
35
Reheating
turbine
Orenda Aerospace
30
ABB
GE
o
i
t 25
a
r
Small size
e
Allision
r β
u 20 Hitachi
zosen
s
s
e
r
P
15 Kawasaki
Pratt &
Mitsubishi H.l.
10 Whitney
Heavy-duty
5
10 Mashproekt
10 Siemens
Nuovo pignone
Turbomeca
Solar Turbines European GT
Figure 7. Pressure ratio vs. power of existing 0
gas turbines. Three different groups are evi- 0.1 1
100 1000
dent: small size, aeroderivative, and heavy-
duty gas turbines.
Power (MW)
0.45
Westinghouse Rolls-Royce Siemens
Allison
Mitsubishi HI
0.40
GE
Solar Turbines
European GT
0.35 Mashproekt
ABB
Orenda
y Aerospace Nuovo Pignone
c 0.30
n
e
i
c Turbomeca
i
f
f 0.25
E
•
Fabrication methods
GAS TURBINE COMPONENTS AND •
Instrumentation and diagnostic systems
TECHNOLOGICAL INNOVATIONS •
Testing approaches
The main components of OBC gas turbines are (see Fig. 1): As
previously shown one of the major factors in OBC effi-
an axial flow compressor, combustion chamber, and axial flow ciency
advancement has been the continuously increasing
turbine (expander). Only for small systems are the compres- level of
turbine inlet temperature (Fig. 2), made possible by
sor and/or expander of the radial type (centrifugal compres- materials
technology and turbine blade cooling techniques.
sor; centripetal turbine) instead of the axial type. There
are several approaches to the problem associated
To design more effective components, research and devel- with high
gas temperature; in general they can be categorized
opment activities have been carried out mainly in the follow- as
developing suitable materials and cooling systems.
ing areas:
Materials
650
Pratt & Whitney Allison Nuovo Pignone
ABB
Siemens
600 Hitachi Zosen
Kawasaki HI
550
Mitsubishi HI
) 500
C
°
(
p
m 450
e
Westinghouse
t
t Turbomecca
s
u 400
a Solar Turbines Rolls-
Royce
h
x Orenda Aerospace
E 350 GE
300
European GT Mashproekt
250
Power (MW)
heavy-duty gas turbines) are not evident.
1000
900
800
700
)
W
k
/ 600
$
(
t
s
o 500
c
c
i
f
i
c 400
e
p
S
300
200
the turbine first stage fixed blades (nozzles) and moving surface as
well as the purity and temperature of the hot gas.
blades. Heat resistent materials and precision casting are two For
surface temperatures up to around 800#C, chromium
recent advances, largely attributable to developments in air- based
diffusion coatings have proved to be quite effective and
craft engines (Fig. 11). Cobalt based alloys have been used are still
being used today in the rear blade turbine stages
for the first stage fixed blades. These alloys are now being that run
at lower temperatures. Blades operating at higher
supplemented by vacuum-cast nickel-based alloys that are
temperatures require coatings that are thicker and more com-
strengthened through solution and precipitation hardened plex in
composition. Here MCrAlY coatings (M comprises Co
heat treatment. For the moving blades, cobalt based alloys and/or Ni)
are used worldwide for this purpose. The coatings
with high cromium content are now used (4). Ceramic materi- are
applied using the vacuum plasma-spray process. All the
als are also being developed, especially for combustor and tur- coatings
mentioned are consumed while the turbine is in use.
bine inlet fixed blades. A problem here is inherent brittleness, At
present, the service life is still much shorter than actual
which causes fabrication problems and raises uncertainties blade
material and, consequently, the coatings need refur-
about the mechanical properties of ceramic materials (10). bishing
after some 25,000 h of operation.
To ensure the integrity of the blades against chemical at- A
different type of coating is the thermal barrier coating
tack and increasingly higher operating temperatures, protec- (TBC)
(11,12); this has been used for several years in high
tive coatings are utilized (11). Indeed, the intensity of chemi-
performance aero-engines. The importance of such coatings
cal attack is dependent on the temperature of the blade’s will
increase significantly over the coming years because their
1200
engines
PWA 1484
1100
PWA 1480
)
C
MM200 = Hf
°
(
Trend 5 °C/Year
e
MM 247
r 1000
u IN
100
t
MM 247
a 713 – U
r
e
R 80
p U – 500
SC ALLOYS
m
B1900
e 900
t
e X – 40
d
Waspaloy GTD 111
a
DS GTD 111
l
RENE 77
B Nimonic 80 A
U
500
800
IN 738
INCO X-750
General Electric
Figure 11. Aeroengine and heavy-duty N 80 A
heavy-duty
S 816 M
252
gas turbine blade temperature trend. The
700
blade temperature increase is largely at-
1940 1950
1960 1970 1980 1990 2000
tributable to development in aircraft en-
gines.
Year
Turbomachinery
crease in the cycle thermodynamic efficiency of gas turbines
was described earlier. However, to obtain very high tempera-
Significant improvements in existing gas turbines have also
ture coupled with satisfactory blade material temperature, been
obtained through improving the performance of turbo-
cooling must be utilized. The most widespread cooling tech-
machinery. For example, the general trends seen in axial
nique is air cooling, which includes: convection, film, or tran-
compressor design over the past 30 years are: higher speeds,
spirational cooling (1). It is believed that combined convection higher
spool pressure ratio, higher pressure ratio per stage,
and film cooling offers the most promise for air-cooling ap- higher
aerodynamic loading, lower blade aspect ratios, higher
proaches. At current levels of turbine inlet temperature, three blade
row solidities, and improved configurations and blade
or four stages of a turbine rotor may be cooled, and air would shapes
(16). The trend toward higher tip speeds, higher pres-
be bled from earlier stages of the compressor to cool the later sure
rise per stage, and lower aspect ratios is seen for the
stages of the turbine. Bleeding air from earlier stages reduces first
three compressors in Table 1, each of which has a pres-
the work input required to pressurize the cooling air, with sure
ratio between 12 : 1 to 13 : 1. A time of about 20 years
beneficial effects on the net output. is
covered by these three machines while the Ge-NASA E3
1.8
1.6
y
t
i
d
i
l
o
s 1.4
e
g
C
a
r
e
v
a 1.2
e
g
a
t
S
1
0.8
0.5
q 0.45
/
p
g
n
i
d
a 0.4
o
l
e
g
a
r
e
B
v
A
0.35
0.3
25
6
Spool
pressure ratio
5
20
o
o
i
i
t
4 t
a
a
r
r
e 15
t
c
r
e
u
Average aspect ratio p
s
s
3 s
e
a
r
e
p
g
l 10
o
a
r
o
A 2 e
p
v
S
A
1
Figure 12. Axial flow compressor design param-
eters trend: higher speed, higher spool pressure
0
0
ratio, higher aerodynamic load, lower blade as-
1950 1955 1960
1965 1970 1975 1980 1985 1990 1995
pect ratios, higher blade row solidities (see also
Table 1).
Year
Lower aspect ratios, plus higher solidity and higher stagger pactness
and ruggedness, blading arrangements, and higher
blading, are the major design advancements that make this stage
loading with good performance.
possible. The trend toward lower blading aspect ratios in Similar
considerations can be carried out for the gas tur-
multistage compressors over the past 30 years is clearly bine
expander, particularly for blade stages, transonic and su-
shown in the figures. Average aspect ratios of about 1.3 to 1.4 personic
design, tip clearance, and endwall effects.
are now common. Similar considerations can be carried out
for solidity values.
Combustion
Chamber
In conclusion, the application of advanced design methods
and research and development results has led to the compres- Figure 13
shows a simplified representation of a gas turbine
sor system having: improved efficiency and stall-margin, com- combustion
chamber: the diffuser, the swirler, the liner, the
x 2 2
x
Air
specified. For dry low NOx systems three different designs
ADVANCED CYCLES
Figure 13. Simplified lay-out of a gas turbine combustion chamber.
As
already stated, the simple OBC gas turbine performances
theless, the effect of the colder air in the last stages of the
• Selective catalytic reduction (SCR)
compressor, which is very useful for turbine cooling purposes,
120
30
Temperature range
) 100
for low emissions 25 )
v
NO v
m
x m
p
p
p
p
(
(
80
20
e
n
d
e
i
CO g
x
o
o
r
60
15 t
n
i
o
n
m
f
o
n
o 40
10 s
e
b
d
r
a
i
x
C 20
5 O
B
A
C
2
1′ 2′′ 3 3′ 3′′ 4
2′
LPC
HPC HPT LPT
Wu
m cooling
3′
3′
2"
4
Figure 15. Three modifications of OBC simple cycle T
Q reg.
5
heating (A: intercooler, B: regenerator, C: combustion
2
chamber, LPC and HPC low and high pressure com-
1′ 1
pressor, LPT and HPT low and high pressure
turbine.
S
0.6
Near-term combined
Combined cycles
#####
0.55 cycles β = 30
24 18 15 12 3 lev. + RH.
2 lev.
Reheating cycles, #####
β hpt = 2
1 lev.
0.5
β =
30 27 24 21 18
15
Regenerative cycles G /G =
0.127 0.152 0.167 0.186
w a
0.212
y 0.45 10 12
0.249
c 8
65
n β = 80
e 15
50
i β = 6
c
i
40
f
f 0.4 18 β = 36
30 STIG cycles
E 30
β = 30 24
24
24
18 18
0.35 18
15 15
Intercooled cycles
OBC 12
(Simple cycles) 12 β = 30
0.3 24
Reheating cycles,
10
8 18
β hpt = SQRT( )β
15
6
0.25
250 300 350 400 450
500 550 600 650 700
Specific work
(kJ/kg)
similar to the simple cycle gas turbine, but not so high as to tive
systems. Another interesting aspect of the steam injec-
justify the technological difficulties associated with the re- tion
cycle is the very high performance when electricity is
combustion process. However, the associated combined cycle
generated together with heat—cogeneration power plant—
performance is very high, about 58%, thanks to the high tur- and
also when water recovery is utilized (24). However, the
bine outlet temperature values (#600#C) and the economic re- steam
injection solution requires very clean water (demineral-
sults also seem to be very interesting. ized
water; careful control of the compressor surge margin
Obviously, several combinations of nonsimple solutions limit;
new compressor-expander matchline; mechanical stress
can be utilized, with intriguing performances. An interesting
limitation due to the high power increase (see Fig. 16). An
aspect of the regeneration and intercooled or reheated cycle
offshoot of steam injection in OBC systems is the coupling
is that regeneration avoids the negative thermodynamic as- with
the intercooling steam injection gas turbine cycle (IS-
pect of intercooling or reheating previously discussed (more TIG);
in this case the efficiency can be higher than 50%.
heat entering in the cycle).
New cycles are also under investigation all around the
Other systems based on OBC technology are steam injec- world,
and, at this moment, the most promising is the HAT
tion turbines (STIG (20), CHENG (21)), humid air turbines (humid
air turbine) or CHAT (cascaded humid air turbine).
(HAT (22)), cascaded humid air turbine (CHAT (23)) or more The
HAT cycle is a regenerative cycle with a complex lay-out:
complex solutions. the
high temperature heat at the expander outlet is trans-
The most utilized is the steam injection cycle, where the ferred
to the hot and water saturated air at the compressor
steam is generated by recovering the waste heat of the ex-
outlet. Some preliminary performance evaluations for the
haust gas at the expander exit, through the use of a heat HAT
show a cycle efficiency value around 55% and very high
recovery steam generator. Normally the steam pressure is
specific work values (600–650 kW/kg/s), and higher for the
1.25–1.30 times the maximum OBC pressure to allow steam CHAT
solution.
injection in the combustion chamber and in the power turbine A
last very promising solution is the possible coupling be-
stages. The high cp value of steam or water increases the out- tween
the gas turbine simple-cycle with solid oxide fuel cells
put of the gas turbine for fixed temperature drop (the ex- (SOFC)
(25). In this case an efficiency value higher than 70%
pander working fluid is a mixture of combustion gases and seems
to be possible.
steam) and decreases the combustion flame allowing a reduc-
tion of No emissions.
x
has taken place mainly in the last twenty years, and corre-
18. T. Korakianitis and K. Beier, Investigation of the part-load per-
lates closely to the research and development activities in the
formance of two 1.12 MW regenerative marine gas turbines,
aeroengine field.
ASME Paper 92-GT-86, ASME Trans. J. Eng. Gas Turbines
However, at this moment, very important reasons for the
Power, 1994.
gas turbine power plant’s success are also the low cost of en-
19. W. Jury and D. Searles, Process optimization of an integrated
ergy generation utilizing natural gas, the reduced environ-
combined cycle. The impact & benefit of sequential combustion,
mental impact, the wide availability and reliability, and fi-
ASME Paper 97-GT-490, 1997.
nally the possibility to be utilized in the combined cycle
20. D. H. Brown and A. Cohn, An evaluation of steam injected com-
configuration (the heat of the high temperature exhaust
bustion turbine systems, ASME Trans., J. Eng. Power, 103: Janu-
ary 1981.
gas—see Fig. 9—is recovered in a bottoming steam Rankine
cycle). The combined cycle solution shows efficiency greater
21. M. A. Saad and D. Y. Cheng, ‘‘CHENG CYCLE II’’ recent ad-
than 55%, and in the short term period close to 60% (see Fig.
vancements in gas turbine steam injection, Flowers ’92, Proc., I:
83–96, 1992.
16), while an increase in the specific work of about 60% is
evident compared to the OBC simple cycle.
22. A. D. Rao, United States Patent 4,829,763: Process for Producing
340, 1997.
3. C. F. McDonald, The future of the closed-cycle gas turbine for
terrestrial applications, Proc. 1987 ASME Cogen Turbo Symp.
ARISTIDE F. MASSARDO
IGTI, 1: 238–246.
`
Universita di Genova
4. A. F. Massardo and C. F. McDonald, Closed cycle gas turbine
perspective: enabling technologies from combustion turbines the
key factor for future deployment, VIII Advanced Energy Systems
Conf., Bologna, Italy, 1995.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3021.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Geothermal Power
Abstract
Future Developments
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3021.htm17.06.2008 16:40:02
GEOTHERMAL POWER
Electric
power was first produced from geothermal energy in
1904 when
a 560 W generator was placed in operation at the
THE
NATURE OF GEOTHERMAL RESOURCES
Geopressured resources
9
turbine but smaller in size for the same output power.
stalled capacity, and produces some 17 # 10 kWh of electric-
Conversion
Geoscience
Discovery
Effluent
abatement
Geoscience
modeling
Numerical
modeling
Well drilling
Injection
and
completion
wells
Production
wells
Turbine
Flash
generator Cooling
tower
separator
From
production
wells
Gas
removal
To
injection
To injection
wells
wells
Condenser
(a)
Turbine
generator Cooling tower
Heat exchanger
From
production To
Installed and
Breakeven
Annual Plant Cost
O&M Costs Fuel Costsa Busbar Price
pacity.
Environmental Considerations
Power Plants (Davis, CA, 1991); L. McLarty, DynCorp EENSP, Inc. (Alexan-
dria, VA, August 1995); G. W. Huttrer, International Geothermal Association,
The exploitation of geothermal energy has a net positive im-
Proceedings of the World Geothermal Congress (Florence, Italy, 1995).
u
but in finding both adequate permeability and fluids that are
recoverable in amounts sufficient to supply a commercial-size
Wnet Wnet
power plant. In any geothermal exploration program, an ade-
ηu = #B = m · #B
Advanced
flash system—even for high-temperature resources—because 100
conventional
of its higher efficiency and environmental acceptability.
technology
A geothermal combined cycle power plant, commercialized
Base case:
by ORMAT, efficiently extracts the energy contained in the
Today’s technology
typical mixture of steam and brine flowing from geothermal
s
wells. In this system the geothermal fluid flows directly into r
a
l
l
Problem
a steam separator with the separated high-pressure steam o 10
burdened
used to drive a back-pressure turbine. Low-pressure steam, S
U
which exits the back-pressure turbine, flows into the vapor- 4
izer of an organic cycle binary system wherein its heat of con- 9
9
1
densation is added to the thermal energy of the separated f
o
brine to vaporize an organic fluid, which is used to drive a s
Linear drilling
n
technology
binary turbine. o 1
i
l
l
System Application. Geothermal power plants are generally i
M
baseload systems, but may sometimes be used in a load-fol-
Oil and gas
lowing mode. Current contractual capacity factors for most
average
geothermal plants are on the order of 80%. However, actual
capacity factors for some operating plants approach 100%
0.1
5000 10000 15000 20000 25000 30000
(24). System capacity factor is defined based on nameplate
(ft)
rating:
0 2 4 6 8 10
kWh output per year
Well depth (km)
Capacity factor =
(Nameplate kW)(8760 hours per year) HDR
actual Hydrothermal actual JAS ultra-deep SPE oil and gas
wells
from the same drill pad, rather than drilling at several
Advances in Geothermal Technology
Existing
flash
tank
Steam
Central steam
turbine plant
Existing low pressure two-phase flow ###
############
###
############
Brine
Steam
back
############
Flash into
orifice Separated existing
steam
############
system
############
############
############
############
two flow turbine
############
Geothermal
wells
Recuperative Ammonia–water
heat exchanger working fluid
Turbine
Air-cooled
Recuperative condenser
preheater
Brine Brine
heat heat
exchanger exchanger
Pump
Figure
6. Schematic of Kalina cycle system, which utilizes
Hot brine To reinjection well
ammonia–water working fluids and cascaded recuperative
from geothermal well heat
exchangers to increase power conversion efficiency.
2 x x
ment of hot dry rock and magma resources are underway in 20.
Anonymous, Managing maturity, in United States Geothermal
Europe, Japan, and the United States. If successful, these ef-
Operations, Los Angeles: Unocal Corp., 1995, p. 2.
forts will make abundant geothermal energy universally 21.
M. Dellinger, Turning community wastes into sustainable geo-
available to humankind.
thermal energy—the S.E. Geysers effluent pipeline project,
23.
G. Shulman, Low temperature flashed steam power generation,
1. P. D. Blair, T. A. V. Cassel, and R. H. Edelstein, Geothermal En-
in Proceedings of the World Geothermal Congress, 1995, Taupo,
ergy: Investment Decisions and Commercial Development, New
N.Z.: International Geothermal Association, 1995, Vol. 3, pp.
York: Wiley, 1982.
2059–2061.
1987.
3. D. Duchane, Geothermal energy, in Kirk-Othmer Encyclopedia of
Chemical Technology, New York: Wiley, 1994, 4th Ed., Vol. 12, 25.
J. W. Tester and H. J. Herzog, Economic Predictions for Heat
pp. 512–539.
Mining: A Review and Analysis of Hot Dry Rock (HDR) Geothermal
18. J. W. Tester, Energy conversion and economic issues for geother- 39.
R. H. Wallace et al., Assessment of geopressured–geothermal re-
mal energy, in Handbook of Geothermal Energy, Houston, TX:
sources in the northern Gulf of Mexico basin, Geol. Surv. Circ.
Gulf Publishing, 1982, pp. 471–588.
(U.S.) 790: 1978.
41. USGS, Potential of hot dry rock geothermal energy in the eastern
United States, Geol. Surv. Open-File Rep. (U.S.) 93-377: 1993.
Reading List
JOHN E. MOCK
Technology Transfer International
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Homopolar and Inductor Machines
● Email this page Enter words or
Standard Article
Abstract
Homopolar DC Machines
Inductor Machines
Appendix
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3004.htm17.06.2008 16:40:21
Electric machines are energy converters in which force and torque are normally
produced by a load current
in the presence of a magnetic field. This article is devoted to a
special class of machines in which the load
winding is induced by a homopolar flux that generates dc voltage or by a flux
modulation due to the variable
reluctance of one member that produces ac voltage. These machines differ in
construction from conventional
dc and ac machines, and they are designed mostly for special applications. The
following paragraphs describe
the concepts of heteropolar and homopolar excitation and the derived machine
concepts within the scope of
this article.
Heteropolar Excitation
Rotating electrical machines are mostly heteropolar, where magnetic north and
south poles are placed
with alternating polarity along the circumference. In synchronous structures,
the magnetic flux is supplied
by one member, which is generally the rotor. In conventional dc machines which
belong to the synchronous
structures, the flux supplying member is the stator. Asynchronous structures, where
the field generated by ac
currents in the primary member is also heteropolar, are not considered here.
In machines with alternating pole polarity, the airgap field curve does not
contain a homopolar component.
The voltage induced in the armature winding on the other member of the machine
during rotation is determined
by the law of induction. Its basic integral form, valid for a conductor loop of
area A with boundary line s moving
with velocity v in a field with magnetic flux density B , is given here:
Note that B , v, A , and s are vectors. Applied to a coil in a machine, the first
term on the right side of Eq. (1)
represents the transformer voltage component due to time-variable flux, and the
second term, due to the speed
of the coil moving in the field, denotes the rotational voltage component.
Heteropolar structures have a pole
pitch defined by the field exciting member. In a dc machine where axial conductors of
active length l move with
constant tangential velocity v in a radial field B , the voltage induced in a coil
having N c turns of 180◦ pitch is
where
Hence for constant speed, the ac voltage waveform reflects the flux
density waveform. For sinusoidal field
distribution and constant speed, the generated voltage is also a sine
wave. In dc machines the alternating
Type (2): cyclic machines, homopolar inductor machines, and hybrid motors
or alternatively
HOMOPOLAR AND
INDUCTOR MACHINES 3
Although the medium-frequency generators, well known as Lorenz and Guy type
machines, have lost their
market to static power electronic equipment, claw-pole machines and especially
transversal-flux machines are
under consideration for direct-driven, low-speed generators or vehicle motors, see
Ref. 17. Within the scope
of this article, of the heteropolar structures only the inductor machine is
considered because its behavior is
similar to the homopolar inductor machine.
Machines with Homopolar Excitation
voltage u = −d#/dt in this circuit is given by the rate of change of the flux linked
with area A . The boundary
of A represents one turn, and the flux linkage is equal to the flux, # = # =B ·
A . Because the flux density is
constant, u = −B · dA/dt. There is only one portion of A which changes due to
rotation. When the rotor surface
speed is v, the winding area decreases so that dA/dt = −lv, and we end up with:
This equation is formally the same as for one armature conductor of a conventional
dc machine; see also Eq.
(2). Here with constant flux density and speed the equation describes a dc voltage.
The inherent absence of
voltage harmonics is the reason that homopolar acyclic machines are recommended as
tacho generators.
Cyclic Induction. For the cyclic types of machine, consider Fig.
3. Both excitation 1 and armature
winding 2 are placed on the same stationary member. The rotor has a salient
structure. The homopolar flux
passes from stator to rotor in the left part I and from rotor to stator in the
right part II. The pole pitch of the
ac winding is related to the number of rotor teeth. Depending on the rotor
position, the flux linkage of a coil
varies between a maximum and a minimum value. Suppose the variable reluctance has
only a fundamental
#0 is the mean value, and #a the resulting flux component linked with the ac
winding coil. When a machine
that has zr rotor projections rotates with constant angular speed #, the induced
voltage in a coil of N c turns is
given by
These machines utilize only up to half of the power per volume compared with
heteropolar machines because
twice the amplitude of the ac flux density component cannot exceed the homopolar flux
density.
----------------------- Page 218-----------------------
HOMOPOLAR AND
INDUCTOR MACHINES 5
Homopolar DC Machines
HOMOPOLAR
AND INDUCTOR MACHINES 7
Inductor Machines
and
where λ0 is the average specific magnetic permeance of the air gap and γ is the
modulation factor.
Because the period of λ is the rotor tooth pitch, it consists only of uneven
components. Note that in Eq. (4)
field harmonics above the fundamental were neglected. Then the machine model is
similar to a conventional
turbo rotor synchronous machine, except that the armature reaction needs special
consideration when describ-
ing the mutual inductance of the field and the ac winding. The direct and quadrature
axis (d,q)-components of
the armature flux are given in terms of the armature current d,q-components and the
field current by
◦
◦
on the dc side. The firing angle was α = 143.4 . The overlap of u = 0.4
was almost negligible. This results in a
margin angle of γ = 37◦ with an equivalent of 342 µs for the thyristors to
recover. Under these circumstances
Fig. 13. Comparison of motor voltages and currents (a) for dc side commutated
inverter (DCI ); (b) for load-commutated
inverter (LCI ).
synchronous reactance was X d = 0.9 p.u. The firing angle was adjusted to
yield the same power factor cos
φ = 0.8.
The analysis of the harmonic contents of machine current and
terminal voltages shows remarkable
differences between the LCI and the DCI performance. In Table 1 the
fifth- and seventh-order harmonics
reveal only a remarkably small deviation from a sinusoidal shape with the DCI.
Consequently, the pulsating
torque component generated by a DCI drive is also smaller than with a LCI
configuration.
A deeper insight into the characteristics of the converter fed machine can be
obtained by inspecting the
trajectories of voltages and current in the space-phasor plane. It is well known
that the steady-state, 120◦ block
HOMOPOLAR AND
INDUCTOR MACHINES 13
Of these types the hybrid motor allows for small step angles and displays
large holding torques. Hence
it is preferred for industrial applications. Figure 16a shows the principle of
construction. Note that the cogged
rotor disks embracing the axially magnetized permanent magnet are displaced by one-
half of the tooth pitch.
The step angle is given by
◦
The relatively simple motor depicted in Fig. 16(b) has m = 2 phases and zr = 9
rotor teeth. Hence α = 10 . Phase
numbers up to m = 5 are in use.
Because a hybrid motor is also a synchronous machine, the relevant analysis
can be adapted. Assuming
a two-phase symmetrical system at constant synchronous speed with
sinusoidal waveforms of voltages and
Fig. 14. Space phasor trajectories for steady-state operation of a DCI: (a,b) in
stator frame; (c,d) in synchronous rotating
frame.
currents, the voltage equations in d,q-components can be written as
where
τ = L/R is the winding time constant with L = (Ld +Lq )/2
q = (Ld −Lq )/(Ld +Lq ) is the reluctance factor
ω = zr · dα/dt is the motor speed in electrical rad/s
HOMOPOLAR
AND INDUCTOR MACHINES 15
This model does not take into account iron losses due to alternating magnetization.
It needs the addition of
damper windings to describe the motor behavior in constant current operation.
Stepping motors are electromagnetic energy converters that render a
certain angle increment on the
application of one impulse. The number of steps which the rotor proceeds
in 1 s corresponds to the step
frequency f z . Two-phase motors can be supplied in full-step or half-step
operation (Fig. 18).
In practice the maximum torque decreases with increasing step
frequency. It is usual to specify the
performance of a motor by three limiting torque curves which imply synchronous
operation on a sinusoidal
supply (see Fig. 19):
HOMOPOLAR
AND INDUCTOR MACHINES 17
Fig. 18. Step rate and excitation pattern for two-phase, full-
step and half-step operation.
◦
instance a 200 step (1.8 ) motor can be run with five intermediate positions per
full step by controlling the
phase currents with a supplementary unit.
The standard operation of stepping motors is in open-loop control.
When loaded above the maximum
available torque, the motor loses steps. In recent times strategies for a closed-
loop operation have been intro-
duced by which the performance is made similar to an electronically commutated
(EC ) motor and the main
disadvantage of stepping motors is overcome. This is especially true when available
methods for sensorless
angle determination are applied.
Appendix
Homopolar Machine A machine in which the magnetic flux passes in the same
direction from one member to
the other over the whole of a single air-gap area (411-31-02)
HOMOPOLAR AND
INDUCTOR MACHINES 19
Acyclic Machine A direct current homopolar machine (411-31-03)
Heteropolar Machine A machine having successive physical or effective poles
of opposite polarity (411-31-04)
Inductor Machine A synchronous machine in which one member, usually
stationary, carries armature and
excitation windings or permanent magnets effectively disposed relative to each
other, and in which the other
member, usually rotating, is without windings but carries a number of regular
projections (411-31-11)
BIBLIOGRAPHY
MANFRED H. STIEBLER
KEZHONG GUO
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3022.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
HVDC Power Converters
Canada
❍ Advanced Product
Copyright © 1999 by John Wiley & Sons, Inc. All rights
Search
reserved.
❍ Search All Content
DOI: 10.1002/047134608X.W3022
❍ Acronym Finder
Article Online Posting Date: December 27, 1999
Abstract | Full Text: HTML PDF (366K)
Abstract
Configurations
Environmental Considerations
DC Converter Operation
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3022.htm17.06.2008 16:40:44
The question is often asked, “Why use dc transmission?” One response is that losses
are lower, but this is not
strictly correct. The level of losses is designed into a transmission system and is
regulated by the size of the
conductor selected. Dc and ac conductors, as overhead transmission lines or
submarine cables, can have lower
losses but at higher expense because the larger cross-sectional area generally
results in lower losses but costs
more.
When converters are used for dc transmission in preference to ac
transmission, it is generally by economic
choice driven by one of the following reasons:
(1) An overhead dc transmission line with its towers can be designed to cost less
per unit of length than an
equivalent ac line designed to transmit the same level of electric power.
However the dc converter stations
at each end are more costly than the terminating stations of an ac
line, and so there is a breakeven
distance above which the total cost of dc transmission is less than
its ac transmission alternative. The
dc transmission line can have a lower visual profile than an
equivalent ac line and so contributes to a
lower environmental impact. Other environmental advantages to a dc transmission
line are dc instead of
ac electric and magnetic fields.
(2) If transmission is by submarine or underground cable, the breakeven distance
is much less than overhead
transmission. It is not practical to consider ac cable systems exceeding 50 km,
but dc cable transmission
Configurations
The integral part of an HVDC power converter is the valve or valve arm. It is
noncontrollable if constructed
from one or more power diodes in series and controllable if constructed from one or
more thyristors in series.
Figure 1 depicts the International Electrotechnical Commission (IEC) graphical
symbols for valves and bridges
(1). The standard bridge or converter connection is defined as a two-way connection
comprising six valves or
valve arms which are connected as illustrated in Fig. 2. Electric power flowing
between the HVDC valve group
and the ac system is three-phase. When electric power flows into dc valve group from
the ac system, then it is
considered a rectifier. If power flows from the dc valve group into the ac system, it
is an inverter. Each valve
consists of many series-connected thyristors in thyristor modules. Figure 2 depicts
the electric circuit network
for the six-pulse valve group. The six-pulse valve group was usual when valves were
mercury arc.
Twelve-Pulse Valve Group. Nearly all HVDC power converters with
thyristor valves are assembled
in a converter bridge of twelve-pulse configuration. Figure 3 illustrates the use of
two three-phase converter
transformers with one dc side winding, one as an ungrounded star connection and the
other a delta configura-
tion. Consequently the ac voltages applied to each six-pulse valve group of the
twelve-pulse valve group have
a phase difference of 30◦ which is utilized to cancel the ac side fifth and seventh
harmonic currents and, the
Fig. 2. Electric circuit configuration of the basic six-pulse valve group with its
converter transformer in star–star connec-
tion.
the outline around each of the three groups of four valves in a single vertical
stack. Known as “quadrivalves,”
they are assembled as one valve structure by stacking four valves in
series. Because the voltage rating of
thyristors is several kilovolts, a 500 kV quadrivalve may have hundreds of
individual thyristors connected in
series groups of valve or thyristor modules. A quadrivalve for a high-voltage
converter is mechanically quite
tall and may be suspended from the ceiling of the valve hall, especially in
locations susceptible to earthquakes.
Fig. 3. The twelve-pulse valve group configuration with two converter transformers,
one in star–star connection and the
other in star–delta connection.
with wall bushings. When applied at dc voltages of 400 kV or greater, wall bushings
require considerable design
and care to avoid external or internal insulation breakdown.
Harmonic filters are required on the ac side and usually on the dc side. The
characteristic ac side current
harmonics generated by six-pulse converters are 6n ± 1 and 12n ± 1 for 12-pulse
converters where n equals
all positive integers. Ac filters are typically tuned to the 11th, 13th, 23rd, and
25th harmonics for 12-pulse
converters. Tuning to the fifth and seventh harmonics is required if six-pulse
converters are permitted. Ac side
harmonic filters may be switched with circuit breakers or circuit
switchers to accommodate reactive power
requirement strategies because these filters generate reactive power at
fundamental frequency. A parallel
resonance is naturally created between the capacitance of the ac filters and the
inductive impedance of the
ac system. For the special case where such a resonance is lightly damped and tuned
to a frequency between
the second and fourth harmonics, then a low-order harmonic filter at the
second or third harmonic may be
required, even for 12-pulse converter operation.
Characteristic dc side voltage harmonics generated by a six-pulse converter
are of the order of 6n and
when generated by a 12-pulse converter, are of the order of 12n. Dc side filters
reduce harmonic current flow
on dc transmission lines to minimize coupling and interference with adjacent voice
frequency communication
circuits. Where there is no dc line, such as in the back-to-back configuration, dc
side filters may not be required.
Dc reactors are usually included in each pole of a converter station. They
assist the dc filters in filtering
harmonic currents and smooth the dc side current so that a discontinuous
current mode is not reached at
low load current operation. Because the rate of change of dc side
current is limited by the dc reactor, the
commutative process of the dc converter is more robust.
Surge arresters across each valve in the converter bridge, across each
converter bridge, and in the dc and
ac switchyards are coordinated to protect the equipment from all overvoltages
regardless of their source. They
may be used in nonstandard applications, such as filter protection. Modern HVDC
substations use metal oxide
arresters whose rating and selection are made with careful insulation coordination
design.
fired thermal generation can be located close to load centers and may delay
development of isolated energy
sources in the near term.
(3) Importing electric energy into congested load areas. In areas where new
generation is impossible to put
in service to meet load growth or replace inefficient or
decommissioned plants, underground dc cable
transmission is a viable means of importing electricity.
(4) Increasing the capacity of existing ac transmission by conversion to dc
transmission. New transmission
rights-of-way may be impossible to obtain. Existing overhead ac transmission
lines upgraded to or overbuilt
with dc transmission can substantially increase the power transfer capability
on the existing right-of-way.
(5) Power flow control. AC networks do not easily accommodate desired power-flow
control. Power marketers
and system operators may require the power-flow control capability provided by
HVDC transmission.
(6) Stabilization of electric power networks. Some widespread ac power system
networks operate at stability
limits well below the thermal capacity of their transmission conductors. HVDC
transmission is an option
to increase utilization of network conductors along with the various electronic
power controllers which can
be applied to ac transmission.
HVDC
POWER CONVERTERS 7
(1) Back-to-back. There are some applications where the two ac systems to be
interconnected are in the same
physical location or substation. No transmission line or cable is required
between the converter bridges
in this case and the connection may be monopolar or bipolar. Back-to-back dc
links are used in Japan for
interconnections between power system networks of different frequencies (50 and
60 Hz). They are also
used as interconnections between adjacent asynchronous networks.
(2) Transmission between two substations. When it is economical to transfer
electric power through dc trans-
mission or cables from one geographical location to another, two-terminal or
point-to-point HVDC trans-
mission is used. In other words, dc power from a dc rectifier terminal is
dedicated to one other terminal
operating as an inverter. This is typical of most HVDC transmission systems
HVDC
POWER CONVERTERS 9
Environmental Considerations
ion density in the air. A static magnetic field is produced by dc current flowing
through the conductors. Air ions
produced by HVDC lines form clouds which drift away from the line when blown by the
wind and may come in
contact with humans, animals, and plants outside the transmission line’s right-of-
way or corridor. The corona
effects may produce low levels of radio interference, audible noise, and ozone.
Field and Corona Effects. The field and corona effects of
transmission lines largely favor dc over ac
transmission. The significant considerations are as follows:
(1) For a given power transfer requiring extra high voltage transmission, the dc
transmission line has a tower
profile smaller than the equivalent ac tower carrying the same level
of power. This can also lead to a
narrower right-of-way for dc transmission.
(2) The steady and direct magnetic field of a dc transmission line near or at the
edge of the transmission right-
of-way has about the magnitude as the earth’s naturally occurring magnetic
field. For this reason alone, it
seems unlikely that this small contribution by HVDC transmission lines to the
background geomagnetic
field is a basis for concern.
(3) The static and steady electric field from dc transmission at the
levels experienced beneath lines or at
the edge of the right-of-way have no known adverse biological effects.
There is no theory or mechanism
to explain how a static electric field at the levels produced by dc
transmission lines could affect human
health. The electric field level beneath a HVDC transmission line has a
magnitude similar to the naturally
occurring static field which exists beneath thunder clouds. Electric fields from
ac transmission lines have
undergone more intense scrutiny than fields generated by dc transmission lines.
(4) The ion and corona effects of dc transmission lines contribute a
low level of ozone production to higher
naturally occurring background concentrations. Exacting long-term measurements
are required to detect
such concentrations. The measurements taken at cross-sections across the Nelson
River dc lines in Canada
failed to distinguish background from downwind levels (4). Although solar
radiation influences the produc-
tion of ozone even in a rural environment, thereby maintaining its level, any
incremental contribution from
a dc line is subject to breakdown, leading to a resumption of background levels
downwind from the line.
Investigations of ozone in indoor conditions indicate that in well-mixed air,
the half-life of ozone is 1.5 min
to 7.9 min. Increases in temperature and humidity increase the rate of decay
(4).
DC Converter Operation
The six-pulse converter bridge of Fig. 2, the basic converter unit of HVDC
transmission, is used equally well
for rectification when electric power flows from the ac to the dc side and for
inversion when the power flow is
from the dc to the ac side. Thyristor valves operate as switches which turn on and
conduct current, when fired
on receiving a gate pulse, and are forward-biased. A thyristor valve conducts
current in one direction and, once
it conducts, turns off only when it is reverse-biased and the current falls to
zero. This process is known as line
commutation and is discussed in more detail below.
An important property of the thyristor valve is that once its conducting
current falls to zero, when it is
reverse-biased and the gate pulse is removed, too rapid an increase in the
magnitude of the forward-biased
voltage causes the thyristor to turn on inadvertently and conduct. The
design of the thyristor valve and
converter bridge must ensure that such a condition is avoided in useful inverter
operation.
Commutation. Rectification or inversion for HVDC converters
is accomplished through a process
known as line or natural commutation. The valves act as switches so
that the ac voltage is sequentially
switched to provide always a dc voltage. With line commutation the ac voltage at
both the rectifier and inverter
must be provided by the ac networks at each end and should be three-phase and
relatively free of harmonics, as
depicted in Fig. 8. As each valve switches on, it begins to conduct current while
the current begins to fall to zero
in the next valve to turn off. Commutation is the process of transferring current
between any two converter
valves, and both valves carry current simultaneously during the process.
Consider the rectification process. Each valve switches on when it receives a
firing pulse to its gate and
its forward-bias voltage becomes more positive than the forward-bias
voltage of the conducting valve. The
current flow through a conducting valve does not change instantaneously as it
commutates to another valve
because the transfer is through transformer windings. The leakage reactance of the
transformer windings is
also the commutative reactance so long as the ac filters are located on the primary
or ac side of the converter
transformer. The commutative reactance at the rectifier and inverter is shown as an
equivalent reactance X C
in Fig. 8. The sum of all of the valve currents transferred to the dc side and
through the dc reactor is the direct
current, and it is relatively flat because of the inductance of the dc reactor and
converter transformer.
At the inverter, the three-phase ac voltage supplied by the ac system
provides the forward- and reverse-
bias conditions of each valve in the converter bridge to allow commutation of
current between valves, as in
the rectifier. The inverter valve can only turn on and conduct when the positive
direct voltage from the dc line
is greater than the back negative voltage derived from the ac
commutative voltage of the ac system at the
inverter.
Because of the line-commutative, valve-switching process, a nonsinusoidal
current is taken from the ac
system at the rectifier (Ivr in Fig. 8) and is delivered to the ac system at the
inverter (Ivi in Fig. 8). Both Ivr and
Ivi lag the alternating voltage. This nonsinusoidal current waveform consists
of the fundamental frequency
ac component plus higher harmonics taken from and injected into each ac
system. The ac filters divert the
harmonics from entering the ac system by offering a low impedance bypass path that
allows the commutative
voltage to be relatively harmonic-free ( ULr and ULi in Fig. 8).
where I dN is the rated direct current and UVN is the rated phase-to-phase
voltage on the valve or secondary side
of the converter transformer. Usually the dc converter bridge power rating is known
from its rated dc current
I dN and rated dc voltage UdN . The valve and converter bridge design are very
dependent on the commutative
reactance X C , and so consequently its value is established and known. In modern
HVDC converter bridges it
2
is usually in the range 0.1 <X C < 0.15 per unit where 1.0 per unit is (UVN ) /SN
#.
A reasonably good approximation for the power factor of a converter bridge at
the ac commutating bus is
given by the following expression for a rectifier. Note that the delay angle α is
usually known or determined.
◦ ◦
For example, the normal steady-state range of the delay angle for a
rectifier may be 10 < α < 18 , and the
◦
lowest normal operating power factor occurs when α = 18 :
losses in the converter bridge, the power flowing through the bridge Pd is given by
where I d is the operating direct current through the converter bridge and Ud
is the operating direct voltage
across the converter bridge. Having calculated the power factor angle θ from Eq.
(4) or (5) and the throughput
power of the converter bridge from Eq. (6), the reactive power QL
demanded by the converter bridge at the ac
commutating voltage busbar at either the rectifier or inverter is given by
It may be that the rated phase-to-phase voltage UVN on the valve or secondary side
of the converter transformer
is not known. It is possible to compute what it should be if the power factor cos θ
from Eq. (4) or (5) is known
at the converter bridge rating. Then a good estimate of UVN is given
by
Once UVN is known, it is possible to find the converter transformer rating from
Eq. (3).
It may be necessary to determine the overlap angle µ. At the rectifier, the
following approximate expression
can be applied when the delay angle α per unit of commutating reactance X C
and the dc load current I d are
known:
Similarly at the inverter, the extinction angle γ is usually known for steady-state
operation so that
The delay angle α at the inverter may not be inherently known, but once the
extinction angle γ and the overlap
angle µ have been determined, then
where X C is the commutating reactance for the converter bridge per unit, φ = α
for a rectifier, and φ = γ if an
inverter. I dN is the rated dc current for the converter bridge and UdN is its
rated dc voltage.
Equations 1 to 13 are the steady-state and reasonably accurate expressions
that define the state of a six-
pulse converter bridge under ideal conditions. More exacting expressions can be
found in Refs. (6,7,8,9,10,11)
and can be used if the network data are known with sufficient accuracy to justify
precise mathematical formu-
lation. Defining the performance and operation of a converter bridge under dynamic
or transient conditions
requires a suitable electromagnetic transients simulation program with
capability of modeling the valves,
converter transformer, the control system which produces the firing pulses to the
valves, and the associated ac
and dc networks.
Short Circuit Ratio. The strength of an ac network at the bus of a
HVDC substation can be expressed
by the short circuit ratio (SCR), defined as the relationship between the short
circuit level in MVA at the HVDC
substation bus at 1.0 per unit ac voltage and dc power in MW.
The capacitors and ac filters connected to the ac bus reduce the short circuit
level. The effective short
circuit ratio (ESCR ) is the ratio between the short circuit level
reduced by the reactive power of the shunt
capacitor banks and ac filters connected to the ac bus at 1.0 per unit voltage and
the rated dc power.
A lower ESCR or SCR means more pronounced interaction between the
HVDC substation and the ac
network (9,10). Ac networks can be classified in the following categories according
to strength:
In high ESCR systems, changes in the active/reactive power from the HVDC
substation lead to small or
moderate ac voltage changes. Therefore the additional transient voltage control at
the busbar is not normally
required. The reactive power balance between the ac network and the HVDC substation
is achieved by switched
reactive-power elements.
In low and very low ESCR systems, the changes in the ac network or in the
HVDC transmission power
could lead to voltage oscillations and a need for special control strategies.
Dynamic reactive-power control at
the ac bus at or near the HVDC substation by some form of electronic reactive-power
controller, such as a static
var, compensator (SVC) or static synchronous compensator (STATCOM), may be
necessary (12). In earlier
times, dynamic reactive-power control was achieved with synchronous compensators.
Commutation Failure. When a converter bridge operates as an
inverter, as represented at the receiving
end of the dc link in Fig. 8, a valve turns off when its forward current commutates
to zero and the voltage
across the valve remains negative. The period for which the valve stays
negatively biased is the extinction
angle γ, the duration beyond which the valve then becomes forward-biased. Without a
firing pulse, the valve
ideally stays nonconductive or blocked, even though it experiences a forward bias.
All dc valves require removing the internally stored charges
produced during the forward-conducting
period (defined by period α + µ at the inverter in Fig. 8) before the
valve can successfully block a forward
bias. Therefore the dc inverter requires a minimum period of negative
bias or minimum extinction angle γ
for successful forward blocking. If forward blocking fails and conduction
is initiated without a firing pulse,
commutation failure occurs. This also results in an immediate failure to maintain
current in the succeeding
converter arm as the dc line current returns to the valve which was previously
conducting and has failed to
sustain forward blocking (13).
Commutation failure at a converter bridge operating as an inverter
is caused by any of the following
conditions:
(1) When the dc current entering the inverter experiences an increase in magnitude
which causes the overlap
angle µ to increase, the extinction angle γ is reduced and may
reach the point where the valve cannot
maintain forward blocking. Increasing the inductance of the dc current path
through the converter by the
dc smoothing reactor and commutating reactance reduces the rate of change of dc
current. This has the
greatest effect on commutation failure onset.
(2) When the magnitude of the ac side voltage in one or more phases
reduces or is distorted causing an
inadequate extinction angle as commutation is attempted.
(3) A phase angle shift in the ac commutating voltage can cause commutation
failure. However, the magnitude
of ac voltage reduction and not the corresponding phase shift is the most
dominant factor determining the
onset of commutation failures for single-phase faults.
(4) The value of the predisturbance steady-state extinction angle γ also affects
the sensitivity of the inverter
◦
◦ ◦
to commutation failure. A value of γ = 18 is usual for most
inverters. Increasing γ to values of 25 , 30 ,
or higher reduces the possibility of commutation failure (at the expense of
increasing the reactive-power
demand of the inverter).
(5) The value of valve current before commutation failure also affects the
conditions at which a commutation
failure may occur. A commutation failure may more readily happen if the
predisturbance current is at full
load compared to light load current operation.
HVDC transmission systems must transport very large amounts of electric power which
is accomplished only
under tightly controlled conditions. Dc current and voltage are precisely
controlled to effect the desired power
transfer. Therefore it is necessary to measure system quantities continuously and
precisely. These quantities
at each converter bridge include the dc current, its dc side voltage, the delay
angle α, and for an inverter, its
extinction angle γ.
Two-terminal dc transmission systems are more usual, and they have in common
a preferred mode of
control during normal operation. Under steady-state conditions, the inverter is
assigned the task of controlling
the dc voltage. It may do this by maintaining a constant extinction angle γ which
causes the dc voltage Ud
to droop with increasing dc current I d , as shown in the minimum constant
extinction angle γ characteristic
A-B-C-D in Fig. 10. The weaker the ac system at the inverter, the steeper the
droop.
Alternatively, the inverter may normally operate in a dc voltage-controlling
mode which is the constant
Ud characteristic B-H-E in Fig. 10. This means that the extinction angle γ must
increase beyond its minimum
◦
setting, depicted in Fig. 10 as 18 .
If the inverter operates at a minimum constant γ or constant Ud
characteristic, then the rectifier must
◦
control the dc current I d . It can do this so long as the delay angle α is not at
its minimum limit (usually 5 ). The
steady-state constant current characteristic of the rectifier is shown in Fig. 10 as
the vertical section Q-C-H-R.
The operating point of the HDVC system is where the rectifier and inverter
characteristic intersect, either at
points C or H.
The operating point is reached by the action of the on-line tap changers of
the converter transformers. The
inverter must establish the dc voltage Ud by adjusting its on-line tap changer to
achieve the desired operating
level if it is in constant minimum γ control. If in constant Ud
control, the on-line tap changer must adjust its
The on-line tap changers on the converter transformers of the rectifier are
controlled to adjust their tap
settings so that the delay angle α has a working range between approximately 10◦
and 15◦ to maintain the
constant current setting I order (see Fig. 10). If the inverter is operating in
constant dc voltage control at the
operating point H and if the dc current order I order is increased so that
the operating point H moves toward
and beyond point B, the inverter mode of control reverts to constant extinction
angle γ control and operates on
characteristic A-B. Dc voltage Ud is less than the desired value, and so the
converter transformer on-line tap
changer at the inverter boosts its dc side voltage until dc voltage control is
resumed.
Not all HVDC transmission controls have a constant dc voltage control, such
as depicted by the horizontal
characteristic B-H-E in Fig. 10. Instead, the constant extinction angle γ control
of characteristic A-B-C-D and
the tap changer provide the dc voltage control.
Current Margin. The dc current order I order is sent to both
the rectifier and inverter. It is usual to
subtract a small value of current order from the I order sent to the inverter.
This is known as the current margin
Imargin and is depicted in Fig. 10. The inverter also has a current controller, and
it attempts to control the dc
current I d to the value I order −Imargin , but the current controller at the
rectifier normally overrides it to maintain
the dc current at I order . This discrepancy is resolved at the inverter
in normal steady-state operation as its
current controller cannot keep the dc current at the desired value of I order
−Imargin and is forced out of action.
The current control at the inverter becomes active only when the current control at
the rectifier ceases because
its delay angle α is pegged against its minimum delay angle limit. This is readily
observed in the operating
characteristics of Fig. 10 where the minimum delay angle limit at the rectifier is
characteristic P-Q. If for some
reason or other, such as a low ac commutating voltage at the rectifier end, the P-Q
characteristic falls below
points D or E, the operating point shifts from point H to somewhere on the vertical
characteristic D-E-F where it
is intersected by the lowered P-Q characteristic. The inverter reverts to current
control, controls the dc current
I d at the value I order −Imargin , and the rectifier effectively controls dc
voltage as long as it is operating at its
minimum delay angle characteristic P-Q. The controls can be designed so that the
transition from controlling
the current by the rectifier to controlling the current by the inverter is automatic
and smooth.
Voltage-Dependent Current-Order Limit (VDCOL). During
disturbances when the ac voltage at
the rectifier or inverter is depressed, it is not helpful to a weak ac system if the
HVDC transmission system
attempts to maintain full-load current. A sag in ac voltage at either end also
results in a reduced dc voltage.
The dc control characteristics shown in Fig. 10 indicate that the dc current order
is reduced if the dc voltage is
lowered. This can be observed in the rectifier characteristic R-S-T and in the
inverter characteristic F-G in Fig.
10. The controller which reduces the maximum current order is known as a voltage-
dependent current order
limit or VDCOL (sometimes called a VDCL). If invoked by an ac system disturbance,
the VDCOL control keeps
the dc current I d to the lowered limit during recovery which aids the
corresponding recovery of the dc system.
Only when the dc voltage Ud has recovered sufficiently does the dc current return
to its original I order level.
Figure 11 is a schematic diagram showing how dc transmission system controls
are usually implemented.
(1) To limit dynamic and transient overvoltage to within permissible limits defined
by substation equipment
specifications and standards.
(6) The static synchronous compensator or STATCOM to use gate turn-off thyristors
in the configuration of the
voltage source converter bridge. This is the fastest responding voltage
controller available and may offer
limited capability for an increased short circuit ratio.
Because each ac system with its HVDC application is unique, the voltage
control method applied is subject
to study and design.
Special Purpose Controls. There are a number of special purpose
controllers which can be added to
HVDC controls to take advantage of the fast response of a dc link and help the
performance of the ac system.
Ac System Damping Controls. An ac system is subject to power
swings due to electromechanical
oscillations. A controller can be added to modulate the dc power order or
dc current order to add damping.
The frequency or voltage phase angle of the ac system is measured at one or both
ends of the dc link, and the
controller is designed to adjust the power of the dc link accordingly.
Ac System Frequency Control. A slow responding controller can also
adjust the power of the dc link to
help regulate power system frequency. If the rectifier and inverter are in
asynchronous power systems, the dc
controller can draw power from one system to the other to assist in frequency
stabilization of each.
Step Change Power Adjustment. A noncontinuous power adjustment
can be implemented to take
advantage of the ability of a HVDC transmission system to reduce or
increase power rapidly. If ac system
protection determines that a generator or ac transmission line is to be tripped, a
signal can be sent to the dc
controls to change its power or current order by an amount which compensates for
the loss. This feature is
useful in helping maintain ac system stability and easing the shock of a
disturbance over a wider area.
Ac Undervoltage Compensation. Some portions of an electric
power system are prone to ac voltage
collapse. If a HVDC transmission system is in such an area, a control can be
implemented which on detecting
the ac voltage drop and the rate at which it is dropping, can effect a fast power
or current order reduction of
the dc link. The reduction in power and reactive power can remove the undervoltage
stress on the ac system
and restore its voltage to normal.
Subsynchronous Oscillation Damping. A steam turbine and
electric generator can have mechanical
subsynchronous oscillation modes between the various turbine stages and the
generator. If such a generator
feeds into the rectifier of a dc link, supplementary control may be required on the
dc link to ensure that the
subsynchronous oscillation modes of concern are positively damped to limit
torsional stresses on the turbine
shaft.
a collector system of a wind farm where cable delivery and optimum and individual
speed control of the wind
turbines is desired for peak turbine efficiency.
The second immediate application for VSC converter bridges is in a back-to-
back configuration. The back-
to-back VSC link is the ultimate transmission and power flow controller. It can
control and reverse power flow
easily and control reactive power independently on each side. With a suitable
control system, it can control
power to enhance and preserve ac system synchronism and act as a rapid phase-angle
power-flow regulator
with a 360◦ range of control.
BIBLIOGRAPHY
DENNIS A. WOODFORD
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3023.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Hydraulic Turbines
● Email this page Enter words or
Standard Article
● Print this page phrases
Dan Levy1
1University of Limerick, Limerick, Ireland
Abstract
Hydro-Plants
Electrical System
Stand-Alone Generators ()
Protection
Cost
Conclusion
List of Symbols
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3023.htm17.06.2008 16:41:02
HYDRAULIC TURBINES
Hydro-Plants
Large hydroelectric installations run at almost constant hydraulic head and grid
synchronous speed. Opti-
mizing the power output under these conditions is relatively easy. Small
installations, such as “run of river
systems,” on the other hand are subject to much more variable hydraulic conditions,
and optimum power (or
efficiency) from such an installation is much more difficult to achieve.
In most large installations, flow of water through the turbine is varied using
a governor system to maintain
constant generator speed. This is achieved by (1) varying the guide vane angle in
Francis and other reaction
turbines, (2) adjustment of blade angle and possibly entry vane angle to Kaplan and
propeller turbines, and (3)
adjustment of capacity in the cross-flow turbine. In impulse turbines such as the
Pelton wheel a variable inlet
nozzle is used. With this control method at constant head, the delivered torque to
the generator is proportional
to the flow, and the turbine speed is that required for synchronous generation at
the particular grid frequency.
This type of installation is a constant speed, constant frequency (CSCF) system;
and ideally, optimum power
(measured in real time) corresponds to constant speed operation.
In smaller installations in which the hydraulic head is relatively low and
variable over a wide range,
normally it is not possible to obtain optimum power at constant speed; and, for
coupling to a grid, the variable
speed but constant generated frequency (VSCF) requirement has, in the past, made
such small sites nonviable
for mains electricity generation. Power produced in these sites has been suitable
only for local heating which
has a much lower commercial value than grid frequency electricity.
Today, however, small hydroelectric systems will become more financially
attractive by virtue of modern
developments in low-cost power converters (from 100 W upwards), special VSCF
generators, and cheap com-
puting units for on-line power measurement and optimizing control. All these will
lead to different approaches
to system design, and this article discusses some of the relevant techniques.
2 HYDRAULIC TURBINES
while the turbine flow depends on load, speed, vane setting, and a control variable,
um (which has in the past
been supplied by a mechanical governor):
Assuming no control action, the solution of Eq. (1) may or may not be stable
depending on the relative slopes
of the turbine and generator torque characteristics at the operating condition, and
this can be shown by either
of the following:
HYDRAULIC TURBINES 3
(1) Linearizing Eq. (1) for small disturbances, δw, around the mean operating
speed assuming all other pa-
rameters are constant, that is (20),
which corresponds to the point of intersection of the curves Tt and Tg (for the
case K = 0).
Under closed-loop control conditions where control actions um
and or ue are applied, system stability is
much more difficult to determine, particularly when several generating systems are
supplying a common grid
since Eq. (1) becomes a multivariable equation.
Optimal Operation. The optimal operating speed, wopt , for maximum
power (or maximum efficiency if
so desired) will vary with hydraulic conditions (e.g., head), and the control
system on the turbine or generator
(or both) should be designed to maintain this optimal operation. Such a
closed-loop control system must
4 HYDRAULIC TURBINES
continuously monitor the generator output power (or system efficiency) and use the
control variables ue and/or
um to adjust the speed so that maximum power (or efficiency) under the prevailing
conditions is obtained.
In older installations a mechanical governor was used to provide control
action, um , so that constant
(synchronous) speed was maintained. This governor could be replaced by a variable-
speed (or range-speed)
governor so that the speed set point corresponded to the optimal speed
as determined by the optimizing
computer. However, it is more satisfactory in modern installations to minimize the
mechanical adjustments
and to use the computer to provide control action ue on the generator or
electrical side, although care should
be taken to ensure that overspeed protection is still provided on the mechanical
side.
Electrical System
HYDRAULIC TURBINES 5
then the alternator will not be able to absorb all the turbine power, speed runaway
will occur, and the system
will be said to be out of synchronism.
Induction machines with small slips can also be considered as constant-speed
systems. (Note: Percent slip
is the difference between synchronous and actual speed. For a four-pole 50 Hz
induction machine, synchronous
speed is 1500 rev/min. Used as a motor at 1485 rev/min, the “positive slip” = (1500
− 1485)/1500 × 100 = 1%).
Used as a generator and run at 1515 rev/min, the “negative slip” is 1%). An
induction generator can operate
on an infinite bus at slip speeds 1% to 5% above synchronous speed, a small
variation compared with typical
shifts in turbine speed. Hence, induction systems, which maintain constant speed
due to electrical constraints
imposed by a grid irrespective of turbine speed, are sometimes also classified as
CSCF systems even though
they are in fact VSCF in a narrow range above synchronous speed. Slip
higher than 5% is not practical,
since the electrical losses increase substantially and the efficiency deteriorates.
In particular, this is true for
large generators. The torque–speed characteristic of an induction generator
is shown in Fig. 5. Similar to
the synchronous machine, if the turbine torque increases above the peak, Eq. (3)
will no longer be satisfied
and the system becomes unstable, speed runaway occurs, and the machine generates no
power. In induction
machines, this maximum torque is also called “runaway” or “pull-out” torque. The
induction generator can also
be classified as a variable-torque, constant-frequency machine since at synchronous
speed its torque is equal
to zero and increases with speed.
Since the characteristic in Fig. 5 is constant and cannot be varied for a
squirrel cage generator, the control
variable ue is constant. An important characteristic of an induction generator is
that it uses reactive power
usually drawn from the grid in order to enable it to generate real power. This
means that grid failure would
6 HYDRAULIC TURBINES
prevent generation unless the reactive power was supplied by a capacitor station
system. Since capacitors
always exist on the machine (from grid lines, for example), self-excitation is
still possible and there is a danger
of power being supplied from the induction machine in small hydro-stations.
Fatalities have occurred among
power-line workers due to lines being supplied from unknown sources, and protection
at the generator must
be provided to prevent this. Therefore, even in the induction generator case, there
is no automatic protection,
contrary to what many people may think. In the case of power loss, overspeed
protection of the turbine must
be provided.
If used on a wind station rather than a hydro-station, runaway would be more
acute and although a
speed controller is not necessary, some mechanism to dump excess wind power would
be needed. However,
in the wind station, a useful characteristic of the induction generator is that the
same machine runs as a
motor, drawing power from the grid, when below synchronous speed. This may be used
to advantage on wind
turbines which may not be self-starting and need to be motored up to self-
sustaining speeds. If the rotor of the
induction generator is a wire-wound type, then its torque–speed characteristic can
be controlled with variable
resistances in series with the rotor windings. The result will be a variable-
torque, variable-speed (narrow-
range) characteristic similar to the isosynchronous generator (see the
section entitled “The Double-Stator
(Cascade or Isosynchronous) Induction Machine”).
Although the technical merits of synchronous and induction generators have
been extensively argued in
the literature, they deserve at least a brief summary here. A synchronous generator
runs at constant speed
and hence requires costly speed controls if not run in conjunction with a grid.
However, it can supply negative
reactive power to the grid system, and, if necessary, it operates as a
synchronous condenser, whereas an
induction generator lacks this capability and also needs additional power factor
improvement equipment. On
the other hand, stability problems can arise where the synchronous generator of the
hydro-system is “thinly”
connected to other synchronous machines because of its location in the power
network (a large amount of
reactance exists between the main fossil-fuel generator and the hydro-power
generator). Any additional series
electrical impedance with the output of the machine will alter Te and reduce the
stability range of the operating
point. Induction machines have a distinct advantage in this matter.
It is worth mentioning that the instability may not be in speed (or
frequency) but in power supplied to
the grid, if long power lines with large series impedance are used.
Variable-Speed, Constant-Frequency (VSCF) Systems. In most
cases, variable-frequency power
must be converted to constant-frequency power. The conversion can be made with the
aid of rotary electrome-
chanical machinery or with solid-state converters using transistors and thyristors.
In some cases, nonconven-
tional generators produce constant frequency from a variable speed source. As a
result, generation schemes
involving variable-speed rotors are normally more complicated than constant-speed
systems.
Variable-speed, constant-frequency systems can be summarized as follows:
Ac Generator (Alternator). As shown in Fig. 6, the alternator is a
variable-output-frequency generator
when coupled to a varying-speed prime mover. Its field is derived from a permanent
magnet or dc excited coil.
The output of this machine is a variable-frequency, variable-voltage waveform. If
this voltage is not used for
heating or lighting, it may be processed in oder to be converted to a constant
frequency output. Therefore, the
generator output of varying frequency is rectified to give direct current, which is
converted to ac by an inverter.
The ac–dc–ac conversion is an established technology in high-voltage dc
transmission, and the operation of
inverter technology on the ac power grid is well understood within certain
parameters, including system short-
circuited and voltage stability at the point of interconnection. Application of
this scheme to hydropower is
limited only by the cost of power-processing equipment.
For this system, the control variable ue can be taken as equal to the
field current and may be used to
reduce the field losses at low output power. In small permanent magnet machines, the
field is constant and
therefore ue does not exist.
Field Modulation Techniques, Frequency Down Conversion. The
machine shown in Fig. 7 is basically
a three-phase wire-wound rotor slip rings machine excited by a single-phase grid
frequency (10). The output
HYDRAULIC TURBINES 7
Fig. 6. Alternator and frequency conversion system for producing constant output
frequency from a variable speed prime
mover.
Fig. 7. Field modulation system for producing constant output frequency from
a variable speed prime mover.
8 HYDRAULIC TURBINES
due to its commutator and rotational voltage only, at any speed (without
transformer voltage), since feedback
compensation for its armature voltage is used. Therefore, the characteristics of
this machine (mechanical and
electrical) are linear and very similar to dc machines. The output current of the
machine is not affected by the
rotor inductance due to compensation which is practically equal to the difference
between grid voltage and the
generated voltage divided by the total resistance in the circuit. In ACCG there is
another parameter which can
modify the characteristic of the machine other than the excitation voltage, and
this is the phase between the
stator and grid voltages. If this phase is adjusted (for example, by adjusting the
position of the commutator
relative to the stator), then the torque magnitude and direction can be controlled.
The machine can work both
as a generator and a motor.
For this machine, the control variable ue can be regarded as a vector—that is,
field voltage and field phase.
The disadvantage is of course the commutator. The problem of reliable
commutation in ac machines is
serious. Several proven techniques are available to improve commutation, including
high-resistance brushes or
HYDRAULIC TURBINES 9
Fig. 9. Squirrel cage induction machine with a four-quadrant power converter used
for producing constant output fre-
quency from a variable-speed prime mover.
Fig. 10. Double-output machine with slip power recovery with the aid of
a frequency converter. The figure shows a
two-quadrant converter used for negative slip (supersynchronous) power generation.
ue = # = constant is the triggering
angle.
commutators with many segments. The latter techniques have already been used to
build large ac commutation
motors successfully. These additional features make the machine costly and
according to the best of this author’s
knowledge, no commutator generators are available for sale today on the market.
Squirrel Cage Induction Machine with a Converter. A VSCF
system can be designed from a squirrel
cage induction machine and a converter acting as a frequency changer. In Fig. 9
here the induction generator
works as a CSCF generator in small slip; however, its self-synchronous speed is
dictated and can be adjusted
by the converter.
Although a reliable squirrel cage machine is used, the disadvantage of the
system is that the converter
is a four-quadrant type (i.e., able to drive the machine as a motor as well as a
generator—in other words, able
to absorb and feed power in both directions). Also, the converter should accept
variable frequency at its input
and produce constant frequency (grid frequency) at its output. All solid-state
converters produce undesirable
frequencies, which may be unacceptable if they cannot be filtered properly before
the output is fed to the grid
system (16).
The reactive power in this case is dictated by the converter characteristic
but not by the machine itself. The
control variables in this case are the converter output voltage and frequency.
Therefore ue can be regarded as
a two-variable vector which modifies the characteristic in Fig. 5. This type of
converter is relatively expensive
today and is produced by few companies. An alternative solution is the double-
output induction generator
system.
The Double-Output Induction Generator (DOIG) System (4,5,6,7,8,9).
Figure 10 illustrates the ma-
chine used as a generator is a wire-wound-rotor slip-rings induction machine (or a
cascade induction machine
if a contactless machine is desired without slip rings).
10 HYDRAULIC TURBINES
HYDRAULIC TURBINES 11
connected in cascade. One stator winding is fixed and the other can be manually or
automatically adjusted to
match the mechanical power to the electrical power when used as a generator or as a
motor. The machine itself
is self-protecting, and no current limiting device is needed. It can be coupled
directly to the power grid and to
the prime mover without an interface network or a governor.
In application as a generator, the machine is connected directly to the grid
and to the prime mover through
an appropriate gear ratio, which is designed according to the optimal output of
electrical power to the grid,
which corresponds to an almost constant speed (variable speed in narrow range).
This speed is equivalent to
the ordinary squirrel cage machine optimal speed in the generator mode (for
negative slip) and is independent
of the angles between the two stators. However, the torque is a function of the
angle between the stators, and
it is zero for zero and maximum for an electrical angle of 180 degrees. Therefore,
in this case ue = α, the angle
between the two stators.
This machine is equivalent to an ordinary induction machine with an ideal
variable transformer turn
ratio (Variac) introduced between the machine and grid (ue = turns ratio).
Simulation of the machine can be
made easily in the laboratory using a Variac and an ordinary induction machine. In
practice, this arrangement
is not practical. As stated before, it is also equivalent to a wire-wound-rotor
induction machine controlled by
variable resistors at its rotor ports.
As the turbine power changes due to changes in the flow rate, the generator
angle needs to be varied in
such a way that the power fed to the grid is maximum. This will correspond to
optimal power output from
the turbine. The adjustable stator is normally geared to a single-phase ac
servomotor which permits manual
adjustment as well as automatic closed-loop adjustment with the aid of a computer.
The two machine systems
can also be designed as a single-unit machine with a special-shaped squirrel cage
rotor, but with two, or a
single winding of two, different poles pairs in order to avoid the coupling between
the two ports (4). In an
asynchronous mode, one stator is connected to the grid while the other is connected
to variable resistances
which control its characteristic, as in the case with a wire-wound-rotor induction
machine.
An example is the 15 kVA hydro-plant which incorporates a
double-stator induction generator
(Fig. 12). This plant was commissioned in 1985 by the author in an abandoned water
mill site (Liffey Mills),
Nenagh, County Tipperary, Ireland, as a demonstration small-scale hydro-plant.
Found on the site was an
old nonadjustable gate, “horizontal Francis turbine” in relatively good condition,
and therefore it was used for
this project without any civil work investment. All the required safety devices
were installed and approved
by the power authority. The feature of the installation is as follows: As the
turbine power changes due to
changes in the flow rate, the generator angle (angle between the two stators—a
control variable) is varied by
a microprocessor in such a way that the power fed to the grid is maximum. This will
correspond to optimal
output power operation of the Francis turbine used to drive the generator. The
adjustable stator is geared to a
single-phase ac servomotor which permits manual adjustment as well as automatic
adjustment with the aid of
a microprocessor control circuit. The microprocessor measures the output power from
time to time and adjusts
the servomotor for maximum output power. For the rest of the time, the
microprocessor is used for monitoring
purposes such as in the case of electrical or mechanical power failure, and it will
adjust the angle between the
stators to zero for protection. This cannot be done immediately due to power
failure, but it is done after the
power is again restored and before the machines are connected back to the grid. In
this manner, minimum grid
current is assured which is equal to the magnetization current of the iron only
before the rotation is again
restored. When this happens, the microprocessor adjusts smoothly the angle between
the stators and increases
its speed as it works as a motor to drive the turbine up to the synchronous speed
where the latter in turn takes
over for the negative slip. The torque–speed characteristic of the generator is
shown in Fig. 13, the mechanical
power is shown in Fig. 14, and the electrical power is fed to the grid in Fig. 15.
The effect of adjustment of the angles between the two stators is clearly
indicated in Fig. 15. This effect
is similar to the adjustment of variable resistances in the rotor ports of a wire-
wound-rotor induction machine.
It is worth mentioning that this site cannot justify hydro-installations
other than the present one, due to
uncontrollable turbine, water level fluctuation, and relatively low output power.
12 HYDRAULIC TURBINES
HYDRAULIC TURBINES 13
excitation coil and the output winding is practically equal to zero in order to
prevent direct power feeding from
the exciter to the output windings. In this way, the two coils are practically
decoupled from each other.
A reference frequency (or line frequency) is applied to the excitation coil
and magnetizes a pole on the
rotor as it turns. This is called “writing” a pole. The pole then induces a voltage
in the stator windings at the
same frequency. The output frequency then has the same value as the input
frequency, independent of shaft
speed at steady state (zero acceleration). As rotor speed decreases, the length of
the poles shorten, so more of
them are written around the circumference of the rotor. As rotor speed increases,
the circumferential length of
the poles increases and fewer of them are written around the circumference of the
rotor. On one extreme, there
will be an even-pole synchronous speed where an even number of equal-length poles
are spaced uniformly
around the rotor. At the other extreme, there will be an odd-pole synchronous
speed, where an odd number of
equal-length poles are equally spaced around the rotor. Between these extremes
there will be fractional poles in
the vicinity of the exciter head as poles are being partially rewritten. At the
even-pole synchronous speed, the
poles remain in the same position from one revolution to the next, so no rewriting
of poles actually takes place.
There will be no rotor hysteresis loss in this case, since the rotor iron
magnetization does not change with time.
At the odd-pole synchronous speed, however, every positive pole is being exactly
replaced with a negative pole
during each revolution, so rotor hysteresis losses will be maximum at this speed.
This loss can be minimized
with the proper exciter shape design. For stand-alone power generation, the output
voltage can be controlled by
regulating the exciter head current. For this case uε = exciter current. Figure 17
shows the open circuit output
voltage as a function of speed at constant exciter current. The stator in this case
was compensated at lower
14 HYDRAULIC TURBINES
Fig. 14. Mechanical power–speed characteristic of the cascade machine. The figure
is useful for determining the mechan-
ical output power from the turbine.
For any generator, irrespective of whether it is CSCF or VSCF, the output power per
unit volume of the machine
is proportional to the square of the flux density and the machine speed. This flux
density is constant for a given
lamination material. Therefore, in order to reduce the machine size, the operating
speed should be as high as
possible or even above synchronous speed. For that reason, a gear ratio is always
needed for low-speed turbines
even if in principle a machine can produce power at low speed. The nominal
operating speed of the generator
should therefore be a compromise between the turbine speed and generator size. A
gear ratio of 1:7 is normally
acceptable in small installations and can be used as a guideline. For example, when
coupling a 150 rev/min
Francis turbine to a three-pole-pair induction generator at 5% slip, the generator
will run at 1050 rev/min.
HYDRAULIC TURBINES 15
Fig. 15. Electrical power–speed characteristic of the cascade machine. The figure
is useful for determining the optimal
electrical power output from the generator.
16 HYDRAULIC TURBINES
Fig. 16. Programmed pole (Roesel) machine. The torque–speed characteristic can be
varied by adjustment of the electrical
angle ue = µ0 between the writing head (exciter) current and the grid (output)
voltage in parallel grid connection. At stand-
alone operation, the machine produces constant output frequency from a variable
speed prime mover.
Protection
Fig. 17. Open circuit output voltage of the programmed pole generator compensated
at lower speed. The frequency of this
voltage is constant and equal to the exciter head frequency. The voltage shown is
for constant exciter current, which can
be regulated by variation of the exciter head current.
(5) Over and under speed protection (trip for faults on hydro-feeder)
(6) Fuses
(7) Reverse power relay
(1) Devices which provide a dedicated synchronizing and fault interrupting circuit
breaker for each unit
(2) A synchro-check relay or autosynchronizing on each synchronous unit
18 HYDRAULIC TURBINES
For induction machines operating in parallel with the grid, voltage and
frequency protection are still
necessary since in case of grid failure the induction generator may resonate
(become self excited) with the grid
impedance and produce different voltages at different frequencies (possibly also
harmonics).
Cost
Conclusion
Many run-of-river sites which were considered nonrelevant in the past can be
considered relevant today due
to availability of new types of generators and solid-state power processing
systems. These allow the plant to
operate from a low-variable head at variable speed in order to produce constant
frequency output at optimal
power efficiency. Constant speed and constant frequency optimization are used by
large hydro-plants which
are able to maintain a constant head by switching turbines or by adjusting of the
flow entrance to the turbine.
In the latter, the load on the turbine is varied according to the flow in order to
keep the speed constant.
In plants using closed-loop installation, a sensor may be used to measure
the output power and adjust an
independent control variable in the electrical system in such a way that this power
is always maximum.
There are minimum safety procedures required by law on the operation of a
hydro-plant—in particular,
when it is connected to the grid.
List of Symbols
E Voltage (V)
f Grid frequency (Hz)
H Head (m)
I , I (E , Z , f , w, ue ) Current (A)
J Inertia of the rotating parts (kg ·
m2 )
HYDRAULIC TURBINES 19
BIBLIOGRAPHY
20 HYDRAULIC TURBINES
DAN LEVY
University of Limerick
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3005.htm
} { { } }
HOME
●
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Hydroelectric Generators
● Email this page Enter words or
Standard Article
Abstract
Hydraulic Turbines
Hydroelectric Generators
Other Considerations
Acknowledgments
About Wiley InterScience | About Wiley | Privacy | Terms & Conditions
Copyright © 1999-2008John Wiley & Sons, Inc. All Rights Reserved.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3005.htm17.06.2008 16:41:23
HYDROELECTRIC GENERATORS
Hydraulic generating units for electric power generation normally run at a fixed
speed to provide an electrical
frequency of 50 or 60 Hz for conventional system needs. The turbines operate most
efficiently at only one oper-
ating point, which is the best combination of head, speed, and discharge. Any
deviation from these conditions
can cause a marked reduction in hydraulic efficiency because of flow separation,
secondary flows, turbulence,
and friction, all of which in turn, can cause pressure and shaft torque fluctuations
and cavitation.
High voltage direct current (HVDC) transmission of power for electric
utilities was introduced some 40
years ago. Electrical system planners, however, over the last few years
have reexamined its application to
remote hydro sites to reduce costs associated with the expensive sending-end
conversion equipment. Thus an
electrical arrangement known as the “unit connection” has been developed. This
innovation reduces capital
and operating costs considerably and also allows conventional generating
units to operate at an adjustable
speed/frequency, generally in the range of ±10%, with no increase in turbine cost
and only a modest increase
in generator and converter transformer costs.
When there is no necessity for fixed frequency at the generating station,
hydraulic turbines can operate
without the previous restriction of constant speed, thereby improving
hydraulic characteristics throughout
the normal range of operating conditions. Depending on river hydraulic regimes in
relation to plant size, this
opportunity could represent a large revenue improvement. Then adjustable
speed, in this context, depends
wholly on the conditions that the transmission system is HVDC and that
there is no requirement for fixed
frequency, such as an interconnection with another ac system.
When high voltage ac and dc systems are compared at the planning stage,
recognition should be given
to the additional economic benefit that could result from adjustable speed
operation. Utilities in Canada and
Brazil are exploring the possibility of profiting from the benefits of the HVDC unit
connection with adjustable
speed operation for future hydroelectric developments.
This article describes the basic concept of HVDC unit-connection
schemes leading to the efficiency of
adjustable speed turbine operation (sometimes called “variable speed” in referenced
literature). Then it moves
into a detailed analysis of the effects of HVDC and adjustable speed/frequency on
the hydroelectric generator
that is the basis of the article. Finally, the article continues with a brief
discussion of other related considerations
involved with this method of hydroelectric generation and transmission.
2 HYDROELECTRIC GENERATORS
HYDROELECTRIC GENERATORS 3
4 HYDROELECTRIC GENERATORS
Hydraulic Turbines
HYDROELECTRIC GENERATORS 5
Fig. 3. Normalized hill curves. AA# defines the fixed-speed operating range
for a given head as normalized flow (and
power) change. Copyright International Water Power and Dam Construction. Used with
permission.
Increased energy generation is derived directly from the hill chart curves in
Fig. 3. The vertical line AA#
defines a certain fixed-speed operation range, as flow and output power vary, and Pf
an operating point for that
speed. Adjusting the operating speed to move from point Pf to
operating point Pa increases the power output
in this case by 1% at no cost. The increased efficiency coming from a less turbulent
water flow also results
in inherently quieter operation and makes the unit less prone to vibrations and/or
cavitation. If operation is
constantly optimized for efficiency in this way, net energy gains are obtained (see
Ref. 1, Chap. 3).
In addition to increased efficiency, environmental benefits, such as improved
survival rates for fish passing
through the turbines, are also achieved. Furthermore, a new approach to
reservoir planning and design is
possible if energy gains are traded for flooded areas (see Ref. 1, Chap. 3).
Figure 4 shows normalized turbine performance curves corresponding to an
actual Francis runner design
in the 700 MW range at a net head of 80 m. For adjustable speed operation at rated
head, the turbine efficiency
at part load increases substantially over fixed-speed operation. The peak efficiency
changes only slightly, and
the efficiency for rated conditions is identical. Of course, with a large number of
units it is possible to operate at
near peak efficiency over a large power range by simply scheduling the number of
units in operation. Increased
efficiency with adjustable speed at rated head and part load, in this example, would
be most beneficial for a
station with a small number of large units.
From Fig. 4 it can also be seen that adjustable speed operation at low head
allows the turbine efficiency
at part load and the available peak power to increase quite dramatically compared
to fixed-speed operation.
In addition, the turbine output and efficiency near full load are
virtually identical for both maximum and
minimum head. This efficiency increase is obtained only with adjustable speed. For
fixed-speed operation at
minimum head and part load, draft tube pressure oscillations and
resulting shaft torque fluctuations are
possible. This zone is avoided by adjustable speed operation.
The Limestone generating station on the Nelson River in Canada has 10
propeller turbines rated at 125.4
MW at a rated head of 27.6 m and a speed of 90 rpm. The turbines operate at
constant speed and supply power
to a conventional HVDC system. In view of future developments on the same river,
adjustable speed operation
has been studied in a hypothetical unit-connected system. Figure 5 shows the
normalized performance curves
6 HYDROELECTRIC GENERATORS
for this plant with maximum speed variations of 5%, 10%, and 15%. The performance
curve of a Kaplan turbine
is superimposed for reference. The efficiency range for the Kaplan turbine is larger
than the adjustable speed
propeller turbine with a 15% speed range, but its capital and operating costs can
be on the order of 30% greater.
HYDROELECTRIC GENERATORS 7
Hydroelectric Generators
8 HYDROELECTRIC GENERATORS
General Assessment.
Harmonic Electric Loading. Conventional generators feeding ac systems
are subjected to negative se-
quence current loading. The negative sequence current creates low-order harmonics
as a result of rotor saliency.
Hence, generators are designed to have additional thermal capacity to sustain 8% to
12% of negative sequence
current loading. Generators feeding only HVDC converters in the absence of dc line
current second harmonics
(the normal steady-state operating condition) see virtually no negative sequence
loading. Thus the additional
rating can be used to compensate for the 12-pulse harmonic electrical loading (6).
Torque Harmonics. The presence of pulse-number-related frequencies in
the rotor (i.e., multiples of 600
Hz for a 50 Hz fundamental frequency) will cause a small electromechanical torque
of a few percent. Although
high in the context of large machines, such frequencies might excite natural modes
of resonance either in small
components or, because of nonlinearities, might induce subharmonic vibrations in
larger components.
Adjustable Speed Operation of Hydro Generators. At a higher
speed n (rpm), machine operation will
take place at a correspondingly higher frequency f (Hz). Stray losses
due to windage, mechanical friction,
and harmonics increase. However, there is only a minor influence on the overall
electrical efficiency for the
small increases in speed required to optimize energy conversion in hydro turbines.
The case of downward speed
changes requires more attention as it may impose some additional capacity on
excitation circuitry if full voltage
is required at minimum speed.
Basic Design Frequency and Speed. The present day switching speed
of high power electronic devices,
especially the parameters associated with the current extinction process, allows
HVDC converters to operate
at up to about 120 Hz.
Vibration and Noise. The full pole-pitch, air-gap flux components
caused by current harmonics can
interact with each other and with other components of air-gap flux in a
rather complicated way, causing
vibration forces and contributing to increased noise level.
Generator Power Factor. The absence of ac filters leads to
unit-connected generator operation with
very high fundamental frequency power factor values, as seen from its terminals,
typically 0.9 to 0.94. In the
limit, for a hypothetical transformerless unit connection, the generator terminal
fundamental frequency power
factor is reduced, as a first approximation, to the cosine of the firing angle “α.”
This is so, because the ideal
commutation voltage moves inside the machine. Given the comparatively high
frequency of commutations
(12 per cycle), the voltage behind subtransient reactance, with good approximation,
is the new commutation
voltage (see Ref. 1, Chap. 4). In practice, internalizing the commutation
voltage relieves the machine from
about half the total converter reactive power requirement that it would otherwise
have to provide.
Fundamental Frequency-Generator/Converter Interaction.
The commutation process is a short
circuit between two phases of the generator across the transformer
leakage reactance xt that lasts for only
about 20 electrical degrees. The commutation reactance xc for a unit-connection
arrangement is given by Eq. (1).
x## is the actual operating point subtransient reactance of the generator, xt the
transformer leakage reactance,
The commutation has two main effects on the operation of a unit connection. One is
detrimental in that it
creates an ohmic drop of the dc voltage in which the commutation reactance plays
the part of a nondissipative
resistance. The greater xc , the greater the loss of voltage:
HYDROELECTRIC GENERATORS 9
where
#Ud = dc voltage drop in volts
Ud = dc voltage in volts
IG = generator current in amps
UG = generator voltage in volts
For a given firing angle, the angle of commutation and the voltage drop of the dc
voltage are constant:
The per unit voltage drop increase with the square of the frequency is of little
concern at higher speeds because
generated voltage can easily be increased through modest flux adjustments in such a
situation. At low speeds,
the per unit voltage drop reduction at a quadratic rate is beneficial in that less
overflux is required to keep the
generated voltage at rated levels.
Typical Values of Commutation Reactances for 12-Pulse Unit Connection.
The subtransient
reactance seen by the commutation circuit during operation of a unit connection is
a nonlinear function of direct
and quadrature axis reactances, dependent on the actual air-gap flux magnitude
and spatial orientation at
the specific operating conditions. For planning and station design, the subtransient
reactance x## of hydro units
can be evaluated from the negative sequence reactance x2 . For a generator equipped
with a continuous damper
## ##
winding, the values of the reactances x d and x q are very close (within 10%),
and the following approximation
is suggested:
In practice, the value of the transformer reactance based on the transformer rating
is about half that of the
subtransient reactance, and so with interpolar damper connections the commutation
reactance approximates
10 HYDROELECTRIC GENERATORS
angles become practically zero, and the commutation angles increase substantially.
In this case, either lower
subtransient and transformer reactances are selected, or operation with forced
retard (i.e., with reduced dc
voltage and a communication angle equal to 30% at full load) is accepted.
Damper Winding. Interpolar damper connections are sometimes
avoided to improve the mechanical
##
reliability of salient pole rotors. However, for a machine equipped with a
noncontinuous damper, x q increases
##
and reaches a value of about 1.5x d . This causes an increase in x2
of about 25%. Therefore, for a machine
constructed with noncontinuous damper connections,
Suppression of the interpolar damper links increases the value of the commutation
reactance xc by about 12.5%.
This also increases subtransient saliency and may allow a higher proportion of
stator harmonics to penetrate
the rotor circuits, resulting in increased losses, vibration, and noise. Therefore
full cage damper circuits are
highly recommended for HVDC unit-connected systems.
Reactance Variation Due To Machine Load.
All reactances in a generator vary with the machine
state of saturation. The corresponding effect on xc , however, is about 1%, which
is negligible.
The variation of frequency does not affect the reluctance of the
magnetic circuit significantly over the
considered range. Therefore, xc is practically constant in the
working range for constant U/f . However, if full
voltage is required at minimum speed, that is, U/f is
increased because of overexcitation, saturation may
increase, and the subtransient and commutation reactances are reduced.
Influence of Damper Resistance. In a generator
connected to a converter, the effectiveness of the
dampers is more important against the stator current harmonics. The skin effect
increases the resistance of
the dampers by about 30% within the range of 45 Hz to 75 Hz. However, their
resistance is less than 1% of the
##
##
value of the reactance x d , and so the increase is equivalent to about 0.3% of x
d . Thus damper resistance can
be disregarded in the context of commutation analysis.
HYDROELECTRIC GENERATORS 11
d
[Lq (s)] axis operational inductances plotted from the actual direct and quadrature
axis parameters, obtained
by standard testing of a salient-pole, 194.4 MVA, 13.8 kV hydro generator. The
horizontal axis refers to slip
frequency seen by the rotor (see Ref. 1, Chap. 4).
The diagrams can be derived from a larger number of time
constants, if the relevant test information
is available. This corresponds to a larger number of rotor circuits when deriving
the equivalent circuit and
to a correspondingly larger number of operational inductance poles and zeros. The
characteristic asymptotic
behavior consisting of ever lower “plateaus” remains, nevertheless, unaltered in
shape.
It is very relevant to note from this example that stator harmonics coupled
with rotor frequencies of only
about 50 Hz and up already see operational inductances that remain constant over
the speed ranges considered
for HVDC unit-connected stations. The lowest stator winding frequencies associated
with converter operation
induce slip frequencies of at least fundamental frequency. Therefore, it can indeed
be safely considered that all
machine reactances associated with the commutation process vary linearly with
speed.
Generator Excitation. The most commonly used and simplest system for
generator excitation is the
shunt arrangement. This system is well suited to operation under a constant U/f
(voltage/frequency). A field
flashing system from a battery can be used to provide power to start up
the excitation system. When the
12 HYDROELECTRIC GENERATORS
generator is required to operate with constant (rated) voltage over the entire
range of speed/frequency, it may
be necessary to oversize the excitation system to allow the generator to operate at
low frequencies.
Auxiliary Three-Phase Generator. Oversizing may be avoided by using
an auxiliary three-phase syn-
chronous generator coupled to the main unit shaft. This auxiliary generator,
provided with a shunt excitation
system and with a field flashing system, supplies the main generator excitation.
Because it operates with a
constant U/f , the excitation transformer does not need to be oversized.
Fully Independent Auxiliary Generators. In large, multiunit hydro
stations, it may be more convenient
to provide smaller independent hydro units to feed the excitation systems, station
auxiliaries, and local loads.
In this way all effects on auxiliaries caused by harmonics in the station supply
and by the frequency variations
are eliminated.
Cycling and Fatigue. Excessive cycling of machines over significantly
varied operating conditions may
lead to premature aging and even to failures. Designers usually consider the
influence of voltage, frequency,
and load variation. The last two have a direct bearing on operating temperature.
The amplitude of the cyclic
variations and its periodicity are the key variables for each parameter considered.
HYDROELECTRIC GENERATORS 13
Generator Losses.
Windage Losses. If the generator speed range is not too large, the unit
can be self-ventilated. Then
windage losses are proportional to the third power of speed. When the generator
cannot be self-ventilated at
low speeds, however, a forced cooling system may be required to support
ventilation.
Iron Losses. This is a common designation for a number of phenomena of
different natures in the iron
core. They can be split up roughly as “eddy current losses” and “hysteresis
losses.” For 0.5 mm laminations
in delivery condition and at 50 Hz, for instance, the eddy current losses measured
according to the Epstein
method are about 33% of the total losses. Consequently, the hysteresis losses
correspond, in this case, to the
remaining 67%.
Eddy Current Losses. This is a collective name for losses caused
by eddy currents within the lami-
nations themselves and by eddy currents circulating between laminations and some
other solid or structural
components of the core. Eddy current losses vary roughly in proportion to the
square of both induction and
frequency (≈B2 , ≈f 2 ).
where
PFe = the total iron losses
PFe0 = the total iron losses at rated frequency and nominal induction
f = the actual frequency under consideration
f 0 = the rated frequency
B = the actual induction under consideration
B0 = the nominal induction
kh = the hysteresis share of PFe0
This expression is a fair approximation for 0 <B <B0 when f varies between 25
Hz and 65 Hz and f 0 varies
between 40 Hz and 60 Hz. The numerical value for kh depends, among other
things, on machine design and
lamination treatment during the manufacturing process. The value is usually 0.5 for
0.5 mm steel lamination,
because it is typical that the eddy current loss share increases from the
previously mentioned value of 0.33 for
steel laminations before machining processes to about 0.5 after the machine is
completed. The iron losses due
to voltage harmonics are low and are neglected in the present context.
Stray Losses in Stator Copper. In the windings of ac machines, the
transverse flux in the slots creates
supplementary losses by skin effect in the copper strands in addition to joule
losses. When the generator is
connected to a converter station, the phase currents contain many harmonics that
cause supplementary losses.
These can be computed similarly to those corresponding to the fundamental frequency
and have significant
values up to the 25th. These losses are proportional to the square of the frequency
(f 2 ).
14 HYDROELECTRIC GENERATORS
where kH is a factor that allows for harmonic VA demand due to the rectifier load.
It varies with operating
parameters (firing and commutation angles, full or partial load) chosen to specify
the rating point but is only
a fraction of a percent for large 12-pulse unit connections. Reference 6 recommends
values of 1.01 for twelve-
pulse and 1.05 for six-pulse operation. Reference 1, Chap. 4, considers these
numbers typical for laminated
pole construction. Pd is the dc power of the rectifier load connected to the
generator, and φt1 is the fundamental
frequency power factor angle seen from the generator terminals. The converter
fundamental frequency power
factor angle should not be used here because now the commutation voltage is
internal to the generator. The
values to be used are approximated within a few percent by
where the subscript N indicates nominal (rated) values, UGN = rated generator
voltage and IGN = rated gener-
ator current. It has to be adapted for operation at adjustable speed by
where nN = generator rated speed. No consideration is given here to the fact that
iron and copper losses caused
by the skin effect are frequency-dependent, with the consequence that at n < nN a
higher induction, UG >n/nN
· UGN , is acceptable in a first approximation.
The generator rating for adjustable speed operation must be derived from the
most demanding conditions
over the range of operating points which are defined by the actual requirements for
the specific development.
The ideal requirements are reduced voltage at the lowest speed and maximum power
only at highest head and
HYDROELECTRIC GENERATORS 15
Fig. 8. Runaway and overspeed curves for hydro
generator sets.
speed. This may not be always the case, however, and so the actual operating
conditions must be investigated
to identify all limiting requirements that do not correspond necessarily to just
one operating point. The need
to absorb 12-pulse harmonic loading must be combined with the actual negative
sequence loading expected
from the small imperfections of the actual equipment, and the need for increased
excitation must be checked
against reduced losses at lower speeds.
Generator Voltage. It is generally required in ac and in dc systems
that generators maintain an almost
constant voltage over the full operating range. This requirement cannot be fulfilled
by a generator whose rated
power is required at maximum speed and is operated with constant flux. However, as
mentioned above, higher
induction can be accepted at lower speeds, and if the generator current is also
reduced in proportion to the
voltage increase, a design compromise becomes easier at the lower end of the speed
range, with no or only a
marginal overrating at base speed.
2
Generator Inertia Parameters (GD ). Generator inertia parameters are
important because, together
with the hydraulic circuit and turbine design, they define the overspeed transient
peak. Figure 8 is a diagram
showing the wicket gates closing after a load rejection in conditions of runaway
speed and overspeed. Runaway
occurs when the turbine is left with no load at maximum head and fully open wicket
gates. It is a steady-state
condition with zero efficiency, and speed is directly proportional to the square
root of the head. A total loss
of governor control leads to runaway. The time to achieve runaway speed increases
with the inertia. Design
standards require that hydro generating sets be built to withstand
runaway conditions without structural
damage.
Overspeed, on the other hand, is the peak velocity of a controlled machine
during a load rejection transient.
The slower the wicket gate closing action, the longer the machine accelerates
toward runaway, and the greater
the overspeed. The larger the generating set inertia, the smaller the acceleration,
and the smaller the overspeed.
It is reasonable to question the limits of overspeed that can be specified for
unit-connected generators. The
so-called “natural” inertia of the generator, which results from electrical and
mechanical design considerations
and is adequate for stability of speed regulation, should be checked first. After
that, load rejection simulations
and structural analysis of the generator and exciter mechanical components at the
corresponding overspeed
and runaway conditions must be performed to see if any limit is violated.
Eventually some small increase in
inertia may be required, but it is likely that a lighter and less costly, nearly
“natural inertia” design would be
acceptable.
----------------------- Page 292-----------------------
16 HYDROELECTRIC GENERATORS
New Zealand Field Tests. The New Zealand HVDC system was made
available for harmonic tests
during two periods of converter transformer maintenance in 1993 and 1995.
Under these conditions, the
sending-end HVDC poles and ac filters were disconnected from the ac
system. Operation was uneventful
as a “group connected” unit connection (4). Waveform distortion
measurements confirmed the theoretical
predictions, and the generator operating parameters, such as temperature,
vibration, and noise, remained
within normal limits.
Other Considerations
Overall, there are no technical difficulties that cannot be overcome for speed
regulation in a unit-connected
HVDC scheme operating within a speed range of approximately ±10%.
Modeling. When investigating new concepts, such as adjustable speed HVDC
unit-connection schemes,
it is essential to model the complete system to evaluate the consequences of system
effects on the generator
and the effect of the generator on the system. This involves developing
modeling techniques to simulate
the generator, the hydraulic system, and the electrical system to which
the generator is connected. Thus,
consideration must be given to the level of modeling required if
meaningful results are to be obtained. For
HYDROELECTRIC GENERATORS 17
18 HYDROELECTRIC GENERATORS
The application of diode-equipped rectifiers requires further assessing the
need for dc circuit breakers in
the transmission line (see Ref. 1, Chap. 6).
Acknowledgments
´
speed concepts, Conference Internationale Des Grand Reseaux Electriques A Haute
Tension (CIGRE) Study
Committees 11 (Rotating Machines) and 14 (HVDC links and ac power electronic
equipment) created Joint
Working Group (JWG) 11/14.09 in 1989. The following are the JWG’s main objectives:
BIBLIOGRAPHY
1. CIGRE JWG 11/14.09, Guide for Preliminary Design and Specification of Hydro
Stations with HVDC Unit Connected
Generators, Technical Brochure, No. 116, 1997.
2. D. A. Green, Generating more efficiently with adjustable speed machines, Hydro
Rev. , 13: 18–27, April 1994.
3. L. Ingram, A practical design for an integrated HVDC unit-connected
hydroelectric generating station, IEEE Trans.
Power Deliver. , 3: 1615–1621, 1988.
4. S. J. Macdonald et al., Test results from Benmore HVDC station with the
generators operating in group connected mode,
IEEE Winter Power Meet. , 95 WM, New York: SPWRD, 1995, pp. 242–248.
5. C. Mubemba, K. J. Cornick, The analysis of fast-fronted switching surges in
machine windings-effect of dielectric losses,
Trans. South Afr. EE, 86 (2): 57–64, 1995.
6. P. C. S. Krishnayya, Block and double block connections for HVDC power station
infeed, IEEE PES Summer Power Meet.
and EHV/UHV Conf., Vancouver, Canada, 1973.
READING LIST
HYDROELECTRIC GENERATORS 19
CIGRE JWG 11/14.09, Study of the application of HVDC variable speed unit connected
generators to hydro developments
in the Brazilian Amazon region, CIGRE Int. Colloquium, HVDC and FACTS,
Wellington, NZ, 1993.
CIGRE JWG 11/14.09, Hydro generators reactances and power factor in HVDC unit
connection, Int. Colloquium HVDC
FACTS , Montreal, Quebec, Canada, 1995.
CIGRE JWG 11/14.09, Adjustable speed operation of hydro electric turbine
generators, Electra , 167: August, 1996.
EPRI Report RP3577-01, Application of adjustable speed machines in conventional and
pumped-storage hydro projects,
1995.
K. W. Kanngiesser, Unit connection of generator and converter to be integrated in
HVDC or HVAC energy transmission,
Int. Symp. HVDC Technol. , Rio de Janeiro, Brazil, 1983.
E. Kita et al., A 400 MW adjustable speed pumped storage system, Water Power and
Dam Construction, 43: 37–39, Nov.
1991.
L. Ingram et al., Adjustable speed hydroelectric turbine/generators, Water Power
and Dam Construction, 47: 40–42, July
1995.
G. Merouge, M. Denoncourt, Variable speed hydraulic generators, CIGRE Int.
Colloquium on HVDC Power Transmission,
Recife, Brazil, 1989.
M. Naidu, R. M. Mathur, Evaluation of unit connected variable speed hydropower
station for HVDC power transmission,
IEEE Trans. Power. Syst. , 4: 668–676, 1989.
R. D. Rangel, A. S. Pedroso, J. G. Campos-Barros, Dynamic performance of
salient pole unit-connected generators, Int.
Colloquium on HVDC Power Transmission, CIGRE, Recife, Brazil, 1989.
L. B. Reis, M. A. Saidel, Operation at adjustable speed: Prospectives of higher
energy efficiency and better environmental
integration for hydroelectric plants, Int. Symp. Electr. Power Eng. ,
Stockholm, Sweden, June, 1995.
H. Tanaka, An 82 MW Variable speed pumped-storage system, Water Power and Dam
Construction, 43: 25–26, Nov. 1991.
L. Terens, Schafer, Variable speed in hydro power generation utilizing static
frequency converters, WATERPOWER-An Int.
Conf. Hydropower, Nashville, TN, 1993.
J. G. CAMPOS-BARROS
LINDSAY INGRAM
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3025.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Hydrogen Energy Systems
● Email this page Enter words or
Standard Article
Abstract
Fossil Fuels
Gaseous Hydrogen
Liquid Hydrogen
Slush Hydrogen
Conclusions
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3025.htm17.06.2008 16:41:52
Fossil Fuels
After the invention of the steam engine in the 1860s, when advances brought about
by the Industrial Revolution
started to replace humans’ and beasts’ toil with nature’s energy sources, a bright
future seemed to be certain
for humankind. More and more of nature’s energy, initially in the form of wood and
coal and later as oil and
natural gas, was being harnessed for the benefit of humans. This resulted in mass
production of goods with a
corresponding reduction in prices, thus raising living standards. Communities asked
for factories, railroads,
highways, seaports, and airports. This meant more jobs, more income, more
goods, and more services. The
world’s standard of living was rising. When the Industrial Revolution started, the
annual gross world product
per capita was in some tens of dollars; today, it is about $6600 and rising
exponentially.
Fossil fuels, which fed this amazing economic growth, were the medicine to
cure deprivation, but it was
an untested medicine. As humans consumed more and more fossil fuels, two important
predicaments started
to emerge: (1) fossil fuels would be depleted in a foreseeable future, and (2)
fossil fuels and their combustion
products were causing global environmental problems.
Depletion. The demand for energy continues to rise for two main reasons:
(a) the continuing increase
in world population and (b) the growing demand by the developing countries in order
to improve their living
standards. At the present time, a large portion (about 70%) of the world energy
demand is met by fluid fossil
fuels (petroleum and natural gas) because of the availability of these resources
and their convenience of use.
However, it is expected that the world fluid fossil fuel production will soon peak
and thereafter begin to decrease
(1,2,3,4). Figure 1 shows estimates of the production rates of fossil fuels and the
world demand. It can be seen
that the fluid fossil fuel production worldwide will continue to rise for the next
15 years and then will start
to decrease. The coal production—because of obvious environmental reasons—
is expected to remain nearly
constant for the next decade and then start to decrease.
Meanwhile, as a result of the growing world population and the
desire of people to better their living
standards, the world demand for fluid fuels is rising (Fig. 1). It is expected that
the world population (which is
5.88 billion as of 1998 and rising at 1.5% per year) growth will slow down and
reach about 10 to 12 billion by the
end of the next century (5). Consequently, the world demand for fluid fuels will
slow down and reach around 1.6
× 1012 GJ per year (1 GJ of energy is approximately equal to the energy contained
in 3.4 gallons of petroleum).
There will be a growing gap, starting within the next 10 years, between the demand
and production of fluid
fuels.
Environmental Damage. The second predicament involving fossil
fuels is the environmental damage
being caused by their combustion products. Technologies for fossil fuel extraction,
transportation, processing,
and particularly their end use (combustion) have harmful effects on the
environment, which cause direct and
indirect negative effects on the economy. Excavation of coal devastates the land,
which has to be reclaimed and is
out of use for several years. During the extraction, transportation, and storage of
oil and gas, spills and leakages,
which cause water and air pollution, occur. Refining processes also have an
environmental impact. However,
most of the fossil fuel environmental impact occurs during end use. End use of all
fossil fuels is combustion,
irrespective of the final purpose (heating, electricity production, or motive power
for transportation). The main
constituents of fossil fuels are carbon and hydrogen in addition to some other
ingredients, which are originally
in the fuel (sulfur) or are added during refining (lead, alcohols). Combustion of
fossil fuels produces various
gases (CO , SO , NO , CH), soot and ash, droplets of tar, and other organic
compounds, which are all released
x x x
into the atmosphere and cause air pollution. Air pollution may be defined as the
presence of some gases and
particulates that are not a natural constituent of the atmosphere, or the presence
of natural constituents in
----------------------- Page 299-----------------------
Because of the environmental problems fossil fuels are causing, energy researchers
are looking at all possible
sources of energy to replace these fuels. There are quite a number of primary
energy sources, such as ther-
monuclear energy, nuclear breeders, solar energy, wind energy, hydropower,
geothermal energy, ocean currents,
tides, and waves, that are available. At the consumer end, about one-quarter of the
primary energy is used as
electricity and three-quarters is used as fuel. The primary energy sources already
mentioned must therefore
be converted to these energy carriers needed by the consumer. In contrast with
fossil fuels, none of the new
primary energy sources can directly be used as a fuel (e.g., for air and land
transportation). Consequently, they
must be used to manufacture a fuel or fuels as well as to generate electricity.
Comparison of Fuels. Because we need to manufacture a fuel for the
post fossil fuel era, we are in a
position to select the best possible fuel. There are many candidates, such as
synthetic gasoline, synthetic natural
gas (methane), methanol, ethanol, and hydrogen. The fuel of choice must satisfy the
following conditions (7):
where E is the energy generated by the fuel, M is the mass of the fuel, V is the
volume of the fuel, and the
subscript h refers to hydrogen. The higher the motivity factor, the better the fuel
for transportation. Table 3
lists the pertinent properties of some fuels, as well as the motivity factors
calculated using Eq. (1). It can be
seen that among the liquid fuels LH2 has the best motivity factor, whereas
methanol has the lowest motivity
factor. Among the gaseous fuels, GH2 has the best motivity factor.
Consideration of the utilization efficiency advantage of hydrogen further
improves hydrogen’s standing
as the best transportation fuel. Of course, this is one of the reasons why hydrogen
is the fuel of choice for the
space programs around the world, even though presently it is more expensive than
fossil fuels.
Versatility. At the user’s end, all fuels must be converted through a
process (such as combustion) to other
forms of energy (thermal, mechanical, or electrical energies). If a fuel
can be converted through more than
one process to various forms of energy at the user’s end, it becomes more
versatile and more convenient to
use. Table 4 lists various fuels and processes by which they can be converted to
other forms of energy at the
user’s end. It can be seen that most fossil fuels can be converted through one
process only, that of combustion.
In certain cases, natural gas could be used in catalytic combustion and
in fuel cells, in addition to flame
combustion. Hydrogen, however, can be converted to other forms of energy in five
different ways; in addition
to flame combustion, it can be converted directly to steam, converted to heat
through catalytic combustion,
be used as a heat source and/or heat sink through chemical reactions, and be
converted directly to electricity
through electrochemical processes (9). In other words, hydrogen is the most
versatile fuel.
Utilization Efficiency. In comparing the fuels, it is important to
take into account the utilization effi-
ciencies at the user’s end. For utilization by the user, fuels are
converted to various energy forms, such as
mechanical, electrical and thermal. Studies show that in almost every instance of
utilization, hydrogen can be
converted to the desired energy form more efficiently than other fuels. Table 5
presents the utilization efficiency
factors, defined as the fossil fuel utilization efficiency divided by the hydrogen
utilization efficiency for various
applications. It can be seen that hydrogen is the most efficient fuel. This results
in conservation of resources,
in addition to conserving energy.
Safety. The safety aspects of fuels involve their toxicity on the one hand
and the fire hazard properties
on the other. In addition to the toxicity of their combustion products, the fuels
themselves can be toxic.
Toxicity increases as the carbon-to-hydrogen ratio increases. Hydrogen and
its main combustion product,
water or water vapor, are not toxic. However, NOx , which can be produced through
the flame combustion of
hydrogen (as well as through the combustion of fossil fuels) displays toxic
effects.
The Best Fuel. When we look at the fuel options critically under the
criteria given earlier, it becomes
clear that hydrogen is the best transportation fuel, the most versatile
fuel, the most efficient fuel, and the
safest fuel. In summary, hydrogen is the best fuel.
Hydrogen Energy System. As a result of the aforementioned
observations, it becomes clear that it
would be expedient to manufacture hydrogen using any and all primary energy
sources, in order to make up
for their shortcomings. Such an energy system is then called the hydrogen energy
system.
Figure 2 presents a schematic diagram of the proposed hydrogen energy system.
In this system, hydrogen
(and oxygen) is produced in large industrial plants where the primary
energy source (solar, nuclear, and
even fossil) and water (H2 O), the raw material, are available. For large-scale
storage, hydrogen can be stored
underground in ex-mines, caverns, and/or aquifers. Hydrogen is then
transported, by means of pipelines or
super tankers, to energy consumption centers. It is then used in
electricity, transportation, and industrial,
residential, and commercial sectors as a fuel and/or an energy carrier. The by-
product is water or water vapor.
If flame combustion of hydrogen is used, then some NOx is also produced.
Water and water vapor are recycled
back, through rain, rivers, lakes and oceans, to make up for the water used in the
first place to manufacture
hydrogen.
The oxygen produced in the industrial plant making hydrogen could either be
released into the atmosphere
or be shipped or piped to industrial and city centers for use in fuel cells
(instead of air) for electricity generation.
This would have the advantage of increasing the utilization efficiency. The oxygen
could be used by industry
for nonenergy applications and also for rejuvenating the polluted rivers
and lakes, or speeding up sewage
treatment.
It should be noted that in the hydrogen energy system, hydrogen is not a
primary source of energy. It is
an intermediary or secondary form of energy or an energy carrier. Hydrogen
complements the primary energy
sources and presents them to the consumer in a convenient form at the desired
location and time.
Details of the hydrogen energy system, including production, storage,
transportation, distribution, uti-
lization, environmental impact, and economics, can be found in the proceedings of
the THEME and the World
Hydrogen Energy Conferences (10,11,12,13,14,15,16,17,18,19,20,21,22,23).
When one considers early 2000s, it can be expected that about half of the
thermal energy will be supplied
by natural gas, and the rest will be supplied by petroleum fuels (fuel
oil and residual oil) and coal. Coal is
assumed to be the main energy source for electricity generation, gasoline for
surface transportation, and jet
fuel for air transportation. This is of course a simplified version of the fossil
fuel energy system, but it is close
enough to the present patterns of energy consumption and can be used as a basis for
comparisons.
Coal/Synthetic Fossil Fuel System. Reserves of fossil fuels
are finite, particularly those of oil and
natural gas. Known reserves of oil and natural gas are about 8000 EJ (1 EJ = 1018
J), which would be enough for
the next 40 years at the current consumption rate (25). If the exponential
population growth and the demand
for growth are taken into account, they would only last about 25 years.
Even if the estimated additional
undiscovered resources were added, that would satisfy energy needs for fluid fuels
an additional 30 years or
so. Coal reserves are much larger, known reserves are about 20,000 EJ, but
estimated ultimately recoverable
resources add up to 150,000 EJ. These large amounts of coal could eventually be
used to produce synthetic
liquid fuels, allowing society to continue employing the present energy
system. Such a system is called the
coal/synthetic fossil fuel system because coal is to be used to manufacture
synthetic fossil fuels, as well as to
be directly used for electricity generation.
In this case, it can be assumed that the present fossil fuel system will be
continued by the substitution
with synthetic fuels derived from coal wherever convenient and/or necessary.
Patterns of energy consumption
are also assumed to be unchanged (see Fig. 4). Coal will be used extensively for
thermal as well as electric
power generation because it is much cheaper than synthetic fuels.
However, some end uses require fluid
fuels. Therefore, it has been assumed that synthetic natural gas (SNG) will be used
for some thermal energy
generation (primarily in the residential sector) and also as fuel for surface
transportation, where it will share
the market with synthetic gasoline. Synthetic jet fuel will be used in air
transportation.
Solar Hydrogen Energy System. If solar energy, in its direct
and/or indirect forms (e.g., hydro, wind,
etc.), is used to manufacture hydrogen, then the resulting system is called the
solar hydrogen energy system.
In this system, both the primary and secondary energy sources are renewable and
environmentally compatible,
resulting in a clean and permanent energy system. Figure 5 presents a schematic of
the solar hydrogen energy
system.
In this case, it is assumed that the conversion to hydrogen energy
will take place, and one-third of
the hydrogen needed will be produced from hydropower (and/or wind power)
and two-thirds by direct and
indirect (other than hydropower) from solar energy forms. The same percentage of
energy demands sectors
as the above systems will be assumed. It will further be assumed that one-half of
the thermal energy will be
achieved by flame combustion, one-quarter by steam generation with hydrogen/oxygen
steam generators, and
the last quarter by catalytic combustion; electric power will be generated by fuel
cells; one-half of the surface
transportation will use gaseous hydrogen-burning internal combustion engines, and
the other half will use fuel
cells. In air transportation, both subsonic and supersonic, liquid hydrogen will be
used.
These three energy systems will now be compared environmentally and economically.
Pollution. Table 8 lists the pollutants for the three energy systems
described earlier. It can be seen that
the coal/synthetic fossil system is the worst from the environmental point of view,
whereas the solar hydrogen
energy system is the best. The solar hydrogen system will not produce any CO , CO,
SO , hydrocarbons, or
2 x
particulates, except some NO . However, the solar-hydrogen-produced NO
is much less than that produced
x
x
by the other energy systems. This is due to the fact that in the
solar hydrogen energy system only the
flame combustion of hydrogen in air will generate NOx . The other utilization
processes (such as direct steam
generation, use of hydrogen in fuel cells, hydriding processes, etc.) will not
produce any NOx .
Vapor Generation. There is a notion that the hydrogen energy
system would produce more water
vapor than the other energy systems because the fuel is pure hydrogen. When one
considers the problem in
detail, this is not so.
Only the flame combustion of fuels in air or in oxygen will produce water
vapor. In the case of hydrogen,
those other processes mentioned earlier will not produce any water vapor.
Consequently, contrary to popular
belief, the solar hydrogen energy system will produce less water vapor than the
other systems.
Global warming, which is caused by the utilization of fossil fuels,
also causes an increase in water
vapor generation. Assuming that Earth’s mean temperature has increased by 0.5◦C
since the beginning of the
Industrial Revolution (26), this additional water vapor generation and that
produced by the combustion of fuels
have been calculated. The results are presented in Table 9. It can be seen that (1)
the two fossil fuel systems
generate much more additional (above natural) water vapor than the solar hydrogen
energy system; (2) the
additional water vapor generated by global warming is much greater than that
produced by the combustion
of fuels; (3) the amount of water vapor generated by fuels is minimal compared to
that generated naturally;
and (4) the solar hydrogen energy system causes the smallest increase in vapor
generation. Again, when the
additional vapor generation is considered, the solar hydrogen energy
system becomes environmentally the
most compatible system.
Environmental Damage. Table 10 presents the environmental damage
per gigaJoule of the energy
consumed for each of the three energy systems considered and also for
their fuel components in 1998 US
dollars, as well as the environmental compatibility factors, defined as the ratio of
the environmental damage
caused by the hydrogen energy system to that caused by a given energy system. The
environmental damage
for the solar hydrogen energy system is a result of the NOx produced. It can
be seen that the solar hydrogen
energy system is environmentally the most compatible system.
It should be mentioned that hydrogen also has the answer to the
depletion of the ozone layer, which
is mainly caused by chlorofluorocarbons. Refrigeration and air-conditioning systems
based on the hydriding
property of hydrogen do not need chlorofluorocarbons but need hydrogen, and any
hydrogen leak would not
cause ozone layer depletion. Such refrigeration systems are also very
quiet because they do not have any
moving machinery.
Economics. An economical comparison between competing energy
systems should be based on the
effective costs of the services these fuels provide. The effective costs include
the utilization efficiency, the cost
of the fuel, and the costs associated with fuel consumption but that
are not included in its price (so-called
external costs). External costs include the costs of the physical damage done to
humans, fauna, flora, and the
environment as a result of harmful emissions, oil spills and leaks, and coal strip
mining, as well as governmental
expenditures for pollution abatement and expenditures for military protection of
oil supplies.
In economic considerations, it is also important to compare the future costs
of hydrogen (which will be
considerably lower than they are today because of the assumed market and technology
development) with the
future costs, both internal and external, of fossil fuels (which will unavoidably
be higher than today’s prices as
a result of depletion, international conflicts, and environmental impact).
The effective cost of a fuel can be calculated using the following
relationship:
damage caused by the fuel, η the fossil fuel utilization efficiency for
the application k, and η the synthetic
fk
sk
fuel (including hydrogen) utilization efficiency for the same application or end
use.
In order to evaluate the overall cost to society Co , the three scenarios
considered earlier will be used. This
cost can be calculated from the relationship
Using Eqs. (2)–(5), Tables 11, 12, 13 have been prepared for the three energy
scenarios (i.e., the fossil
fuel system, the coal/synthetic fossil fuel system, and the solar hydrogen energy
system) in 1998 US dollars.
Comparing the results, it becomes clear that the solar hydrogen energy system is
the most cost-effective energy
system and results in the lowest overall effective cost to society.
Gaseous Hydrogen
liquid at temperatures below 20.3 K (at atmospheric pressure). Hydrogen has the
highest energy content per
unit mass of all fuels—higher heating value is 141.9 MJ/kg, almost three times
higher than gasoline. Some
important properties of hydrogen are compiled in Table 14.
Hydrogen Production. Hydrogen is the most plentiful element
in the universe, making up about
three-quarters of all matter. All the stars and many of the planets essentially
consist of hydrogen. However, on
Earth free hydrogen is scarce. The atmosphere contains trace amounts of it (0.07%),
and it is usually found in
small amounts mixed with natural gas in crustal reservoirs. A few wells, however,
have been found to contain
large amounts of hydrogen, such as some wells in Kansas that contain 40% hydrogen,
60% nitrogen and trace
amounts of hydrocarbons (27). The Earth’s surface contains about 0.14% hydrogen
(the tenth most abundant
element), most of which resides in a chemical combination with oxygen as water.
Hydrogen, therefore, must be produced. Logical sources of hydrogen are
hydrocarbon (fossil) fuels (C H )
x y
and water (H O). Presently, hydrogen is mostly being produced from fossil fuels
(natural gas, oil, and coal).
2
Except for the space program, hydrogen is not being used directly as a fuel
or energy carrier. It is being
used in refineries to upgrade crude oil (hydrotreating and hydrocracking), in the
chemical industry to synthesize
various chemical compounds (such as ammonia and methanol), and in metallurgical
processes (as a reduction
or protection gas). The total annual hydrogen production worldwide in 1996 was
about 40 million tons (5.6 EJ)
(28). Less than 10% of this was supplied by industrial gas companies; the rest is
produced at consumer-owned
and -operated plants (so-called captive production), such as refineries, and ammonia
and methanol producers.
Production of hydrogen as an energy carrier would require an increase in production
rates by several orders of
magnitude.
Hydrogen Production from Fossil Fuels.
Steam Reforming of Natural Gas. Steam reforming of hydrocarbons
(mainly natural gas) has been the
most efficient, economical, and widely used process for hydrogen production. A
simplified basic flow diagram
of a conventional steam reforming process is shown in Fig. 6 (29). The process
basically consists of three main
steps: (1) synthesis gas generation, (2) water–gas shift, and (3) gas purification.
The feedstock (natural gas) is
mixed with process steam and reacted over a nickel-based catalyst contained inside
a system of alloyed steel
tubes. To protect the catalyst, natural gas must be desulfurized before being fed
to the reformer. The following
◦ ◦
fuel oil. The metallurgy of the tubes usually limits the reaction temperature to
700 to 925 C. The synthesis
gas leaving a catalytic reformer is typically a mixture of H , CO, CO , and CH .
After the reformer, the gas
2
2 4
mixture passes through a heat recovery step cooling it down to about 350◦C and
feeding it into a water–gas
shift reactor to produce additional hydrogen. Next, the cold gas passes through gas
purification units to remove
the CO , the remaining CO, and other impurities in order to deliver purified
hydrogen. Several commercial
2
processes can be used for removing CO2 (and CO), such as wet
scrubbing, pressure swing adsorption, and,
recently, membrane processes.
The efficiency of the steam reforming process, expressed as the
ratio of the heating value of produced
hydrogen and energy input (feedstock, fuel and small amounts of electricity) is
about 65% to 75%. The cost of
produced hydrogen is about $6/GJ but is strongly dependent upon the cost and
availability of natural gas or
other fuel feedstock.
Partial Oxidation. Partial oxidation is used for converting hydrocarbons
heavier than naphtha (for which
steam reforming is not applicable). Most recently, partial oxidation reformers are
being developed for on-board
reforming of natural gas, ethanol, and even gasoline, to be coupled with fuel cells
for automobile propulsion. A
simplified basic flow diagram of an industrial partial oxidation process is shown in
Fig. 7. There are three main
steps: (1) synthesis gas generation, (2) water–gas shift reaction, and (3) gas
purification. The partial oxidation
reactions are typically as follows:
◦
industry. A methane–air flame is used to heat up firebrick to temperatures in the
order of 1400 C. The air is
then turned off, and the methane alone decomposes on the hot firebrick until the
temperature drops to about
800◦C. The micron-size carbon particles are collected in the effluent gas
stream in bag filters. The process
and the design of the reactor. Because of the lower calorific value per unit mass of
biomass as compared to coal,
the processing facility is larger than that of comparably sized coal gasification
plants.
Hydrogen Production from Water. The most logical source for
large-scale hydrogen production is water,
which is abundant on Earth. Different methods of hydrogen production from
water have been or are being
developed. They include
• electrolysis,
• direct thermal decomposition or thermolysis,
• thermochemical processes, and
• photolysis.
◦ ◦
• High temperature steam electrolysis, which operates between 700
and 1000 C and which employs oxygen
ion-conducting ceramics as electrolyte. Electrical energy consumption is
reduced because part of the energy
required for water dissociation is supplied in the form of heat. The water to
be dissociated is entered on the
cathode side as steam which forms a steam–hydrogen mixture during electrolytic
dissociation. The O2 -ions
are transported through the ceramic material to the anode, where they are
discharged as oxygen (38).
An electrolysis plant can operate over a wide range of capacity factors and
is convenient for a wide range
of operating capacities, which makes this process interesting for coupling
with renewable energy sources,
particularly with photovoltaics. Photovoltaics generate low voltage–direct
current, which is exactly what is
required for the electrolysis process.
Performance of photovoltaic–electrolyzer systems has been studied
extensively both in theory and in
practice (39,40,41,42). Several experimental PV-electrolysis plants are currently
operating all over the world,
such as
(1) Stationary Large Storage Systems These are typically storage devices at the
production site or at the start
or end of pipelines and other transportation pathways.
(2) Stationary Small Storage Systems These can be found at the distribution or
final user level, for example, a
storage system to meet the demand of an industrial plant.
(3) Mobile Storage Systems for Transport and Distribution These include
both large-capacity devices, such
as a liquid hydrogen tanker–bulk carrier, and small systems, such as a gaseous
or liquid hydrogen truck
trailer.
(4) Vehicle Tanks These store hydrogen used as fuel for road vehicles.
than expected losses of the working gas in the amount of 1% to 3% per year are not
anticipated. The city of
Kiel’s public utility has been storing town gas with a hydrogen content of 60% to
65% in a gas cavern with a
geometric volume of about 32,000 m3 and a pressure of 80 to 160 bar at a
depth of 1330 m since 1971 (54).
Gaz de France, the French National Gas Company, has stored hydrogen-rich refinery
by-product gases in an
aquifer structure near Beynes, France. Imperial Chemical Industries of Great
Britain stores its hydrogen in
the salt mine caverns near Teeside in the United Kingdom (55).
Above-Ground Pressurized Gas Storage Systems. Pressurized gas
storage systems are used today in
natural gas businesses in various sizes and pressure ranges from standard pressure
cylinders (0.05 m3 , 200
3
bar) to stationary high-pressure containers (over 200 bar) or low-pressure
spherical containers ( >30,000 m ,
12 bar to 16 bar). This application range will be similar for hydrogen storage.
Vehicular Pressurized Hydrogen Tanks. Development of ultralight
but strong new composite materials
has enabled storage of hydrogen in automobiles. Pressure vessels that allow
hydrogen storage at pressures
>200 bar have been developed and used in automobiles (such as Daimler-Benz NECAR
II). Storage density of
higher than 0.05 kg H2/kg of total weight is easily achievable (56).
Metal Hydride Storage. Hydrogen can form metal hydrides with
some metals and alloys. During the
formation of the metal hydride, hydrogen molecules are split, and hydrogen atoms
are inserted in spaces inside
the lattice of suitable metals and/or alloys. In such a way, an effective storage
is created comparable to the
density of liquid hydrogen. However, when the mass of the metal or alloy is taken
into account, then the metal
hydride gravimetric storage density is comparable to storage of
pressurized hydrogen. The best achievable
gravimetric storage density is about 0.07 kg H2/kg of metal, for a high
temperature hydride such as MgH2
as shown in Table 16, which gives a comparison of some hydriding substances with
liquid hydrogen, gaseous
hydrogen, and gasoline (57).
storage.
Hydrogen Conversion Technologies. Hydrogen as an energy
carrier can be converted in useful
forms of energy in several ways, namely:
engine is
where
r = compression ratio and
k = ratio of specific heats (C /C ).
p v
Equation (16) shows that the thermal efficiency can be improved by increasing
either the compression
ratio or the specific heat ratio. In hydrogen engines, both ratios are higher than
in a comparable gasoline engine
because of hydrogen’s lower self-ignition temperature and ability to burn in lean
mixtures. However, the use
of hydrogen in internal combustion engines results in the loss of power because of
the lower energy content
in a stoichiometric mixture in the engine’s cylinder. A stoichiometric mixture of
gasoline and air and gaseous
hydrogen and air premixed externally occupy ∼2% and 30% of the cylinder volume,
respectively. Under these
conditions, the energy of the hydrogen mixture is only 85% of the gasoline mixture,
thus resulting in about 15%
reduction in power. Therefore, the same engine running on hydrogen will have ∼15%
less power than when
operated with gasoline. The power output of a hydrogen engine can be improved by
using more advanced fuel
injection techniques or liquid hydrogen. For example, if liquid hydrogen is
premixed with air, the amount of
hydrogen that can be introduced in the combustion cylinder can be increased by
approximately one-third (65).
One of the most important advantages of hydrogen as a fuel for
internal combustion engines is that
hydrogen engines emit by far fewer pollutants than comparable gasoline engines.
Basically, the only products
of hydrogen combustion in air are water vapor and small amounts of nitrogen oxides.
Hydrogen has a wide
flammability range in air (5% to 75% vol.), therefore, high excess air
can be utilized more effectively. The
formation of nitrogen oxides in hydrogen/air combustion can be minimized
with excess air. NOx emissions
can also be lowered by cooling the combustion environment using techniques such as
water injection, exhaust
gas recirculation, or using liquid hydrogen. The emissions of NOx
in hydrogen engines are typically one
order of magnitude smaller than emissions from comparable gasoline
engines. Small amounts of unburned
hydrocarbons, CO2 , and CO have been detected in hydrogen engines due to
lubrication oil (65).
The low-ignition energy and fast-flame propagation of hydrogen have led to
problems of preignition and
backfire. These problems have been overcome by adding hydrogen to the air
mixture at the point where
and when the conditions for preignition are less likely, such as delivering the
fuel and air separately to the
combustion chamber and/or injecting hydrogen under pressure into the combustion
chamber before the piston
is at the top dead center and after the intake air valve has been
closed. Water injection and exhaust gas
recirculation techniques are also used in hydrogen engines to help control
premature ignition. Note that most
of the research on hydrogen combustion in internal combustion engines has been
conducted with modifications
of existing engines designed to burn gasoline. Redesign of the combustion
chamber and coolant systems to
accommodate hydrogen’s unique combustion properties could be the most
effective method of solving the
problems of preignition and knocking (65).
Hydrogen use in turbines and jet engines is similar to the use of
conventional jet fuel. The use of hydrogen
avoids the problems of sediments and corrosion on turbine blades which prolongs
life and reduces maintenance.
Gas inlet temperatures can be pushed beyond normal gas turbine temperatures of
800◦C, thus increasing the
overall efficiency. The only pollutants from the use of hydrogen in turbines and jet
engines are nitrogen oxides.
Steam Generation by Hydrogen/Oxygen Combustion. Hydrogen
combusted with pure oxygen results
in pure steam; that is,
◦
This reaction would develop temperatures in the flame zone above 3000 C; therefore,
additional water has to
be injected so that the steam temperature can be regulated at a desired level. Both
saturated and superheated
vapor can be produced.
The German Aerospace Research Establishment (DLR ) has developed a compact
hydrogen/oxygen steam
generator (66). The steam generator consists of the ignition, combustion, and
evaporation chambers. In the
ignition chamber, a combustible mixture of hydrogen and oxygen at a low
oxidant/fuel ratio is ignited by means
of a spark plug. The rest of the oxygen is added in the combustion chamber to
adjust the oxidant/fuel ratio
exactly to the stoichiometric one. Water is also injected in the combustion chamber
after it has passed through
the double walls of the combustion chamber. The evaporation chamber serves to
homogenize the steam. The
steam temperature is monitored and controlled. Such a device is close to 100%
efficient because there are no
emissions other than steam and little or no thermal losses.
The hydrogen steam generator can be used to generate steam for spinning
reserve in power plants, for
peak load electricity generation, in industrial steam supply networks, and
as a micro steam generator in
medical technology and biotechnology applications (66).
Catalytic Combustion of Hydrogen. Hydrogen and oxygen in the
presence of a suitable catalyst may be
combined at temperatures significantly lower than flame combustion (from ambient to
500◦C). This principle
• Alkaline fuel cells (AFC ) use concentrated (85 wt %) KOH as the electrolyte
for high-temperature opera-
◦
◦
tion (250 C) and less concentrated (35 wt % to 50 wt %) for lower-temperature
operation (<120 C). The
◦
◦ ◦
Operating temperatures are usually below 100 C, more typically between 60
and 80 C.
• Phosphoric acid fuel cells (PAFC ), use concentrated phosphoric acid (∼100%)
as the electrolyte. The matrix
used to retain the acid is usually SiC, and the electrocatalyst in both the
anode and cathode is Platinum
◦
◦
black. Operating temperatures are typically between 150 C and 220 C (67,68).
◦
carbonates, which are retained in a ceramic matrix of LiAlO2 . Operating
temperatures are between 600 C
and 700◦C where the carbonates form a highly conductive molten salt, with
carbonate ions providing ionic
2 3 2
◦ ◦
electrolyte. The cell operates at 900 to 1000 C where ionic
conduction by oxygen ions takes place (67,68).
2 2 2
Low-temperature fuel cells (AFC, PEMFC, PAFC) require noble electrocatalysts to
achieve practical reaction
rates at the anode and cathode. High-temperature fuel cells (MCFC and SOFC) can
also utilize CO and CH4 as
fuels. The operating temperature is high enough so that CO and CH4 can be
converted into hydrogen through
the water–gas shift and steam reforming reactions, respectively.
The electrolyte not only transports dissolved reactants to the electrode,
but it also conducts ionic charge
between the electrodes and thereby completes the cell electric circuit, as shown in
Fig. 10.
◦
The reversible potential of the above electrochemical reactions is 1.229 V
(at standard conditions, i.e., 25 C
and atmospheric pressure), and it corresponds to the Gibbs free energy according to
the following equation:
where
0 ◦
DG = Gibbs free energy at 25 C and atmospheric pressure
n = number of electrons involved in the reaction
F = Faraday’s constant
0 ◦
E = reversible potential at 25 C and atmospheric pressure (V)
shows a typical fuel cell polarization curve with pronounced regions of predominant
irreversible losses. Figure
12 shows actual polarization curves of some representative fuel cells. The fuel
cells are typically operated in a
range between 0.6 V and 0.8 V. The Space Shuttle fuel cell (alkaline) is designed
to operate at 0.86 V and 410
mA/cm2 (67). PEM fuel cells have the highest achievable current densities,
between 1 and 2 mA/cm2 at 0.6 V
where
hTh = thermal efficiency, which is the ratio between Gibbs free energy of
the reaction and the heating
value of the fuel, DGr/DHfuel (similar to internal combustion engines, the
fuel cell efficiency is often
expressed in terms of the lower heating value);
hv = voltage efficiency, defined as a ratio between the actual voltage (V )
and thermodynamic voltage (E )
(i.e., V/E );
hF = Faradaic efficiency, or the ratio between the actual current and the
current corresponding to the
rate at which the reactant species are consumed [I/nFm, where m is the rate (in
moles/s) at which the
reactants are consumed];
hU = fuel utilization, or the ratio between the amount of fuel actually
consumed in the electrochemical
reaction and fuel supplied to the fuel cell.
and the hydride tanks are connected by water-carrying pipe circuits or loops,
equipped with a set of switches
and valves, so that a hydride tank in a given water loop can be replaced by another
hydride tank.
When the system works as a heater, the heat from a heat source (e.g., solar
collector) is carried to tank
1 at about 100◦C. The heat drives the hydrogen from tank 1 to tank 3, where
hydrogen is absorbed forming
a hydride, and heat is released at 40◦C. The water loop carries this heat to the
building heat exchangers and
heats the air in the building. At the same time, water in the other loop absorbs
heat from the ambient and
carries it to tank 4. This heat drives off the hydrogen from tank 4 to tank 2 where
hydride is formed and heat
is generated at 40◦C. The whole operation of driving hydrogen from tanks 1 and 4 to
tanks 3 and 2 takes about
2 minutes. At the end of this cycle, the hydride tanks are switched from one loop
to the other in cycle II (as
shown in Fig. 14). Now, the solar and ambient heats are used to drive off the
hydrogen in tanks 2 and 3 to tanks
4 and 1, respectively. The heat produced during the absorption processes in tanks 1
and 4 is used for heating
the building. After this, the cycles are repeated.
When the system works as an air conditioner, the building heat exchangers are
placed in the 8◦C water
loop, while the outside heat exchangers are placed in the 40◦C water loops, and the
operation proceeds in two
where the numerator in the exponent term is the hydriding reaction enthalpy, R is
the gas constant, and S is
a constant depending on the hydriding substance. If hysteresis is neglected, Eq.
(25) would hold for both the
absorption and desorption processes. By using heat, it would be possible to
increase hydrogen pressure by a
hundred or more times. The pressure ratio would be
Hydrogen has a very low ignition energy (0.02 mJ), about one order of
magnitude lower than other fuels.
The ignition energy is a function of fuel/air ratio, and for hydrogen it reaches a
minimum at about 25% to 30%.
At the lower flammability limit, hydrogen ignition energy is comparable with that of
natural gas (70).
Hydrogen has a flame velocity seven times faster than that of natural gas or
gasoline. A hydrogen flame
would therefore be more likely to progress to deflagration or even
detonation than other fuels. However,
the likelihood of a detonation depends in a complex manner on the exact fuel/air
ratio, the temperature, and
particularly the geometry of the confined space. Hydrogen detonation in the open
atmosphere is highly unlikely.
The lower detonability fuel/air ratio for hydrogen is 13% to 18%, which is
two times higher than that of
natural gas and 12 times higher than that of gasoline. Because the lower
flammability limit is 4%, an explosion
is possible only under the most unusual scenarios (e.g., hydrogen would first have
to accumulate and reach 13%
concentration in a closed space without ignition, and only then an ignition source
would have to be triggered).
Should an explosion occur, hydrogen has the lowest explosive energy per unit
stored energy in the fuel,
and a given volume of hydrogen would have 22 times less explosive energy than the
same volume filled with
gasoline vapor.
Hydrogen flame is nearly invisible, which may be dangerous because people in
the vicinity of a hydrogen
flame may not even know there is a fire. This may be remedied by adding some
chemicals that will provide
the necessary luminosity. The low emissivity of hydrogen flames means that near-by
materials and people will
be much less likely to ignite and/or be hurt by radiant heat transfer. The fumes
and soot from a gasoline fire
pose a risk to anyone inhaling the smoke, whereas hydrogen fires produce only water
vapor (unless secondary
materials begin to burn).
Liquid hydrogen presents another set of safety issues, such as the risk of
cold burns and the increased
duration of leaked cryogenic fuel. A large spill of liquid hydrogen has some
characteristics of a gasoline spill;
however, it will dissipate much faster. Another potential danger is a
violent explosion of a boiling liquid
expanding vapor in case of a pressure relief valve failure.
In conclusion, hydrogen appears to pose risks of the same order of magnitude
as other fuels. In spite of
public perception, in many aspects hydrogen is actually a safer fuel than gasoline
and natural gas. As a matter
of fact, hydrogen has a very good safety record, as a constituent of the “town gas”
widely used in Europe and
the United States in the 19th and early 20th century, as a commercially used
industrial gas, and as a fuel in
space programs. There have been accidents, but nothing that would characterize
hydrogen as more dangerous
than other fuels has occurred.
One of the most remembered accidents involving hydrogen is the Hindenburg
dirigible disaster in 1937.
However, hydrogen did not cause that accident, and hydrogen fire did not directly
cause any casualties. The
accident appears to have been caused by a static electricity discharge,
and it was the balloon’s lining that
caught fire first (71). Once hydrogen that the balloon was filled with for buoyancy
(instead of helium as it was
originally designed to be filled with) was ignited, it burned (as any fuel is
supposed to). However, hydrogen
fire went straight up, and it did not radiate heat so the people in the gondola
underneath the balloon were
not burned or suffocated. As a matter of fact, 56 survivors walked out of gondola
when it landed and all the
hydrogen and balloon structure burned out. Therefore, even in a worst case scenario
accident, hydrogen proved
to be a safe fuel.
Liquid Hydrogen
Hydrogen may be stored in gaseous, liquid, slush, and solid forms (metal
hydrides). Every mode of storage
has its advantages and disadvantages in terms of cost, stability,
convenience of usage, and energy density.
Liquid hydrogen has several advantages over other storage modes, especially in
terms of energy density and
convenience of utilization. However, several problems stand in the way of its
large-scale production and long-
term storage. This includes problems associated with energy consumed in
liquefaction as well as boil-off losses
the action of an externally applied magnetic field. This effect tends to achieve
saturation in a field of about
1000 Oersted (79577 A/m) (75). In a magnetocaloric liquefier, there is a good chance
of taking advantage of
this phenomenon primarily because both the magnetic ions and the external magnetic
field are available, thus
making a good design the only prerequisite for achieving a fast and economic
conversion.
The use of a catalyst usually results in a larger refrigeration load and
consequently in an efficiency penalty
primarily because the heat of conversion must be removed. The time for which
hydrogen is to be stored usually
determines the optimum amount of conversion. For use within a few hours, no
conversion is necessary. For
example, large-scale use of liquid hydrogen as a fuel for jet aircraft is one of
those cases where conversion is not
necessary because utilization of the liquid is almost a continuous process and
long-term storage is therefore not
needed (80). For some other uses, a partial conversion might be required to create
more favorable conditions. It
should be noted that, for every initial ortho concentration, there exists a unique
curve for boil-off of hydrogen
with respect to time.
The heat of conversion at higher temperatures is small (at 300 K, the heat of
conversion is 270 kJ/kg) and
increases as the temperature decreases, where it reaches 519 kJ/kg at 77K, the nbp
for liquid nitrogen. For
temperatures lower than the nbp of nitrogen, the heat of conversion remains
constant (523 kJ/kg). However, the
equilibrium percentage of para-hydrogen at the nbp is 50%. Therefore, in order to
produce equilibrium liquid
hydrogen, 50% of hydrogen must be converted from ortho to para at temperatures
lower than liquid nitrogen
temperatures. Because the heat of conversion is almost constant for temperatures
lower than those of liquid
nitrogen, the refrigeration load required to remove the heat of conversion should
be almost constant at different
temperatures. Ideal refrigeration requires that the energy needed for the same
amount of refrigeration load
be temperature-dependent. This suggests that, in order to operate at an
optimum power, it is necessary to
have continuous conversion, especially for temperatures lower than those
of liquid nitrogen. By continuous
conversion, up to 15% energy saving may be achieved.
Boil-Off Resulting from Heat Leak.
Reservoir Shape and Size Effect. Hydrogen is usually transported in
large quantities by truck tankers
of 30 m3 to 60 m3 capacity, by rail tank cars of 115 m3 capacity, and by
barge containers of 950 m3 capacity
(82). Liquid hydrogen storage vessels are usually available in sizes ranging from
one liter dewar flasks used
in laboratory applications to large tanks of 5000 m3 capacity. The National
Aeronautics and Space Adminis-
tration (NASA ) typically uses large tanks of 3800 m3 capacity (25 m in
diameter) (83). The total boil-off rate
from such dewars is approximately 600,000 LPY (liters per year), which is vented
to a burn pond. The heat
leakage losses are generally proportional to the ratio of surface area to the
volume of the storage vessel (S/V ).
The most favorable shape is therefore spherical because it has the least
surface to volume ratio. Spherical
shape containers have another advantage. They have good mechanical strength because
stresses and strains
are distributed uniformly. In large sizes, spherical containers are expensive
because of their manufacturing
difficulty. They are often used for on-site large volume storage, however.
Storage vessels may also be constructed in other shapes such as cylindrical,
conical, or any combination of
these shapes. Cylindrical vessels are usually required for transportation of liquid
hydrogen by trailers or railway
cars because of limitations imposed on the maximum allowable diameter of the
vessel. For normal highway
transportation, the outside diameter of the vessel cannot exceed 2.44 m.
From an economics standpoint,
cylindrical vessels with either dish, elliptical, or hemispherical heads are very
good, and their S/V ratios are
only about 10% greater than that of the sphere (84).
Because boil-off losses resulting from heat leak are proportional to S/V as
reported earlier, the evapora-
tion rate will diminish drastically as the storage tank size is increased. For
double-walled vacuum-insulated
spherical dewars, boil-off losses are typically 0.3% to 0.5% per day for containers
having a storage volume of
50 m3 , 0.2% for 103 m3 tanks, and about 0.06% for 19,000 m3 tanks (85).
Obviously, the larger the size of the
dewar, the smaller the cost per unit volume of storage. Interestingly enough, the
rate of evaporation does not
substantially decrease with increasing the size of the container for cylindrical
vessels of constant diameter.
Thermal Stratification and Thermal Overfill. Stratification is the
existence of a fluid at different tem-
peratures within one vessel. Heat transfer from the sidewalls and bottom to a
liquid hydrogen storage vessel
will always result in an increase of the sensible enthalpy of the liquid. When the
heat flux is small enough and
the inside wall of the vessel is smooth, boiling is unlikely to occur (77). In this
case, the added energy tends to
get distributed nonuniformly because the imposed heat flux usually produces a free
convective flow. The warm
liquid hydrogen becomes buoyant and rises to the top. With the passage of time, the
lower boundary of the
stratified region moves toward the bottom of the vessel. Because the warm liquid
rises to the surface, it is not
uncommon to have the temperature at the top layers rise much more rapidly than that
of the bulk liquid (84).
As a consequence, the upper layers are more likely to experience a relatively large
vapor pressure increase,
whereas the lower layers experience little or no vapor pressure change. This
thermal stratification creates a
(1) The average specific enthalpy of the liquid is greater than that of the
saturation temperature, in which
case the liquid boil-off from the vented dewars is found to be nonuniform. This
can be attributed to the
unstable stratification which occurs in vented dewars as a result of
a sudden lowering of atmospheric
pressure (generally, the liquid becomes superheated in the lower
region, while the surface temperature
remains saturated, which leads to generating instant vapor of large
quantities).
(2) The operating pressure of the tank is less than that corresponding to the
saturation temperature of the
feed liquid, in which case only the surface layer is able to respond to the new
operating pressure. If the
surface layer is distributed by some means, for instance by top-loading more
liquid, the underlying liquid
may be brought into rapid equilibrium, and that, in turn, would cause a rapid
boil-off (77,86).
in the system. The air could be thought of as presenting a potential safety problem
in a hydrogen system. Voth
also showed that the freeze–thaw process required either costly equipment to
recover the generated triple-point
vapor or the loss of approximately 16% of the normal boiling-point liquid hydrogen
and approximately 24% of
the normal boiling-point liquid oxygen if the vapor was discarded.
Auger Method. Voth (107) reported on a study for producing liquid–solid
mixtures of oxygen or hydrogen
using an auger. He described an auger used to scrape frozen solid from the inside
of a refrigerated brass tube
in order to produce slush hydrogen. He showed that slush hydrogen could be
continuously produced by this
method, and because it could be immersed in liquid, slush was produced at pressures
above the triple-point
pressure. Voth was also able to produce the increased pressure pneumatically or by
generating a temperature
stratification near the surface of the liquid. He observed that the auger system
produced particles in the size
realm of the particle produced by the freeze–thaw method so that, like the freeze–
thaw-produced slush, the
auger-produced slush could be readily transferred and stored.
According to Voth (107), the surface freezing occurring in the freeze–thaw
production process is thermo-
dynamically reversible. In contrast, the freezing process in the auger is
irreversible because a temperature
difference must exist between the refrigerant and the freezing cryogen, and the
energy added to scrape the
solid cryogen out of the brass tube must be removed by the refrigerant.
Voth argued that in spite of these
irreversibilities, the auger system was found to require less energy to produce
slush hydrogen than a particle
freeze–thaw system. He reported that the temperature difference required
refrigeration temperatures below
the triple-point temperature of hydrogen so that a gaseous or liquid helium
refrigerator was required for the
auger.
Although this method was determined to offer the advantage of operating above
atmospheric pressures
(an important consideration in lowering the possibility of oxygen intrusion
into the slush), it was shown to
have the disadvantage of using expensive helium coolant.
Quality Determination. A thermodynamic analysis of the freeze–thaw
production process for forming
mixtures of slush developed by Mann et al. (109) showed that the measurement of two
quantities—(1) the mass
fraction pumped during the production process, and (2) the heat leak per unit mass
into the production dewar—
were sufficient to determine the quality of the resulting mixture. Because
irreversibilities were observed to
exist in the freeze–thaw process, a comparison of the predicted quality with the
qualities determined by the
measurement of thermodynamic properties was found necessary.
According to Daney and Mann (108), in the freeze–thaw method of
forming a triple-point mixture of
hydrogen, the latter was partially evaporated under the reduced pressure obtained
by a vacuum pump, while a
refrigeration effect (approximately equal to the latent heat of vaporization) was
experienced by the remaining
liquid or liquid–solid mixture. They suggested, by specifying the initial
state and the process or path to be
followed, that the quality of the liquid–solid mixture could be predicted as a
function of the mass of the vapor
removed. According to them, therefore, the problem of quality prediction consisted
of two parts. The first was
the determination of when a specific initial state, triple-point liquid was reached,
whereas the second was the
prediction of the quality once this standard initial state was achieved. They also
reported that the arrival at the
triple-point liquid condition may be determined in several ways: (1) vapor pressure
measurement, (2) visual
observation of the formation of a few solid particles, and (3) prediction from
the mass fraction of the vapor
pumped off in the cooling-down process. Daney and Mann argued that the first two
methods provided a more
accurate means of determining the arrival at triple-point liquid conditions, and
hence a more accurate quality
determination could be realized. However, they also argued that it may be of
interest to see what accuracy in
quality determination could be expected by starting the mass accounting with a
normal boiling-point liquid
instead of a triple-point liquid.
Energy Requirements. A study to determine the energy requirements
of the freeze–thaw and auger
methods was reported by Voth (107) who mentioned that producing slush appears to be
a reasonable way of
increasing the density and heat capacity of cryogens. He observed that when large
quantities of slush were
produced, the energy required to produce them became significant. He therefore
concluded that the production
particle size determines the minimum restrictions permissible in systems using the
mixtures. They concluded
that although particle configuration changed with age, particle size did not change
significantly and that the
distribution of particle size could be satisfactorily described by a modified
logarithmic function. They reported
that size ranged from 0.5 mm to 10 mm, with 2 mm being the most frequently
occurring size. They also reported
over 170 settling velocity measurements on both fresh and aged slush particles and
concluded that the settling
velocity of fresh particles was lower than that of aged particles and was extremely
dependent on the shape of
individual particles.
Sindt (106) reported that mixing of slush hydrogen to maintain homogeneity
was required in most of the
potential applications, and if mixers were to impart velocities to the fluid greater
than the settling velocities,
particles would have been carried along and mixing would have resulted. Sindt used
several methods to mix
the slush in a vessel. One method used a simple propeller with blades designed to
impact velocity to the fluid
with lift. Another method used high-velocity streams of fluid expelling from
openings in a duct. According to
him, both methods of mixing had been used to mix slush of a solid fraction as high
as 0.6. Also, the solid content
in slush could be maintained and even increased by transferring slush
into the vessel and removing liquid
through a screen with 0.6 mm openings. Sindt explained that by using this method
the solid fraction could be
increased until the solids protrude above the liquid–gas interface.
Transfer Characteristics. Most of the studies on transfer
characteristics of slush hydrogen have
been carried out at the Cryogenics Division of the National Institute of
Standards and Technology (NIST ).
Experimental results pertaining to slush hydrogen fluidity will be reported in this
section.
Slush Hydrogen Flow in Pipes. Sindt (106) reported that transfer and
pumping characteristics of slush
hydrogen should be similar to those of liquid hydrogen if slush is to be used in
the existing liquid hydrogen
system and if slush is to be handled as a fluid. During the
experimental study of Sindt et al. (113), slush
of solid fractions as high as 0.6 was transferred through vacuum-insulated pumps of
16.6 mm to 25 mm in
diameter. Transfer pressure losses were determined in triple-point liquid and in
slush of solid fractions up to
0.55 in a 16.6 mm diameter pipe. They estimated slush data to have an uncertainty
occurring at the high solid
fractions and the lowest flow rates. According to them, pressure loss data for
flowing slush revealed that, at
low flow rates, pressure losses in flowing slush were double those of triple-point
liquid if solid fractions were
high. The data also revealed that at high flow rates, pressure losses in flowing
slush were less than the losses
in triple-point liquid if the solid fraction was near 0.35. They also carried out a
more complete analysis of the
losses in flowing slush hydrogen by calculating a friction factor and plotting it
versus the Reynolds number. In
the calculations, they used triple-point liquid viscosity because the viscosity of
slush hydrogen was generally
unknown.
Sindt and Ludtke (112) observed that the characteristics of slush
hydrogen flow losses at the higher
Reynolds numbers were more evident when displayed in terms of the
friction factors than if displayed as
pressure losses. They reported that friction losses in flowing slush of solid
fractions to 0.4 were as much as 10%
less than in liquid and that the friction-loss curve for 0.5 solid fraction slush
appeared to be following a slope
such that it would be below triple-point liquid losses at high Reynolds numbers.
They reported these data for
a 16.6 mm diameter pipe and concluded that low losses in slush aged 10 hours were
4% to 10% higher than
losses in 1-hour old slush.
The experiments of Sindt et al. (113) revealed that a gradient in
the solid concentration may develop
when the slush velocity is 0.5 ms − 1, whereas at a velocity of 0.15 ms − 1, the
solids were likely to settle forming
a sliding bed. Considering these observations, it was concluded that the critical
velocity of slush in the pipe
was about 0.5 m/s.
Slush Hydrogen Flow in Restrictions. Sindt et al. (113)
experimentally determined flow losses in re-
strictions such as valves, orifices, and venturies using slush hydrogen of solid
fractions to 0.5. They determined
the losses in a 3/4 inch nominal copper globe valve with the valve fully opened and
in cases when it was partially
opened and expressed the losses as a resistance coefficient independent of the slush
solid fraction. According
to their study, with the valve partially opened, the losses seemed to be less with
increasing the solid fraction,
whereas with the valve fully opened, the trend was reversed. Also, the loss
coefficients compared favorably
with the typical valve coefficient for water.
Sindt and Ludtke (112) remarked that liquid–solid mixtures of hydrogen flow
through orifices of 6.4 mm
and 9.5 mm in diameter with no plugging and with a pressure drop of
the same magnitude as that of the
triple-point liquid. They observed that the slush mass flow rate was dependent on
the pressure differential
across the orifice as would have been expected for a nearly incompressible,
Newtonian fluid and that when
the minimum pressure in the orifice reached the triple-point pressure, cavitation
with accompanying choking
occurred. At that point, the mass flow became dependent on the upstream pressure and
independent of the
downstream pressure. Flow characteristics of slush flowing through a venturi have
also been determined by
Sindt et al. (113). According to them, flow losses through the venturi were not
affected by the solid fraction of
the slush.
Concluding Remarks on Slush Hydrogen. This section
presented an overview of the state-of-the
art of slush hydrogen production and utilization technologies. It is apparent that
the use of slush hydrogen
should be considered only for cases in which higher density and greater solid
content are really needed. This
is mainly because the production costs of slush hydrogen now and in the
near future are greater than the
liquefaction costs of hydrogen resulting from the larger energy use involved.
Conclusions
As a result of the aforementioned facts, it can be seen that hydrogen as a fuel and
the Solar Hydrogen Energy
System have unmatched advantages when compared with fossil fuels and the Fossil
Fuel System, respectively.
`
Hydrogen. The advantages of hydrogen vis-a-vis fossil fuels can be
summed up as follows:
(1) Liquid hydrogen is the best transportation fuel when compared with liquid
fuels such as gasoline, jet fuel,
and alcohols; and gaseous hydrogen is the best gaseous transportation fuel.
(2) Hydrogen can be converted to useful energy forms (thermal, mechanical, and
electrical) at the user end
through five different processes, whereas fossil fuels can be converted through
only one process (i.e., flame
combustion). In other words, hydrogen is the most versatile fuel.
(3) Hydrogen has the highest utilization efficiency when it comes to conversion to
useful energy forms (thermal,
mechanical, and electrical) at the user end. Overall, hydrogen is 39% more
efficient than fossil fuels. In other
words, hydrogen will save primary energy resources. It could also be termed as
the most energy-conserving
fuel.
(4) When fire hazards and toxicity are taken into account, hydrogen becomes the
safest fuel.
`
Solar Hydrogen Energy System. The advantages of the Solar
Hydrogen Energy System vis-a-vis the
present Fossil Fuel System and the Synthetic Fossil Fuel System can be summed up as
follows:
• 5.When the environmental impact is taken into consideration, the Solar Hydrogen
Energy System becomes
the most environmentally compatible energy system. It will not produce
greenhouse gases, ozone-layer-
damaging chemicals, oil spills, and climate change and will produce little or no
acid rain ingredients and
pollution. It will actually reverse the global warming and bring the Earth back
to its normal temperatures
by decreasing the CO2 in the atmosphere to its level before the Industrial
Revolution.
• 6.The Solar Hydrogen Energy System has the lowest effective cost when
environmental damage and the
higher utilization efficiency of hydrogen are taken into account. In other words,
the Solar Hydrogen Energy
System will cost society least when compared with the present Fossil Fuel System
and the Synthetic Fossil
Fuel System.
BIBLIOGRAPHY
S. A. SHERIF
University of Florida
F. BARBIR
T. N. VEZIROGLU
University of Miami
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3026.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Hydrothermal Power Systems
● Email this page Enter words or
Standard Article
Abstract
Problem Description
Solution Methodologies
Conclusions
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3026.htm17.06.2008 16:42:15
units on an hourly basis over a horizon from one day to ten days. It encompasses
two interrelated problems:
(1) the simultaneous determination of the “unit commitment” (unit up and
down) status and preliminary
hourly generation levels for all units and (2) the “economic dispatch”
of power (i.e., the determination of
hourly generation levels) given the unit commitment status. The second
problem is a special case of the
first. Although hydrothermal scheduling provides a preliminary power output for each
unit, actual generation
levels are decided by an economic dispatch calculation on an hourly basis using the
latest demand and cost
information. The focus of this article is on short-term hydrothermal scheduling.
The approaches developed for
short-term scheduling, however, can often be used for midterm and longterm
scheduling problems.
Problem Description
where p (t) is unit power output at time t, and C0 the no-load cost. For many
units, C [p (t)] can be well approxi-
mated by a quadratic, piecewise quadratic, or by an n-segment piecewise linear and
convex function:
as shown in Fig. 2, where p˜ is the generation capacity and is the minimum
generation. The cost curve in Eq.
(1) is smooth, and the cost in Eq. (2) is computationally convenient for economic
dispatch. The forms of the cost
functions, however, are not essential for solving hydrothermal scheduling problems.
The startup cost of a thermal unit is made up of the fuel and operating costs
to bring the unit on-line; it
is a time-varying functuion because it depends on the unit status (i.e., the time
since last shutdown). In the
literature, this cost is often assumed to be an exponential function of unit down
time as
where t is the time since last shutdown, Cs a positive constant representing the
cold startup cost, and d is the
boiler cool-down coefficient. For certain units, the startup cost is modeled as a
linear function of unit downtime,
while remaining constant after the cold startup time (3). The startup or
shutdown costs for a hydro or a
pumped-storage unit are usually ignored because they are insignificant as compared
to those of a thermal unit.
Frequent startups and shutdowns, however, may increase maintenance costs (4).
Therefore, penalty terms can
be added to discourage excessive startups and shutdowns of hydro and pumped-storage
units.
System-Wide Constraints. The system demand or load balance
equation is written as
where p ti (t), p hj (t), p pk (t) are generation levels of thermal unit i, hydro
unit j , and pumped-storage unit k at
time t, respectively; Pd (t) the total system demand at time t containing the
aggregated customer demand and
losses (approximately accounted for by solving an optimal power flow problem). The
generation of a pumped-
storage unit may take negative values representing pumping. If transactions among
utilities are considered,
the system demand in Eq. (4) should also add the net power sold and deduce the net
power purchased.
The system reserve requirements include spinning and supplemental reserves
which are the extra power
that the system should be able to provide within specified times guard against
disturbances such as unscheduled
outage of a unit or unforeseen load changes. The spinning reserve is the reserve
that on-line units can provide.
Generally, the spinning reserve contribution of an on-line unit is the amount of
additional power that the unit
can provide within, say, 10 min {i.e., r [p (t)] = min[p˜ −p (t), r˜], where p (t)
is the unit’s power output, p˜ is the
generating capacity, and r˜ is the maximum reserve contribution determined by the
unit’s ramping capability}.
The reserve is zero if the unit is down, except for some specifically designated gas
turbines that can be started
very quickly. The system-wide spinning reserve constraints require that the total
reserve contribution should
be no less than a specified requirement Pr (t) at any time; in other words,
The minimum up (down) time constraints specify how long a unit must be
kept on (down) before it can be
shut down (started up). They are imposed on large thermal units because of heat
stress on the equipment, and
on hydro units to prevent frequent startups and shutdowns. Ramp rate constraints
specify a unit’s maximum
allowable generation change between two consecutive time instances, as described by
where # is the unit’s ramp rate. The ramp rate constraints couple the
unit’s generation levels over the
scheduling horizon and are difficult to deal with. In addition, some
units can be “must-run,” or “must-not-
run,” or must generate at predetermined levels during specified periods for
scheduled maintenance or testing
purposes.
The power output of a hydro unit is a function of the amount of water
discharge and reservoir head—the
elevation difference between the upstream and downstream water surfaces of a
reservoir (see Hydroelectric
power stations). A typical power output versus discharged water and head is shown
in Fig. 3. The minimum and
maximum water discharge, and thus the minimum and maximum power output, can also be
head-dependent.
Because the reservoir head is a function of the volume of water stored given the
shape of the reservoir bed,
the power output is thus generally modeled as a function of the volume of water
stored and water discharge.
Forbidden operating regions are common for many units (i.e., a unit may
not operate in certain regions or
may even be limited to operate at discrete points for safety or efficiency reasons).
Pumped-storage units have
characteristics similar to those of hydro units (see Hydroelectric power stations)
and can significantly contribute
to system reserve in both generating and pumping modes.
Scheduling hydro units is generally more difficult than scheduling thermal
units because the reservoirs
in a river catchment or water shed are hydraulically coupled. The water discharged
by an upstream unit affects
the levels of the downstream reservoirs. The water flow in a river catchment can be
described by the following
the water balance equation
where v (t), w (t), and s (t) are, respectively, the reservoir content, water
discharge, and spillage of hydro unitj
j j j
at time t, ζ (t) is the sum of natural inflows to reservoirj at t, τ is the time
required for water discharged fromj to
j j
reach its directly downstream reservoir, and U is the set of direct upstream
reservoirs of reservoirj . Therefore
j
the schedules of hydro units in a river catchment are coupled not only
through system-wide demand and
reserve requirements in Eqs. (4) and (5) but also the water balance in Eq. (8). The
hydro scheduling problem
can also be viewed as a resource allocation problem—there is limited
hydro energy in a cascaded reservoir
system, and the problem is to determine how to allocate the energy to maximize the
benefits or savings.
The water volume in a reservoir is also constrained by the reservoir
capacity and its minimum (or dead)
volume as
programming problem:
where Sti (t), Shj (t), and Spk (t) are, respectively, the startup costs
for thermal, hydro, and pumped-storage
units, subject to system-wide generation, reserve, and possibly transmission
constraints, water flow balance
constraints of each river catchment, and individual unit constraints. This
problem mathematically belongs
to the class of NP (Nonpolynomial) hard combinatorial optimization problems
(i.e., the computational time
required to find an optimal solution escalates exponentially with problem size).
Consistently obtaining optimal
schedules has been proven to be extremely difficult for systems of practical sizes.
Nevertheless, because the
cost to be minimized is the sum of individual unit costs and the system-wide demand
and reserve requirements
are also unit-wise additive, the problem is “separable,” and efficient near-optimal
approaches can be developed
as will be presented later.
For midterm and longterm scheduling problems, since many parameters
are stochastic in nature, un-
certainties may have to be considered. Examples include demand, unit
availability, and natural inflows of
reservoirs. For instance, if a possible flood is not appropriately considered, hydro
energy may be forced to spill
out during the flood. On other hand, if an unusual draught is not properly taken
into account, the hydro energy
may not be used when it is most needed. By dividing a year into several seasons, it
is possible to approximate
the demand and inflows as stationary series using Markov chains, and the problem is
to minimize the expected
value of the cost function in Eq. (10). The resulting solution is a “decision
policy” detailing what to do under
each possible circumstance, as opposed to a sequence of decisions for a
deterministic problem. The decision
policy can, in turn, provide targets or constraints for the short-term scheduling
problem.
Solution Methodologies
The approaches for solving hydrothermal scheduling problem can be classified into
five categories (2):
Fig. 5. The state transition diagram. For clarity, only the transitions from the
state “unit 2 up” are shown, and transitions
from other states are similar to those from state “unit 2 up.”
tree. Many methods can be used to solve problems associated with individual nodes.
One such method is to
allow all 0–1 commitment decision variables to take in real values between 0 and 1
for the problem associated
with the root node. Then, for each child of a node, gradually set
certain commitment variables to either 0
or 1. The optimization problem at each node is thus reduced to a standard linear or
nonlinear programming
problem that can be efficiently solved by using an appropriate continuous
variable optimization algorithm.
Many approximations and modifications have been developed to improve the efficiency
of this method.
Dynamic programming based on backward (or forward) searching of a state space
has been extensively
used in many areas. In its applications to hydrothermal scheduling, stages
correspond to hours, and states are
associated with unit commitment status. Figure 5 illustrates the states of two
consecutive stages for a simple
system with 3 thermal units without considering their minimum up and down times.
Calculating the optimal
fuel cost C [p ti (t)] for a given commitment status requires solving an economic
dispatch problem, typically a
constrained nonlinear programming problem, to produce the generation levels of all
units. The optimal “cost-to-
go” functions are then established for all possible states by working backward (or
forward) along the scheduling
horizon.
The optimal decision at each state is then obtained by traversing from the
initial (or terminal) state. It
can be seen that the number of states at each time unit for this simple
configuration is 2I , a very large number
if number of units I is large. When minimum up and down times come into the
picture, a single up state as
in Fig. 5 cannot fully represent the status of a unit because the unit may not be
able to shut down within the
next few hours. The number of states required to describe the problem will thus be
increased to MI , where
M is the sum of minimum up and down times. Dynamic programming can also be used to
include hydro and
pumped-storage units. Many constraints such as discontinuous operating regions and
even discrete generation
levels can be incorporated. However, because the continuous reservoir levels have
to be discretized, a very large
state space is required.
The idea of Bender’s partitioning or Bender’s decomposition is to
decompose a problem into a “master
problem” involving “complicating” (usually discrete) variables and a “slave
problem” involving other (usu-
ally continuous) variables (2). The slave problem typically corresponds to economic
dispatch with fixed unit
commitment status. The marginal costs obtained from the slave problem are used to
limit the commitment of
the units in the master problem, which can be further decomposed into a set of
subproblems associated with
individual units. The master problem then provides a new set of commitment
decisions. The two problems are
iteratively solved until the solutions converge.
For branch-and-bound, dynamic programming, and Bender’s decomposition, the
“curse of dimensionality”
is apparent when there are many units in the system, and these methods
can hardly be used to schedule
problems of practical sizes. Approximations and heuristics have to be
applied at the sacrifice of optimality.
One typical method to simplify the problem is to decompose it into two subproblems:
hydro subproblem and
thermal. The marginal costs obtained from thermal scheduling are used as a basis to
allocate hydro energy over
the hours. Then in thermal scheduling, hydro contributions are deducted from the
system demand and reserve
requirement. Hydrothermal coordination is realized by iteratively updating the
hydro and thermal solutions.
This iterative coordination scheme is heuristic in nature, thus optimality and
sometimes convergence cannot
be guaranteed.
Heuristic methods are still widely used by many electric utilities because of
their simplicity and compu-
tational efficiency. These methods are often developed based on the
knowledge of experienced operators for
specific systems. One method is the so-called priority list commitment. Thermal
units are first ranked based
on their full-load average costs, and units are committed in the ascending order of
the cost to meet system
demand and reserve requirements. Hydro energy is then allocated to
“shave” the peak of system demand
(having highest incremental costs). Manual adjustment is needed to satisfy
individual unit constraints. The
disadvantages of heuristics are that the methods are ad hoc for specific systems,
manual adjustment can be
very time consuming, and the schedule obtained may be far from optimal.
Lagrangian relaxation and its extensions have consistently been proven to
provide near-optimal solutions
in a computationally efficient manner and can be easily adapted to different problem
configurations (3,5,6,7,8,
10,11,12). They are thus among the most successful and most widely used methods and
deserve more attention
here.
Lagrangian relaxation has been fundamental for deriving necessary and
sufficient conditions for optimal-
ity. Over the decades, it has been the corner stone of resource allocation theory
in economics. In mathematics, it
has also shed much light on the development of efficient algorithms for constrained
optimization. The method is
particularly powerful for combinatorial optimization of “separable problems” where
the objective function and
system-wide coupling constraints are additive in terms of basic decision variables.
For the scheduling problem
under consideration, because the total cost to be minimized is the sum of
individual unit costs, and the demand
and reserve requirements are also unit/transaction-wise additive, the problem is
“separable,” and Lagrangian
relaxation can be effectively applied.
The key idea of the approach, in a nut shell, is decomposition and
coordination. “Hard” system demand
and reserve requirements are first “softened” or “relaxed” by having to pay a
“price” or “penalty” for constraint
violations. The per unit violation penalty at a particular hour is the
hour’s “Lagrange multipliers,” or the
“shadow prices” in the economics literature. Because the original problem is
separable, the “relaxed problem”
can be decomposed into many smaller subproblems, one for each unit or transaction.
Given a set of multipliers,
each unit determines its generation levels across the time horizon to
minimize its cost and subject to its
own constraints, and similarly for transactions. These subproblems are not
NP-hard, and can be efficiently
solved by using “dynamic programming.” The multipliers are then iteratively
adjusted based on the levels of
constraint violation following the market economy concept (i.e., increase the
multipliers for undergenerated
hours and reduce the multipliers for overgenerated hours). At the
termination of such multiplier-updating
iterations, system-wide constraints may still be violated at a few hours.
Simple heuristics are then applied
to adjust subproblem solutions to form a feasible schedule satisfying all
constraints. Because the value of the
dual problem is a lower bound of the optimal cost, the quality of the feasible
schedule can be quantitatively
evaluated.
where λ(t) and µ(t) are, respectively, the “Lagrange multipliers” associated with
system demand and reserve
requirements at time t. By using the duality theory (15) and the decomposable
structure of Eq. (11), a two-level
maximum–minimum optimization problem can be formed. Given the multipliers, the low
level consists of the
following subproblems:
Fig. 7. The state transition diagram for a thermal subproblem. The minimum down
time is assumed to be 3 hours, and
the minimum up time, 5 hours.
Let L∗ti [λ(t), µ(t)], L∗hj [λ(t), µ(t)], and L∗pk [λ(t), µ(t)] denote
the optimal sub-Lagrangians for (Pt − i),
(Ph −j ), and (Pp −k), respectively. Then the high-level dual problem is max
#[λ(t), µ(t)], with
Units with ramp rate constraints are harder to handle because generation
levels of two consecutive hours
are now coupled. Ramp rates can be handled through relaxation by introducing
another set of multipliers to
be updated at the intermediate level.
The hydro and pumped-storage subproblems are usually more difficult to solve
because of the hydraulic
coupling among the units within a river catchment. Network flow is an efficient
algorithm for hydro subprob-
lems. However, it cannot handle discontinuous operating regions (e.g., forbidden
regions) nor discrete operating
states (e.g., no generation). In this case, additional multipliers can be
used to relax hydraulic coupling and
updated at the intermediate level (16). Individual hydro subproblems can
then be solved by dynamic pro-
gramming similar to that for thermal subproblems, and a nonlinear network
flow can be applied when the
commitment status of hydro units is fixed.
The high-level dual problem is to update the multipliers so as to
maximize the dual function in Eq.
(15). Because discrete variables are involved at the low level, the dual function
may not be differentiable at
some points. The subgradient method is commonly used to update λ(t) and µ(t)
because of its simplicity and
performance (9,17,18):
where
where C is the total generation cost of a feasible schedule as defined in Eq. (10)
can be used to quantitatively
measure the solution quality. If ε is small, the schedule is near optimal because
the optimal cost C∗ must lie
between any feasible cost and the dual cost; in other words,
Conclusions
problem will not disappear, but will be transformed into a more complicated bidding
and resource allocation
problem. The results and insights obtained by researchers and practitioners over
the decades on hydrothermal
scheduling have established a solid foundation for us to tackle the new
issues confronting the deregulated
industry.
BIBLIOGRAPHY
XIAOHONG GUAN
PETER B. LUH
ERNAN NI
University of Connecticut
RENHOU LI
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3028.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Magnetohydrodynamic Power Plants
● Email this page Enter words or
Standard Article
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3028.htm17.06.2008 16:42:36
Gas
flow
Figure
2. Illustration of MHD generator concept in showing electri-
cal
conducting gas moving through magnetic flux lines to generate an
electric field. Note that electrodes must be placed in contact with the
gas to
permit the current to flow through an external circuit.
case,
as illustrated in Fig. 2, the electrical field is generated
in
the conducting gas which is passing through a magnetic
field.
In order to utilize this power, electrodes must be in
cathode emits electrons and the anode collects them, the same
applied primarily to
combustion products which are utilized
N N once
and then exhausted to the atmosphere. Thus, it is classi-
fied as
open-cycle MHD.
To
produce the plasma for open-cycle MHD power genera-
tion,
a fossil fuel such as coal, char, oil, or gas is burned at a
S S high
enough temperature to produce the required ionization.
This
temperature required for ionization is too high for a
S
of about 2800 K. To achieve this high a flame temperature
( P = 9 atm
y 10
t
with fossil fuel requires significant preheating of the combus-
i
i
tion air or oxygen enrichment of the air or some combination
t
u
of the two techniques.
d 8
n
Fuels that have been considered for MHD power plants
o
c 6
i
coke. Fuels that have a high carbon-to-hydrogen ratio produce
r
c
better electrical conductivity at the same conditions. This is
e
l
because the water that is formed when hydrogen burns disas-
E 4
(a)
gen, such as natural gas, can be used but require a few tens
of degrees Kelvin higher temperature to achieve the same
2850
electrical conductivity. Calculated electrical conductivity and
flame temperature for coal with 1% potassium added as po-
P = 5 atm
tassium carbonate is shown in Fig. 3(a) and (b), respectively,
2800
P = 7 atm
for a range of air preheat temperatures and pressures.
These )
K
calculations are all for a fuel-rich condition, 85% of theoretical
( P = 9 atm
e 2750
r
oxygen. The reasons for choosing this fuel-rich condition are
u
a
twofold: It optimizes the electrical conductivity, and it is es-
r
p 2700
sential for the control of nitrogen oxide emissions to a very
m
e
low level.
t
e
The actual configuration of the MHD generator, in its sim-
m 2650
a
plest form, is shown in Fig. 4(a). This arrangement is called
l
F
a continuous electrode Faraday generator because the elec-
2600
trodes are continuous along the generator length. This con-
figuration is used in some short-duty time applications, such
as the Russian PAMIR, which generates a pulse of several
2550
Anode less
than 100 kPa, the conductivity has decreased to around
1 S/m
and MHD generation is no longer attractive. However,
the
working gas still contains considerable sensible energy.
large
amounts (10 kppm to 12 kppm) in the high-temperature
Load
combustors of MHD plants, even at fuel-rich stoichiometries.
I
x be
made in a way that avoids much temperature rise to avoid
J × B u Conducting sidewalls lying along the
formation of additional nitrogen oxides.
z
B equipotential planes which are
The MHD plant has a higher particle loading in the ex-
perpendicular to net electric field haust
because of the potassium seed added. Furthermore, the
J size
distribution tends to be smaller due to the higher temper-
(c)
atures having vaporized some of the solids. However, the re-
MHP power
Alternating
current Inverter
current
Direct
Electric power
Generator current
Coal handling
Air heater
Radiant
Regenerated MHD channel
boiler ESP
seed Combustor
Superheater
Magnet
Secondary Particles
Slag
combustor Steam Air
dump
Air Comp
Electric
power
Figure 5. Schematic of
MHD/steam combined cycle power plant.
3
THE MHD TOPPING CYCLE
typically over 120 MW/m . The walls of the combustor rapidly
become covered with coal slag, solidified next to the wall and
The MHD topping cycle consists of a means of generating the
molten on the flame side. In spite of insulation effects of this
plasma (generally a combustor), a nozzle to accelerate the
slag layer, heat fluxes to the walls average about 100 W/cm2 .
eral streams that are integrated with the bottoming cycle in-
chamber.
cluding air preheat and/or oxygen production, recovery and
The advantages of this type of combustor are simplicity,
reprocessing of the potassium seed, and use of boiler feedwa-
low cost, and higher, more uniform electrical conductivity be-
ter for cooling of all topping cycle components.
cause of the time available for the seed to reach chemical and
tube
the heat losses to the walls. There have been two successful
coal combustors demonstrated that differ substantially
in Oxidizer
design.
jet ports
Oil
One of these is a single-stage, full-slag-carryover combus-
Disperser nozzle
tor modeled somewhat after rocket engine technology (4). A
plate
sketch of this combustor is shown in Fig. 6. In this design,
the pulverized coal is injected in a conical pattern from the
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
###################################################################################
############################yyyyy
Refractory
middle of the injector plate. Air is fed through a number of
small holes in the injector plate where a pressure drop is
Figure 6. Single stage MHD coal combustor concept. This design is
##yy#######y#y#y#y#y#y####y#y#y
taken to accelerate the air. These jets of air mix very rapidly
optimized for rapid turbulent mixing and for minimum heat loss to
with the coal and produce high heat release concentrations,
the walls.
density, and the distance over which it acts. From this view-
and additional air are added in the second stage. The objec-
point
alone, the velocity should be high. Other considerations
tive of this design is to remove as much coal slag as possible
such as
convective heat transfer coefficient, size of generator,
before adding the potassium seed to minimize the amount of
and
total amount of power that can be generated from a unit
potassium seed that chemically combines with the slag and
of
fuel, however, place upper limits on the desired entrance
may be more difficult to recover. Slag removal efficiencies of
The MHD
Generator
2880 11.6
Analysis and characterization of the MHD generator involves
the
consideration of both Maxwell’s equations for electromag-
netic
fields and the conservation of mass, energy, and momen-
2870 11.4 tum
equations for fluid flow (7). In addition, of course, an
T l
equations must be solved numerically for analysis or design
a
c
i
10.6 r of the
MHD generator. In cases where the interaction is low,
2830 t
c
e it may
be a satisfactory approximation to neglect the effect of
l
E
the
body force, J
B , in the flow calculations. In those cases
2820 10.4
where
the body force is neglected in the calculation of flow
Body force
ρuA = m˙ (3)
F = J × B where
A # cross-sectional area of duct and
Ohm’s
Law (including only significant terms for this case):
Conservation
Electric Maxwell's of mass, Fluid
#
and momentum and flow
J E u B E J B
magnetic equations equations
JJ = σ (EE + uu ×BB + EE d ) − B (JJ ×BB ) (4)
energy plus
BB
fields
equation of state
where
Ed is an electrical field loss due to plasma to electrode
E V
d d
# =
=
uB uBd
Ohm’s # =
Hall parameter = electron cyclotron frequency times
law
mean time between collision of electron with
heavy particles
J = (u × B × E )
σ
From
Ohm’s law, an equation for J #E can be derived for use
J is current density
in the
energy equation (10)
B is magnetic field flux density
F is force
J 2
u is velocity
J E − J
JJ ·EE = y B (1 − #) (5)
E is electrical field
σ
σ is electrical conductivity
An
equation of state is required. That is, given two thermody-
Figure 8. Illustration of coupling of the gas-dynamic and electro-
namic
variables such as T and P , a relationship is required
magnetic relations.
such
that the other thermodynamic variables can be calcu-
lated.
This is typically a table of values calculated from a
generator. Either case then gives all the electrical and ther-
# # density of the plasma
R V
K = L or L
where q # heat flux to walls.
RL + Rg VG
w
where a
diverging duct, with (a) the wall angles chosen to be small
P − P
ated power is dissipated by the current flowing through the
Cp = 2 1
1 2
gas. If a higher K is chosen, the generator becomes longer and
2 ρu
more efficient. The correct choice, in general, will depend on
the overall plant economics in terms of levelized cost of elec-
where
tricity being minimized. The MHD generator tends to be rela-
tively cheap, but making the magnet longer adds considerable P2
is static pressure at the outlet
expense. A loading factor, K , of 0.88 to 0.90 is typical. P1
is static presure at the inlet
For a given load factor and gas conditions, the amount of #
is fluid density at the inlet
power that can be generated depends on the pressure ratio u
is fluid velocity at the inlet
across the generator. The exit pressure must be determined
by the conditions needed at the diffuser entrance in order to The
coefficient of performance for practical diffusers in MHD
recover to the pressure needed at the furnace inlet (it may be
topping cycles are expected to be 0.5 to 0.6. Heat loss in the
slightly less than atmospheric for induced draft designs).
diffuser walls is incorporated into the boiler feedwater heat-
Some common efficiency definitions that apply to the MHD ing
chain and, in some aggressive designs, may be used for
generator are given below
boiling surface.
The
first difference from a conventional steam plant is that
Note that since the only energy loss from the gas is electrical the
hot gases enter from a single duct rather than distributed
power output and heat transfer to the walls, we obtain
burners. The gas is at a higher temperature (2000 K to 2200
K,
3140#F to 3500#F) and is reducing. The gas is reducing
Generator efficiency = Electrical power output
because the gases exiting the MHD cycle must be cooled be-
Decrease in enthalpy across generator fore
completing the combustion. At this point they contain
P0
carbon monoxide, hydrogen, and hydrogen sulfide, which are
=
(h0 − h0 )m˙
known to be corrosive to power plant materials. The basic
in out
99.9 99.5%
removal efficiency. With such a fine particulate, a ma-
Data jor
problem with greatly improved efficiency is reintrainment
99.5 of small
particles when the plates are rapped to remove the
99 Spline
95
collected particles. An attractive alternative to the dry ESP
90 seems
to be the wet ESP, in which the plates are washed
% 75
continuously with water. A baghouse is also a feasible choice.
e The bag
fabric must be chosen to avoid blinding. A Gore-Tex
v
i
t 50 fabric,
manufactured by W. L. Gore & Co., Knoxville, TN, was
a
l
u used
successfully in the pilot plant.
m
u If a
dry ESP is used, it has been suggested that if should
C
10 0.81 MMD be
designed to operate at a high enough gas temperature so
5 that it
can be placed before the economizer/low temperature
1 air
heater surface. Since the temperature difference between
0.5 flue gas
and feedwater or air is low in this region, a large
0.1 amount
of convective surface is required and substantial sav-
0.1 1 10 50 100 ings
result from having a cleaner gas and lower fouling factor.
Diameter ( m)
µ
desirable range for good ESP operation and tends to partially tassium
formate for reuse. The recovered potassium carbon-
compensate for the very heavy mass loading and small size. ate can
be separated from the sulfate and chloride by differ-
A four-field dry ESP with a design specific collection area ences in
their solubility in water. Separation of sulfate from
(SCA) of 475 ft2/ 1000 acfm performed in the pilot plant with
carbonate was performed well in the pilot plant program by a
1011
Illinois #6
1010
)
m
c
Ω- 109
(
y
t
i
v
i
t 108
s
i
s
e
R
107
106
Figure 10. Ash resistivity from MHD pi- 2.6 2.4
2.2 2.0 1.8 1.6 1.4 1000/K
lot plant as a function of temperatures. 111 144
181 227 282 352 441 °C
Resistivities from 109 to 1010 # # cm are de- 233 290
359 440 540 665 826 °F
MAGNETORESISTANCE 269
rotary vacuum filter working on the water from the wet ESP.
17. Technical assessment guide (TAG), EPRI Report No. P-2410-SR,
The system was saturated with carbonate, and the sulfate
1981 ed., Palo Alto, CA: The Electric Power Research Institute,
was removed by the filter.
1982.
Institute
BIBLIOGRAPHY
4. R. C. Attig et al., Design report for the combustor for the low
mass flow coal fired flow facility, Report No. DR-2.1.2-79-02, Tul-
lahoma, TN: The University of Tennessee Space Institute, 1979.
12. Anonymous, Steam: Its Generation and Use, 39th ed., New York:
Babcock & Wilcox, 1978.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W7212.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Multirate Filterbanks
Abstract
Introduction
Summary
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W7212.htm17.06.2008 16:43:00
MULTIRATE FILTERBANKS 35
a a s
(a)
(b)
signal processing.
tion) by factor M . By doing so, the new rate for x(Mn) is 1/M
MULTIRATE FILTERBANKS
x(n) with the data rate the same as x(n) by inserting M
1
ˆ
been providing higher quality services than before, making
D/A converter. The analog signal xa(t) can be mathematically
the information highway possible, and revolutionizing our liv-
represented as
ing standards. In this section, we briefly describe concepts of
analog-to-digital (A/D) and digital-to-analog (D/A) converters,
xˆa (t) = #x (n) sin π(f st − n) (2)
n
as transmultiplexing and multiresolution image analysis and
coding, and finally the outline of this article.
ˆ
INTRODUCTION
Fig. 4, where the data rate of x(n) is N/M times of the data
(TDMA), where different users use different time slots; (2) fre-
mation that the original analog signal xa(t) has, it is necessary
ers use different frequency slots; and (3) code division multi-
f s not to be below the Nyquist frequency, i.e., twice of the
bandwidth (Hz) of the analog signal xa(t):
ple access (CDMA), where different users use different codes.
x (n) x
fs fs
–
Analog Analog
2 2 (b)
signal Analog to Digital signal Digital to signal
digital (A/D) processing analog (D/A)
(a)
converter (DSP) converter
36 MULTIRATE FILTERBANKS
^
and reconstruction are built upon tree-structured multirate
x (n) x (n)
N M
filterbanks as illustrated in Fig. 7(a), which are used along
From the two multirate filtering applications, one can see the
Channel
^
x0 (n)
x0 (n)
F (z) Ho (z) M
M
o
^
x 1 (n)
x 1 (n)
H (Z)
F (z) H (z)
1 M
M
1 1
⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅
⋅ ⋅ ⋅ ⋅ ⋅
^
xp –1 (n)
xp –1 (n)
F + H (z)
(z) Fp –1 (Z)
Figure 5. Transmultiplexer which is used in many M
p –1 p –1 M
communication systems including multiuser systems.
MULTIRATE FILTERBANKS 37
2π /M
⋅ ⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅
1 2 M t 1 2 M ω
2π ω Figure 6. Multi-access characteristics having
j #
denote scalar values and scalar-valued sequences; capital let-
and X (e ) is its discrete time Fourier transform (DTFT).
x (n)
H0 (z) 2 H0 (z) 2 ⋅ ⋅ ⋅
H0(z) 2
H 1(z) 2
H 1(z) 2
H 1(z) 2
^
x (n)
2 F0 (z) + ⋅ ⋅ ⋅ 2 F0 (z) +
2 F0 (z) +
2 F 1(z) 2 F 1(z)
2 F 1(z)
(a)
^
x (n)
x (n)
H0 (z) 2 2 F0 (z) +
38 MULTIRATE FILTERBANKS
An
M-fold decimator (or downsampling) and L-fold expander
H (z ) = # H z −n
(or upsampling) are depicted in Fig. 11(a) with an example
n
n in
Fig. 11(b), and Fig. 11(c) with an example in Fig. 11(d),
respectively, where
where Hn are constant matrices with same size. For a square
matrix polynomial H(z), det(H(z)) denotes its determinant.
yD (n) = x (Mn )
T †
For a matrix A, matrices A , A , and A* denote the transpose,
transpose conjugate, and conjugate, respectively. For a matrix
†
˜
polynomial H(z), its tilde operation H(z) denotes H (1/z*), that
is,
0
˜ # † n # −n
H (z ) = H z , if H (z ) = H z (3)
n n
n n
x (n) yD (n)
M
A matrix polynomial H(z) has FIR inverse if and only if there
M = 2
exists a matrix polynomial F(z) such that F(z)H(z) # I, where
I is the identity matrix.
(a)
0
WM = (Wmn )0≤m ,n ≤M −1
M
where Wm # e
j 2#/M .
(b)
x (n) yE (n)
(c)
L = 2
(d)
MULTIRATE FILTERBANKS 39
j ω j
ω
Y (e ) Y (e
)
E D
j ω M = 2
X (e ) L = 2
1
1 1
2
–π 0 π ω –π 0 π ω –π 0
π ω
Figure 12.
The frequency domain representation ex-
(a) (b) (c)
amples of the decimator and expander.
and
Polyphase Representations
#
The polyphase representation was first invented by Bellanger
x[n/L], if n is a multiple of L ,
yE (n) =
et al. (14) and Vary (15) and first recognized by Vaidyanathan
0, otherwise.
and Vetterli in the simplifications of multirate filterbank the-
M −1
# #
j ω −j ωn 1 j (ω−2π k )/M
N −1
Y (e ) = y (n)e = X (e ) (4)
# −l N
D D M
H (z ) = z E (z ) (5)
l
n k =0
l =0
and
where
# −n L j ω j ωL
Y (z ) = y (n)z = X (z ), and Y (e ) = X (e )
#
E E E
n
E (z ) = h[Nn + l]z −n
j # j #/M
N −1
#
1. Stretch X (e ) by a factor M to obtain X (e );
−N + 1+l N
H (z ) = z R (z ) (6)
2. Obtain M
1 copies of this stretched version by shifting
l
l =0
it uniformly in successive amounts of 2#;
3. Add all these shifted and stretched versions to the origi-
where Rl(z) # EN
1
l(z), which is called the Type 2 polyphase
j #/M ), and divide
nal unshifted and stretched version X (e
representation of H(z). For l # 0, 1, . . ., N
1, El(z) and
by M .
Rl(z) are called the lth Type 1 and Type 2 polyphase compo-
a multirate filterbank.
Noble Identities
x (n) x (n)
y 2 (n)
K H (z) H (zk) K
(a)
40 MULTIRATE FILTERBANKS
x (n)
x (n)
H (z) M N F (z)
H (z) M M F (z)
0 0 0 0
0 0
H (z) M 1 N F (z)
H (z) M F (z)
1 1 1
1 M 1
⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅ ⋅
^
^
x
(n)
x (n)
HM–1 (z) MM–1 NM–1 FM–1 (z) +
HM–1 (z) M M FM–1 (z) +
3. when M0 # M1 # # # # # MM
1 # N0 # N 1 # # # # # filterbank is called FIR. Examples of 2-
channel PR filterbanks
NM
1 M, the filterbank is called over decimated; were
first obtained by Smith and Barnwell (28) and Mintzer
4. when Mk and Nl are not all equal, the filterbank is
(29) independently. Some early studies on multirate filter-
called nonuniformally decimated.
banks can also be found in Refs. 29a–c.
X (z ) = cz X (z ) (7)
lately, such as applications of nonmaximally decimated
ˆ # l
on the first case, that is, maximally decimated multirate fil-
X (z ) = A 0 (z )X (z ) + A (z )X (zWM ) (8)
l
terbanks.
l = 1
H0 H 1 H2 HM – 1
⋅ ⋅ ⋅
Figure 15. M-channel analysis filter frequency
response example.
0 2π
ω
where
ter Fm(z) in Fig. 16, let Rl,m(z) be its lth Type 2 polyphase
component, for 0 # m, k, l # M
1. Let
M −1
#
1
A (z ) = H (zW l )F (z ), 0 ≤ l ≤ M − 1 (9)
l M k M k
E (z ) = (Em ,k (z ))0≤m ,k ≤M −1, and R (z ) = (Rl ,m (z ))0≤l ,m ≤M −1
k =0
Clearly, the second term in the right hand side of Eq. (8) is
which are called the polyphase matrices of the analysis bank
the aliasing term. For the PR property Eq. (7), we need
and the synthesis bank in Fig. 16, respectively. Then, it is not
M ˜ M
where h(z) # (H T
1
let
0(z), . . ., HM
1(z)) , e(z) # (1, z , . . .,
M #1 T T
0 M
1
t (z ) = [cz−n0 , 0, . . ., 0]T
tion of e(z). Thus, by using the Noble identities, the multirate
T
filterbank in Fig. 16 is the same as the one shown in Fig.
f (z ) = [F (z ), F (z ), . . .,F (z )]
0 1 M −1
17(a), which is called the polyphase representation of the
H (z )f (z ) = Mt (z ) (12)
Theorem 1 deals with the AC matrix H(z) while Theorem
x (n)
Theorem 1. An FIR multirate filterbank in Fig. 16 is perfect
M M
reconstruction if and only if its AC matrix H(z) has FIR in-
z–1 z–1
verse.
–1 M M –1
z z
E (z) R (z)
⋅ ⋅ ⋅
One can see that the AC matrix H(z) is a structured ma-
⋅ ⋅ ⋅
trix, where its components are not free but related. This lim-
⋅ ⋅ ⋅
–1
–1 z
its the study and construction of PR multirate filterbanks. We
z ^
x (n)
next want to use the polyphase representations and Noble
M M +
identities introduced in the second section and convert the
AC matrix to the polyphase matrix in which all components
Analysis bank Synthesis bank
are free.
(a)
M M
The analysis in the previous subsection is a direct analysis of
z–1 z–1
the relationship between the input and the output in Fig. 16.
We next want to first simplify the block diagram in Fig. 16 by
z–1 M M z–1
R (z) E (z)
using some properties of building blocks studied in the second
⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅
M M +
ders of decimator/expander and FIR filters, and then convert
the multirate filterbank into a multi-input and multi-output
(b)
(MIMO) system.
For each analysis filter Hm(z) in Fig. 16, let Em,k(z) be its
Figure 17. Polyphase representation of M-channel maximally deci-
kth Type 1 polyphase component, and for each synthesis fil-
mated multirate filterbank.
42 MULTIRATE FILTERBANKS
n
construction and factorization discussed later.
only if its determinant is cz n for a nonzero constant c and an
In some applications, such as the cross-talk cancellation in
integer n0.
transmultiplexers in Fig. 5, PR may not be necessary as long
as the aliasing (cross-talk) is cancelled in a multirate fil-
terbank, i.e., the second term at the right hand side of Eq. (8)
Let H(z) be an FIR causal matrix polynomial with an FIR
is zero. A necessary and sufficient condition on such fil-
inverse. If det(H(z)) # cz
# , then # is called its McMillan de-
torization of
FIR matrix polynomials with FIR inverses, let
P (z ) P (z ) · · · P (z )
us first introduce three types of elementary row (column) op-
0 M −1 1
erations:
z −1P (z ) P (z ) · · · z −1P (z )
1 0 2
. . . .
P (z ) = . . . .
. . . .
Type 1. Interchange two rows (or columns).
−1 −1
z P (z ) z P (z ) · · · P (z )
−1 M −2 −1 M −3 M −1
Type 2. Multiply a row (or column) with a nonzero con-
z P (z ) z P (z ) · · · P (z )
M −1 M −2 0
stant c.
i
+ · · · + z −M + 1PM −1 (zM ))
†
FACTORIZATION AND CONSTRUCTION
A = I + (e − e )(e − e ) ,
i j j i
i i
L
E (z ) = #E z −n ,L is a nonnegative integer
for certain c #
1 and a certain i, 1 # i # M .
n
n=0
A Type 3 elementary matrix U(z) can be written as
i j
of analysis banks have FIR inverses, the corresponding syn-
thesis banks can be obtained by using the inverses for the PR
multirate filterbanks, i.e., PR FIR multirate filterbanks are
where #(z) is a polynomial of z
1 and i # j with 1 # i, j # M .
MULTIRATE FILTERBANKS 43
H (z ) = Vρ (z ) · · ·V (z )Aσ Uσ (z ) · · ·A U (z ) (16)
γ0 (z ) 0 · · · 0 0 · · · 0
1 1 1
0 γ (z ) · · · 0 0 · · · 0
1
where # is the McMillan degree of H(z),
is a certain nonneg-
. . . . . . .
. . . . . . .
ative integer, V (z) # O for i # 1, 2, . . ., #, A # A
and
i i
. . . . . . .
U (z) # U for i # 1, 2, . . .,
.
H (z ) = W (z ) 0 0 · · · γ (z ) 0 · · · 0 U
(z ) i
p
0 0 · · · 0 0 · · · 0
In this subsection, we
introduce paraunitary matrix polyno-
0 0 · · · 0 0 · · · 0
mials and corresponding multirate filterbanks, which are spe-
and only if
where W(z) and U(z) are products of some elementary matrix
polynomials with sizes M #M and N # N , respectively, #i(z)
˜
are polynomials of z
1, # (z) divides # (z), for i # 0, 1, . . .,
H (z )H (z ) = dIN , for all complex values z
i i #1
p
1, i.e.,
#i+ 1 (z )
H† (ej ω )H (ej ω ) = dIN , for all real values ω
γ (z ) =
i # (z )
i
j #
equivalent to paraunitary. When H(z) exists for all z # e
but
where (z) # 1, (z) for i 0 is the greatest common divisor
0 i
not all complex values z , lossless is not equivalent to parauni-
of all the i # i minors of H(z).
n
n polynomials in this subsection. An example of paraunitary
composition has the form of diag(c z 1, . . ., c z M )
with
1 M
matrix polynomial is:
nm # 0 and cm # 0 for 1 # m # M, and this is a complete
characterization of all square causal matrix polynomials with
# −1 + 1 z −1 − 1#
z
FIR inverses. The factorization in Eq. (15) is, however, not
H (z ) = −1 −1
convenient to be incorporated in the optimal design studied
z − 1 z + 1
# #
˜ z + 1 z − 1
H (z ) =
Class I.
z − 1 z + 1
O # V(z): V(z) # I
vv† # z
1vv† where v is an M # 1
H0 (z) M M H0(1/z )
Let U(z) # I # #z
me e† # U . Then, its inverse U
1(z) #
i j
I
#z
m †
e e .
i j
M M H * *
Class III.
H 1 (z) 1(1/z )
A # A : A # I # (e
e )(e
e )† for certain i # j and 1
i j j i
†
# i, j # M or A # I # ce e for certain c #
1 and a ⋅ ⋅ ⋅
⋅ ⋅
i i
⋅ ⋅ ⋅ ⋅ ⋅
certain i, 1 # i # M
. ⋅ ⋅
⋅ ⋅ ⋅
Let A # I # (e
e )(e
e )† # A . Then, its inverse
^
i j j i
x (n)
1 †
1 H (z) M M H * *
+
A # A . Let A # I # ce e # A . Then, its inverse A
# M–1 (1/z )
i i
M–1
I
c/(c #1)e †
e .
i i
Analysis bank Synthesis bank
44 MULTIRATE FILTERBANKS
4 0.3537539 0.1864025
0.3537496
the following, we want to present a complete characterization
5 0.2672266
0.0000020 0.2672007
of all paraunitary matrix polynomials obtained by Vaidyana-
6 0.0870758
0.3543303
0.0870508
than (9,33–35).
7
0.0521155
0.0000363
0.0520909
8
0.0875973 0.3564594 0.0875756
Theorem 5. An M # M causal FIR matrix polynomial H(z)
9
0.0427096
0.0000049
0.0427067
is paraunitary if and only if it can be factorized as
10 0.0474530
0.1931082
0.0474452
13
0.0232765
0.0000026
0.0232749
where d is a positive constant, # is the McMillan degree of
14 0.0000022 0.0 0.0000022
H(z), H is an M # M unitary constant matrix, V (z) # O for
0 i
filters have good filter properties, i.e., with good passband and
M −1
factorization is simplified as the following lattice representa-
# j ω 2
min |Hm (e ) | dω (18)
tion (9).
m=0 mth stopband
1 0 0 ±1
zation, the optimized 3 analysis filter impulse responses are
− sin θ cos θ 0 z −1
i i
α –c –c –c
x (n)
0 1 ρ
–s –s –s
z–1
0 1 ρ
s0 s 1 s
ρ
+ c0 c 1 c
analysis bank.
–α ρ
MULTIRATE FILTERBANKS 45
0 1
0
PM
1(z)).
jω jω jω
H (e ) H (e ) H (e )
0 1 2
By the study in the third section, it is clear that the DFT
–10
–20
Pm (z ) = cmz −nm , cm =# 0, 0 ≤ m ≤ M − 1,
–30
for some integer nm
–40
or
–50
M −1
–60
P (z ) = # cmz −Mnm −l , cm =# 0 (20)
Normalized frequency ( /2 )
ω π
22, which is the reason for the name of the DFT filterbanks.
One can see that the condition on the prototype filter P(z)
cation systems, it is however quite often that all M analysis in
Eq. (20) for the PR property is very restrictive and usually
filters are derived from a single prototype filter and therefore
limits their applications. There are three ways to get around
have the same shape of their Fourier spectra. The advantage
this condition. The first one, which is also the most intuitive
of such systems is the implementation simplicity. In this sec-
one, is to design P(z) with excellent lowpass property. Then,
tion, we introduce two kinds of such filterbanks. One is the
the DFT filterbank is almost PR because the whole frequency
discrete Fourier transform (DFT) filterbank, where analysis
band is almost divided with a wall-cut manner by M analysis
filters are single-sided shifts of a prototype filter in the fre-
filters. The second way is to use nonmaximally decimated
quency domain (or exponential modulation). The other is the
DFT filterbanks, that is, the decimation factor is less than the
cosine modulated filterbank, where analysis filters are dou-
number of channels (or users), which corresponds to
ble-sided shifts of a prototype filter.
oversampled short-time Fourier transforms or discrete Gabor
n=0
modulations to construct M analysis filters with real coeffi-
M M
n=0
and for 0 # m # M
1,
mn H (z ) = a U (z ) + a V (z )
that M analysis filter coefficients hm(n) # WM p (n) are no
m m m m m
longer real even when the prototype filter coefficients p (n)
L −1
are real.
= # 2 real(amW −(m+0.5)n )p (n)z −n
2M
Let P (z), 0 # l # M
1, be the Type 1 polyphase compo- n=0
l
E (z ) = W† diag(P (z ), P (z ), . . .,P (z ))
am = W (m+0.5)(L −1)/2+ (−1)mπ /4
M 0 1 M −1
2M
46 MULTIRATE FILTERBANKS
x
(n) x (n)
M P0(z) M
–1
–1
z
z
–1
M P 1(z) z–1 M
z
WM P (z) WM
⋅
⋅ ⋅ ⋅
⋅
⋅ ⋅ ⋅
⋅
⋅ ⋅ ⋅
–1
–1
z
M PM–1 (z) z M
terbank with the prototype filter in Fig. 23(a). One can see
ficient condition for the PR property (9,10,12,43).
terbanks are:
phase components of a prototype filter P(z) of length L with
2M total components. If L # 2KM # 1 for some positive inte-
Pm+M (z )
x (n)
M M
Linear-Phase Perfect Reconstruction Filterbanks
–1 –1
z z
In some applications, such as image/video processing, it is de-
WM W *
multirate filterbank factorization and construction. Linear-
M M M
⋅ ⋅ ⋅ ⋅
phase multirate filterbanks were first studied independently
⋅ ⋅ ⋅ ⋅
⋅ ⋅ ⋅ ⋅
by Nguyen and Vaidyanathan (49,50), and Vetterli and Le
Gall (51). For more details, see, for example Refs. 7, 9, 11, 12,
–1 –1
z z
45, 49–54.
M M +
^
x (n)
MULTIRATE FILTERBANKS 47
0 0
–10 –10
) )
B –20 B –20
d d
( (
e –30 e –30
s s
n n
o –40 o –40
p p
s s
e e
r –50 r –50
e e
d –60 d –60
u u
t t
i i
n n
g –70 g –70
a a
M –80 M –80
–90 –90
–100 –100
0 0.1 0.2 0.3 0.4 0.5 0
0.1 0.2 0.3 0.4 0.5
Normalized frequency
Normalized frequency
JPEG, MPEG1 and MPEG2, H.261 and H.263 in blocked DCT and 80.
For more properties on multidimensional basic build-
transform coding, one first decomposes an image on a rectan- ing
blocks, such as decimator and expander, delay chain sys-
gular region into 8 # 8 blocks, and then implement 8 # 8 tems,
and constructions, see for example, Refs. 67–85, 85a.
DCT on each block. Then, one quantizes the blocked DCT co- For
multidimensional cosine modulated filterbanks, see Refs.
efficients to get compression gain. Due to the truncation of an 46–48.
For multidimensional LOT, see Refs. 64, 65, 86, 86a.
image in the block decomposition, the blocking effects degrade
the compression performance at low bit rates. The blocking
effects basically come from a hard truncation between blocks, Optimal
Multirate Filterbanks in Quantization
the
complex field, i.e., all signals and filter coefficients are
Multidimensional Multirate Filterbanks
complex
values. These values can certainly be in any finite
Multirate filterbanks have been used in multidimensional sig- field. Due
to the incompleteness of a finite field, it is impossi-
nal processing, such as image/video processing. One way to ble to
factorize all FIR PR filterbanks over finite fields. A
use multirate filterbanks is to use one-dimensional fil- counter
example was given in Ref. 94. Although this is the
terbanks at each dimension of signals. Another way to use case, a
completely study on LOT (order one multirate fil-
multirate filterbanks is to use multidimensional filterbanks terbanks)
over finite fields was obtained in Ref. 61. It is found
directly to multidimensional signals, which has the potential that
nonmaximally decimated multirate filterbanks over fi-
to take advantage of multidimensional signals themselves, nite
fields are the same as convolutional codes for the error
such as nonhorizontal or nonvertical lines in image analysis. control
purpose (31,32,61,95). For filterbanks over finite
This motivates the study of nonseparable multidimensional fields,
see also Refs. 96, 97.
multirate filterbanks. It is known that all one-dimensional
paraunitary filterbanks can be decomposed into degree-one
48 MULTIRATE FILTERBANKS
2.
M. Vetterli, Perfect transmultiplexers, Proc. IEEE Int. Conf.
Nonmaximally Decimated Multirate Filterbanks
Acoust. Speech Signal Process., pp. 2567–2570, Tokyo, Japan,
1986.
Nonmaximally decimated multirate filterbanks have been re-
cently applied as precoders in communication systems, such 3.
R. D. Koilpillai, T. Q. Nguyen, and P. P. Vaidyanathan, Some
as intersymbol interference (ISI) cancellation (16–21,98,99)
results in the theory of cross-talk free transmultiplexers, IEEE
1993.
5.
A. Said and W. A. Pearlman, A new and efficient image codec
Time-Varying, Vector Filterbanks and Multirate
based on set partitioning in hierarchical trees, IEEE Trans. Cir-
Filterbanks with Block Sampling
cuits Syst. Video Tech., June 1996.
6.
O. Rioul and M. Vetterli, Wavelets and signal processing, IEEE
Because of the nonstationarity of signals, such as speech,
Signal Process. Magazine, 8: 14–38, 1991.
audio, and images, time-varying filter banks have been stud-
7.
M. Vetterli and C. Herley, Wavelets and filter banks: Theory
ied to exploit the nonstationarity, where filter banks change
and design, IEEE Trans. Signal Process., SP-40: 2207–2232,
dynamically with the nature of a signal. An important prob-
1992.
lem in time-varying filter bank theory is how to construct 8.
A. Akansu and R. Haddad, Multiresolution Signal Decomposi-
time-varying filter banks with PR property from local PR fil-
tion: Transforms, Subbands, and Wavelets, Boston: Academic
ter banks. For this problem, there have been several discus-
Press, 1992.
sions (60,102–114). 9.
P. P. Vaidyanathan, Multirate Systems and Filter Banks, Engle-
ley, 1994.
Conventional filter banks are all linear transformations. It is
˘ ´
11.
M. Vetterli and J. Kovacevic, Wavelets and Subband Coding,
well-known that all linear transformations may smear edges
Englewood Cliffs, NJ: Prentice-Hall, 1995.
that are important in image/video processing. Since nonlinear 12.
G. Strang and T. Q. Nguyen, Wavelets and Filter Banks, Welles-
transformations, such as median filtering, may preserve
ley-Cambridge Press, 1996.
edges better than linear transformations do, nonlinear filter
12a. P. P. Vaidyanathan et al., Improved technique for design of
banks have been recently studied (115–118).
perfect reconstruction FIR QMF banks with lossless polyphase
1042–1056, 1989.
SUMMARY 13.
A. N. Akansu and M. J. T. Smith, Subband and Wavelet Trans-
MULTIRATE FILTERBANKS 49
26. J.-J. Lee and B. G. Lee, A design of nonuniform cosine modu- 42a.
S. Qian and D. Chen, Joint Time-Frequency Analysis, Prentice-
lated filter banks, IEEE Trans. Circuits Syst. II Analog Digit.
Hall, New Jersey, 1996.
Signal Process., 42: 732, 1995. 43.
R. D. Koilpillai and P. P. Vaidyanathan, Cosine-modulated FIR
50 MULTIRATE FILTERBANKS
tion time-varying FIR multirate filter banks with overlaps, Proc. 82.
M. Ikehara, K. Sawaaki, H. Isobe, A. Yamashita, and H. Kur-
IEEE Int. Symp. Time-Freq. Time-Scale Anal., Philadelphia,
oda, Design of 2-D perfect reconstruction filter banks for arbi-
Oct., 1994.
trary sampling lattices—Design of 2-D N-th-band digital filters,
61. S.-M. Phoong and P. P. Vaidyanathan, Paraunitary filter banks
Electron. and Commun. Japan. Part 3, Fundamental Electron.
over finite fields, IEEE Trans. Signal Process., SP-45: 1443–
Sci., 76: 85, 1993.
1457, 1997. 83.
B. L. Evans, R. H. Bamberger, and J. H. McClellan, Rules for
62. R. Coifman and Y. Meyer, Remarques sur l’analyse de Fourier
multidimensional multirate structures, IEEE Trans. Signal Pro-
` ˆ
cess., SP-42: 762–771, 1994.
a fenetre, C.R. Acad. Sci. Paris, 312, Serie I: 259–261, 1991.
nent filter banks, Proc. IEEE Int. Symp. Circuits Syst., 1996.
1990.
˘ ´ 92.
A. Delopoulos and S. Kollias, Optimal filter banks for signal re-
74. J. Kovacevic and M. Vetterli, The commutativity of up/down-
1990.
Signal Process., SP-40: 882–893, 1992.
77. A. A. C. M. Kalker, Commutativity of up/down sampling, Elec- 95.
X.-G. Xia, Applications of multirate filterbanks in error correc-
tron. Lett., 28: 567–569, 1992.
tion codes: Partial response channels, Proc. of SPIE: Wavelet Ap-
MULTIVARIABLE SYSTEMS 51
SIFICATION.
100. Z. Cvetkovic and M. Vetterli, Oversampled Filter Banks, IEEE
AGE CLASSIFICATION .
101. Z. Cvetkovic and M. Vetterli, Robustness of overcomplete ex-
pansions, in J. Zeevi and R. Coifman (eds.), Signal and Image
MULTISTAGE DECISION MAKING. See DYNAMIC PRO-
Representation in Combined Spaces, Boston: Academic Press,
GRAMMING.
1996.
MULTITARGET TRACKING. See SONAR TRACKING.
102. K. Nayebi, T. P. Barnwell, and M. J. T. Smith, Analysis-synthe-
MULTITHREADED ARCHITECTURES. See DATA-FLOW
sis systems with time-varying filter bank structures, Proc. IEEE
111. W. Li and Y.-Q. Zhang, A proposal for very low bit rate video
coding standard using vector subband coding (VSC), ISO/IEC
JTC1/SC29/WG11, MPEG-94-209, July 1994.
112. V. M. Gadre and R. K. Patney, Vector multirate filtering and
matrix filter banks, Proc. IEEE Int. Symp. Circuits Syst., pp.
1360–1363, 1994.
113. X.-G. Xia and B. M. Suter, Multirate filter banks with block
sampling, IEEE Trans. Signal Process., SP-44: 484–496, 1996.
XIANG-GEN XIA
University of Delaware
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3029.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Ocean Thermal Energy Conversion
● Email this page Enter words or
Standard Article
Abstract
History of OTEC
Principles of Operation
Closed Cycle OTEC
Kalina Cycle
Environmental Considerations
Economics of OTEC
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3029.htm (1 of 2)17.06.2008 16:43:23
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3029.htm
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3029.htm (2 of 2)17.06.2008 16:43:23
OCEAN
THERMAL ENERGY CONVERSION
Ocean
thermal energy conversion (OTEC) indirectly converts
solar energy
into electricity. From a thermodynamic perspec-
tive, OTEC
power cycles operate as continuous heat engines
driven by
the transfer of energy between a thermal source
and sink.
OTEC expends renewable solar energy; therefore,
recurring
costs to generate electrical power are minimal.
However, the
fixed or capital costs of OTEC systems per kilo-
watt of
installed generating capacity are very high. This is a
consequence
of the low theoretical efficiency of OTEC, which
demands
large components, such as heat exchangers and
pipelines,
to accommodate the thermal energy transfers nec-
essary to
produce small amounts of electricity. The high fixed
costs
dominate the economics of OTEC to the extent that it
currently
cannot compete with conventional power systems
except in
limited niche markets. Toward this end, consider-
able effort
has been expended over the past two decades to
develop OTEC
by-products, such as freshwater, air condition-
ing, and mariculture, that might offset the cost penalty of between
the surface and 1000 m, this requirement limits the
electricity generation. number of
candidate OTEC sites. One of the principal techni-
Warm surface waters of tropical oceans make up the ther- cal
challenges of recent years, therefore, has been to devise
mal resource used by OTEC. The oceans, which cover more
strategies to export the benefits of the renewable OTEC re-
than 70% of the earth’s surface, intercept solar radiation source
to locations outside the tropics, possibly through the
passing through the atmosphere. Even though a portion of
production of synthetic fuels, such as the generation of hydro-
this energy is reradiated directly back to space, a significant gen gas
by electrolysis.
fraction is retained by seawater at the lower latitudes, heat-
ing the upper mixed layer of tropical oceans to an average
year-round temperature of approximately 28#C. This energy HISTORY
OF OTEC
is subsequently advected to higher latitudes—where radiant
losses exceed gains—by oceanic and atmospheric circulation The OTEC
concept was first proposed in 1881 by the French
(the net transfer of energy between the earth and its sur- engineer
J. A. D’Arsonval. D’Arsonval recommended using
roundings must, of course, be zero in order to maintain equi-
pressurized, liquefied gases as working fluids in a heat engine
librium). The amount of solar radiation retained by the operating
across the temperature difference between the sur-
oceans is enormous. Each day, energy equivalent to approxi- face and
deep waters of the equatorial oceans. A half-century
mately 250 billion barrels of oil is absorbed over the 60 mil- later,
in 1930, G. Claude, D’Arsonval’s former student, field-
lion km2 of tropical seas (1). This is more than three orders of tested a
variation of the OTEC concept in northern Cuba.
magnitude greater than the current daily energy consumption Claude’s
power cycle used warm seawater evaporated in a
of the world’s population. partial
vacuum as a working fluid. His shore-based, 50 kW
Inasmuch as the second law of thermodynamics dictates plant
consumed more power than it generated because of an
that only a fraction of the energy extracted from warm sea- overly
conservative cold seawater system that pumped 10
water can be converted to usable work by a power cycle, a times
more water than required. The plant operated for 11
thermal sink must be available to accept waste heat. The days
before the cold water pipe failed in a storm. In spite of
oceans provide such a sink in the form of a bottom layer of its
shortcomings, the experiment was the first to demonstrate
cold water lying beneath the warm, well-mixed (by wind and clearly
the technical feasibility of OTEC. Encouraged by the
waves) surface zone. This reservoir of cold water forms in the Cuba
tests, Claude launched a second attempt in 1933 to de-
polar regions and descends to flow along the sea floor toward velop
OTEC, this time in the form of a plantship moored
the equator (2). The warm surface layer, which can extend to about 100
km off the Brazilian coast. The power system was
depths of 100 m, is separated from the deep cold water by a sized to
produce 2 MW of turbine shaft power. Rather than
thermocline. The vertical gradient in temperature below the attempt
to export this power to the coastline directly, Claude
mixed layer is usually substantial. The temperature differ- decided
to produce ice, which had a favorable market value.
ence between the surface and 1000 m depth ranges from 10
Unfortunately, the expensive vertical cold water pipeline was
to 25#C, with larger differences occurring in equatorial and lost
during deployment, resulting in cancellation of the proj-
tropical waters. ect.
Although Claude continued to champion OTEC, the goal
The performance of OTEC power cycles depends ultimately of an
operating commercial plant was to elude him.
on the available difference in the temperatures of the warm
Following the 1933 project, several OTEC design studies
and cold seawater
T. The rule of thumb is that a
T of about were conducted. In 1956, a French team was preparing to in-
20#C is necessary to sustain viable operation of an OTEC stall a
shore-based, 5 MW Claude cycle plant at a site on the
power station. As shown in Fig. 1, which maps average
T west coast of Africa. The project was abandoned when it was
engine
tracted several key supporters in the early 1970s, notably W.
E. Heronemus of the University of Massachusetts and C. Ze-
∆QL
ner of Carnegie-Mellon University. In 1972, the National Sci-
ence Foundation awarded a grant to the University of Massa-
Cold
chusetts to assess the technical and economic feasibility of the
Thermal sea water
OTEC process. A second grant was awarded a year later to
sink temp. = TL
OTEC, the plant was anchored off the western coast of the
L H H
also has successfully produced potable water.
Although funding for recent OTEC programs has waxed Assuming
a maximum ocean surface temperature TH of about
and waned over the more than 20 years since the first oil 30#C
(303 K) and a typical deep seawater temperature TL of
embargo, research and development activities have signifi- 4#C (277
K), the limiting performance of OTEC power systems
cantly advanced OTEC technology. Unfortunately, none of the is
programs to date has taken the critical step of constructing
and operating the first commercial-scale (MW scale) pilot
η = 1 − (277/303) = 0.086 or 8.6% (4)
OTEC power station. Until this happens, the vast OTEC en-
ergy resource will continue to remain untapped. This
implies that more than 90% of the thermal energy ex-
tracted
from the ocean’s surface is ‘‘wasted’’ and must be re-
jected
to the cold, deep seawater. The low efficiency of OTEC
PRINCIPLES OF OPERATION
necessitates large heat exchangers and seawater flow rates to
sustain
the
Q’s needed to produce relatively small amounts
OTEC power systems operate as cyclic heat engines. They re- of
electricity. In contrast, the Carnot efficiency of a state-of-
ceive thermal energy through heat transfer from a resource, the-art
combustion steam power cycle, which taps a much
here, surface seawater warmed by the sun, and transform a higher
temperature energy source, may exceed 60%. The
portion of this energy into electrical power. The Kelvin-Planck lower
thermal quality of the OTEC resource, therefore, im-
statement of the second law of thermodynamics precludes poses a
significant penalty on the heat engine that ultimately
complete conversion of the thermal energy into electricity. A is
manifested in high capital costs.
(P − P ) = ρg (z − z ) (6)
which, although not shown, accounts for most of the parasitic
2 1 1 2
with
system materials, toxicity, and environmental hazard.
Warm sea
Cold
water in
sea water
discharge
Noncondensable
Desalinated
water
gases
Deaeration Vacuum chamber vapor
(optional) flash
Turbogenerator Condenser Vent compressor
evaporator
Desalinated
water
(optional)
Cold sea
Noncondensable Warm sea
water in
gases water
discharge
Lift
bine may employ a direct contact condenser (DCC) in which
process
cold seawater is sprayed over the vapor, or a conventional
surface condenser that physically separates the coolant and
Warm sea
Hydraulic
the condensate. DCCs are inexpensive and have good heat
water in turbogenerator
transfer characteristics because they lack a solid thermal
Sea level
boundary between the warm and cool fluids. Surface condens-
ers are expensive and more difficult to maintain than DCCs;
Mixed water
however, they produce a marketable freshwater by-product.
discharge
Effluent from the condenser must be discharged to the en-
Cold sea
vironment. Liquids are pressurized to ambient levels at the
water in
point of release by means of a pump, or, if the elevation of the
(a)
condenser is suitably high, they can be compressed hydrostat-
ically. As noted previously, noncondensable gases, which in-
clude any residual water vapor, dissolved gases that have
come out of solution, and air that may have leaked into the
system are removed by the vacuum compressor.
Open cycle OTEC eliminates expensive heat exchangers at
Sea level Warm sea Mixed water
the cost of low system pressures. Partial vacuum operation
water in discharge
Hydraulic
in-leakage and promotes the evolution of noncondensable
turbogenerator process
gases dissolved in seawater. Power must ultimately be ex-
pended to pressurize and remove these gases. Furthermore,
as a consequence of the low steam density, volumetric flow
rates are very high per unit of electricity generated. Large
components are needed to accommodate these flow rates. In
Cold sea
particular, only the largest conventional steam turbine stages
water in
have the potential for integration into open cycle OTEC sys-
(b)
tems of a few megawatts gross generating capacity. It is gen-
erally acknowledged that higher capacity plants will require
Figure 6. Schematic diagram of (a) mist lift and (b) foam lift OTEC
a major turbine development effort (6).
systems.
Ammonia
Steam
condenser
Noncondensable
gases
##yy##yy
liquid
practice. Warm
water pipe inner diameter 2.5 m
Both open and closed cycles require similar volumes of sea- Cold
water pipe outer diameter 1.6 m
water to produce a unit of electrical power. For megawatt-
Mixed discharge pipe outer diameter 3 m
scale plants operating over a seawater temperature difference Cold
water pipe intake depth 1000 m
of 20#C, about 3.5 m3/s of warm surface water must be sup-
Gross power 1838 kW
2 2 2 2
Warm water return temperature 22.9 #C been
investigated in consideration of its role as a greenhouse
Cold water return temperature 10.3 #C gas.
It is believed that open cycle plants will release between
Ammonia flow rate through turbine 274 kg/s 6 g
and 38 g of CO2/kWh, whereas closed cycle systems have
Turbine inlet temperature 21.1 #C
an
upper bound of 17 g/kWh. These values compare very fa-
Turbine inlet pressure 890 kPa
ECONOMICS OF OTEC
Cold seawater pump power 1150 kW
Mixed discharge seawater pump power 680 kW
Ammonia pump power 190 kW
Studies conducted to date on the economic feasibility of OTEC
agencies.
Freshwater
is important, therefore, that release back into the ocean be
conducted in a manner that minimizes disruptions to the The
condensate of the open and hybrid cycle OTEC systems
ocean mixed layer biota and avoids inducing long-term sur- is
desalinated water, suitable for human consumption and ag-
Agriculture 8. C.
B. Panchal and K. J. Bell, Simultaneous production of desali-
nated water and power using a hybrid cycle OTEC plant, J. Solar
An idea initially proposed by S. Siegal of the University of
Energy Eng., 109: 156–160, 1987.
Hawaii involves the use of cold seawater for agriculture. This 9. L.
A. Vega and G. C. Nihous, Design of a 5 MW OTEC pre-com-
concept involves burying an array of cold water pipes in the
mercial plant, Proc. Oceanology, 94: 5, 1994.
ground to create cool weather growing conditions not found in 10. G.
C. Nihous, M. A. Syed, and L. A. Vega, Conceptual design of
tropical environments. In addition to cooling the soil, the sys- an
open-cycle OTEC plant for the production of electricity and
STEPHEN M. MASUTANI
PATRICK K. TAKAHASHI
University of Hawaii
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W4408.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Optical Waveguide Transitions
● Email this page Enter words or
Standard Article
Abstract
Branches
Waveguide Horns
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W4408.htm (1 of 2)17.06.2008 16:43:42
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W4408.htm
OPTICAL
WAVEGUIDE TRANSITIONS
Most
useful integrated optical waveguide devices use a series
of
waveguide transitions where the cross section of the wave-
guide
structure changes in the direction of propagation. In
the Mach
Zehnder interferometer and in the directional cou-
pler,
both of which are commonly used, waveguide branches
are used
on both sides of a phase shift region to form an elec-
trooptic
modulator or switch. The performance of these de-
vices
depends on how well the transitions perform their in-
tended
function. To understand and optimize device
(a)
Ψ Ψ
i j
Figure 3. Electric field profiles for the first two normal modes of a
j
understanding the behavior of waveguide transitions and for
# and # .
i j
giving approximate design information. Computer solutions
are useful for more precise results, looking at the effects of
Local Normal Modes
radiation, and in cases where a two-mode formalism cannot
be employed.
By definition the normal modes of a structure are orthogonal,
i j
−∞
Normal Modes
x
#
z0
z
y
z
#
Figure 4. A local normal mode is defined by replacing the slowly
# x
varying waveguide structure z0 at (solid
line) with a structure con-
#
stant with z (dashed line), and obtaining the normal mode solutions.
Figure 2. A channel waveguide horn is used as a transition struc-
Local normal modes are used as approximations to normal modes in
ture to connect single-mode and multi-mode sections of waveguide.
transition structures.
x
Φ
a
D D + ∆D
3 3 3
z
#y#y##yy##yy
0 1
D3 ∞ (a)
Figure 5. Abrupt transition at which the waveguide separation in-
creases at a plane. This model represents a section of a branching or
separating waveguide.
Φb
(b)
ADIABATIC AND ABRUPT TRANSITIONS
occur.
K = K e−γ3D 3 (2)
The opposite extreme is a fast or abrupt transition in
0
which a transition between two waveguide structures is made
where
K0 is a constant and #3 is the transverse component of
so abruptly that the maximum amount of power transfer be-
2K
lated from a consideration of boundary conditions at the tran-
sition. We
show below that, in this approximation, power transfer
(b)
mode field profiles are denoted by # # # and # # # , and
a ia b ib
Ψ Ψ
are called the modes of the uncoupled waveguides.
i j
j 1i0
0 i0
i1i0 i1 j 1i0 j 1
#i ≈ d #a + e#b (4a)
output field is
#j ≈ −e#a + d #b (4b)
#1 = Si 1i0 #i 1 + Sj 1i0 #j 1 (8)
Normalization is provided by d2
e2 # 1. Since the normal
2 (X + 1)
Sj 1i0 Sj 1j 0 P 1/2ej φj 1i 0 (1 − P )1/2ej φj 1j 0
e = # 1 #1 − 2 X 1/2 ##1/2 (5b)
with
2 (X + 1)
0 1
batic or abrupt, or intermediate between these two cases. The
0 i0 0 0
may be in separate waveguides, or a mode may have its power
put modes are given by Eqs. (4a) (for # ) and (4b) (for # ),
distributed between two waveguides. Back reflections, radia-
i1 j 1
1 1
tion, and losses are neglected, so that only forward scattering
is accounted for. The scattering matrix S can be defined by
(10):
#1 = (1 − P )1/2ej φi 1i 0 #i 1 + P 1/2ej φj 1i 0 #j 1 (10)
#j 0 0 0 Sj 0i 1 Sj 0j 1 #j 0
the boundary. We can pick three phases to be zero and the
= (7)
# S S 0 0 #
fourth must be chosen to satisfy Eq. (9b). This choice is re-
i 1 i 1i0 i 1j 0 i 1
S
lated to the boundary conditions, and may be chosen consis-
# S 0 0 #
j 1 out j 1i0 j 1j 0 j 1 in
tent with the boundary conditions from a knowledge of
----------------------- Page 413-----------------------
tions given by
j 1 ij i0 jj j 0
i 1 ii i0 ji j 0
I
boundary, that is, # # # . This yields
i0,j 1
0 1
c ≈ (15)
ij √I I
i0,i0 j 1,j 1
(f 1 − f 0 )2
P = (11)
and I # ## # dx, for #,# # i,j . In Eq. (14), A is the mode am-
(f 1 − f 0 )2 + (1 + f 1 f 0 )2
#,# # # #
jj ii
where f # e /d and f # e /d , and f (X ) is given by
0 0 0 1 1 1
priate substitution of i andj in Eq. (15), and c # #c , a conse-
ji ij
quence of power
conservation. These equations are appro-
e
f = = −X + (X 2 + 1)1/2 (12)
priate for a single step. By dividing a transition into many
d
small steps and iteratively applying these equations one can
exp [ #j (# z
# )]. We generalize to a step in which an unspec-
i i
One commonly used numerical approach to treat waveguide
ified parameter p varies from p to p
#p across the step.
transitions is to divide the transition into a series of small
Thus, p may be the channel width in an expansion horn or
abrupt steps, connected by regions of constant waveguide pa-
the waveguide separation in a separating waveguide struc-
rameters, and calculate the mode propagation through this
ture. We expect that cij will be proportional to the change #p
idealized geometry (13,14). For the single step of Fig. 5, the
across the step, and remove this dependence by defining a
waveguide modes of interest are assumed to be incident on
new coefficient Cij
side 0, and the transmitted mode amplitudes are computed
by imposing boundary conditions at the interface, i.e., the
Cij = lim #cij # (17)
be written as
= C A − j β A (18a)
dz i j dz i j j
where
dAi dp
= −Cij A − j β A (18b)
a (z ) = βz + φ (13b)
dz dz j i i
ij ij 3 i0 j 1
−∞
dp = γ ##βij # (19)
dz Cij
where the parameter p is replaced by #D , the increase in
γ3X
dai −ju
Cij = 2 (24)
du = −γ e aj (20b)
2 (X + 1)
where u # z
which gives Cij as a function of X . This result is valid for sym-
# ## dz #. Equation (20) is a set of conventional
0 ij
metric branch shapes. For asymmetric separating wave-
coupled mode equations with constant coefficient # for the
a
j
given by Eq. (6).
(4γ 2 + 1) 2
BRANCHES
Φ
branch behavior depends only on the parameter ##/ ##3
and
a
Φb
Shaped Branches
(a)
mode theory approximations for ##ij and Cij are assumed. For
Figure 8. The limiting behavior of (a) mode splitting and (b) power
2 2 dz γ3 X 2
dividing branches for single local normal mode input i or j. The mode
splitting branch separates the local normal modes; the power dividing
branch divides input optical power while maintaining relative phase
This result can be integrated to give the branch shape
comes very large when the ratio Cij/ ##ij is small, at the begin-
whether
5
This dependence on the quantity ##/ ##3 can also be dem-
4
branch with separation D3 between the arms, we have
# #
dD3/dz, for small angles. The coupling coefficient K is given
by Eq. (2), with D3 # 0 at z # 0. Then, from Eq. (3), X can be
3
expressed as X # X exp[# D ] where X # ##/2K is the
value
0 3 3 0 0
–
of X at z # 0. The coupled normal mode equations [Eq. (18)]
3 –– x0 = 0.01
D – 2
3 –
with p # D3, and with the reduced mode amplitudes intro-
γ –
2
duced previously, can be written for constant # as
x0 = 0.1
daj 1
= a exp[−ju (X ,X )]
(26a) 1
dX 2 (X 2 + 1) i 0
x0 = 0.5
da 1
i
= − 2 aj exp[ju (X ,X0 )]
(26b)
dX 2 (X + 1)
0
βγ
∆ z
u(X ,X0 ) = #β # X √X #2 + 1 dX #
(26c) Figure 9. Normalized plot of branch arm separation versus direction
θγ3 X 0 X #2
of propagation z for shaped structures with initial conditions X . The
shape is symmetric about a center line between the two guiding re-
In Eq. (26) we have used coupled mode theory representa-
gions. This shaped structure maintains a constant level of
power
tions for ## and C . It is clear from this representation that
transfer between normal modes throughout its length.
ij ij
given by taking D3 # 0
1
l = (30)
b √ 2
2γ #β X0 + 1
2 a2
a b
i j
|Aj |2 = 4γ 2 sin2 #u #4γ 2 + 1# (31a)
|A i0 |2 4γ 2 + 1 2
–b
with
2 a
1
−1 −1
u = (tan X − tan X0 ) (31b)
(b)
2γ
1 2 1 2
in Fig. 10 plots of # versus z for comparable shaped and linear
with equal phase [Fig. 11(a)]. We introduce a new set of mode
adiabatic branches. The linear branch has a constant # equal
amplitudes a and b defined by a1 # a
b and a2 # a # b, and
to #m [obtained from Eq. (29)] for the shaped branch. The
decompose the original set of modes into a new set with equal
branches are chosen so that they produce comparable mode
in-phase (a) and equal out-of-phase (b) amplitudes, as shown
conversion. The length advantage for the shaped device is
in Fig. 11(b). The in-phase mode amplitudes will combine to
clear. However, as noted before, the shaped devices may have
form local normal mode i in the output waveguide, and the
to be modified to maintain acceptable radiation loss, which
out-of-phase mode amplitudes will combine to form local nor-
may somewhat reduce their advantage.
mal mode j . Conservation of power will show that the output
i j
Figure 10. The parameter # is plotted versus z for shaped and linear
metric index change makes ## large, so the device acts as an
branches designed to be mode splitters. The magnitude of # shows
adiabatic mode splitter. By reversing the sign of the applied
where power transfer between normal modes occurs in each branch.
voltage, the asymmetric index change reverses, and power ex-
creasing
coupling problem between the modes as W increases
in a linear
horn with constant #. A solution to this problem is
to shape
the horn as described subsequently.
###yyy For an
analytic solution to the coupled normal mode equa-
yy
## tions
using a shaped horn (16), the required horn shape is
obtained by
substituting p # W in Eq. (19). Using the expres-
sions
obtained for ##ij and Cij in that result yields the local
θ
m horn angle
# (Fig. 13)
1 dW 4π λ
θ ≡ = −γ g (34)
###yyy
2 dz 3 W
yy
##yy##
where # is
a constant. The shape of this horn will be parabolic
with
yy
##yy##
0
one output arm. Switching is achieved by changing the polarity of the
Arguments for the length efficiency of this shape are the
applied voltage. same as
those made earlier for the shaped branch. With a
parabolic
shape # # 2Cij#/ ##ij remains constant throughout
the horn,
and the strength of mode coupling between i and j
is evenly
distributed along the length. This is not true for
its the other arm, giving switching action. Minimizing the other horn
shapes where # will be a function of z . For the
length of this adiabatic device has generated considerable same
amount of mode conversion, in horns with the same
theoretical and experimental (24) interest, with the shaped expansion,
the parabolic horn should be shorter than other
branch solution (25,26) playing a significant role. shapes,
within the approximations made.
WAVEGUIDE HORNS
DIRECT
COMPUTATIONAL METHODS
2πλ
#βij = W2g (32)
θ
####
3
Cij = − 4W (33)
x
####
channel where the modes are far from cut off. The ratio Figure 13.
Parabolic channel waveguide horn. Using the parabolic
#C / ## # increases as W increases. This will result in an in- geometry
allows one to analytically calculate horn transmission.
ij ij
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W6213.htm
} { { } }
HOME
ABOUT US
CONTACT US
●
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Power System Control
● Email this page Enter words or
Standard Article
Abstract
Deregulation
Facts
system variables and this is translated into a control action via the actuator.
With the control action initiated
by the actuator, the plant moves to a new state or operating point and the cycle
continues.
Figure 2 will help to illustrate the control problem in modern power systems.
The principal function of
the power system control unit is to adjust the output of the available generators
to meet the system load while
meeting all constraints imposed by engineering design, regulatory, and financial
considerations. This desire
can be expressed as
If Eq. (1) were satisfied all of the time, then there would be no need for a
governor control system. However,
loads change according to the needs of the consumers and are strongly dependent on
weather patterns and
work day schedules. Since system load is a dynamic quantity that cannot be
precisely predicted, the governor
control systems are often reacting to correct for the mismatch of generation to
load. The generating units with
governors are acting to ensure the synchronous network can withstand these
imbalances without collapsing.
Therefore, the governor control system provides primary frequency control to the
synchronous network.
The result of this mismatch of generation to load is a frequency deviation
from the nominal value. Under-
generation or an increase in load will cause the system frequency to drop, while
over-generation or a decrease
in load will cause the frequency to rise. Equipment constraints limit the maximum
frequency deviation from
nominal values. Larger deviations lead to problems such as unacceptable vibration
levels in steam turbine
blades.
Governor control or primary frequency control is a short time frame (seconds)
mechanism for ensuring
generation meets load. It will not, however, reset frequency to nominal after a
disturbance. Automatic gen-
eration control (AGC ) operating in the tie line bias mode provides secondary
frequency control on the longer
time frame (minutes) and is responsible for resetting system frequency to nominal
after a sudden frequency
shift has been arrested by the primary frequency control system. The two systems
work together to provide
stable operations at or near setpoint frequency on the synchronous network. In
general, governor control can
be thought of as the high-frequency component of frequency control and AGC as the
low-frequency and/or bias
component. They should be thought of as two different control loops.
The governor control system can be illustrated from Fig. 1. Considering a
thermal unit, the system input
is steam flow, the actuator controls a steam admission valve, the plant block
represents the unit’s prime mover
(turbine), a device for measuring speed provides an input to the
feedback loop, and the system output is
mechanical power to the generator. The key to this control system is the speed-
droop characteristic illustrated
in Fig. 3. This characteristic shows that when there is a load increase on the
generator that it will slow down.
The slope of this characteristic, R , is referred to as the droop of the generator.
Then R is the steady-state
relationship between system frequency and generated power. Normally, all of the
individual units should have
their droop characteristics set to be about equal to minimize turbine governor
oscillation. Also, with the same
droop characteristic, each unit will share a change in system load in proportion to
the individual unit’s MW
rating. Values for droop ideally range from 3 to 5%. However, according to a recent
EPRI survey, the true
measured droop characteristic of generators actually varied from 1% to 21% with 5%
being the most common
(1).
The governor control system is an integral controller with a feedback loop
gain of R . Therefore, in operation
when the speed of the turbine is sensed to be slowing down (i.e., the load on the
machine has increased), the
feedback loop will provide a signal to open the steam admission valve. When more
steam is admitted into the
turbine this will result in more mechanical power out of the turbine and eventually
more electrical power out
of the generator to serve the load increase. When an increase in speed occurs (load
decreasing), the opposite
actions will be taken and the electrical output of the generator will be decreased.
More detailed analysis of the
governor control system can be found in Refs. 2,3,4.
Figure 5 indicates that the units available for dispatch are selected by the
economic dispatch (ED ) program.
When units are dispatched to a new generation level there are a number of equality
and inequality constraints
that can be applied to the control problem. Equation (1) illustrates the simple
equality constraint that gener-
ation must equal load. However, this simple constraint is complicated in reality
when consideration is given
to the fact that impedance of the transmission and distribution network are between
the generation and load.
Therefore, there are real power losses associated with the transmission network
that must be added into the
load term as
or
There are also inequality terms such as minimum and maximum power outputs for the
individual plants.
Figure 2 also shows that each plant has a fuel cost associated with it (F i ).
Therefore, it is possible to apply the
principles of optimal control theory to this problem and develop an optimal
dispatch level for the units subject
to some defined constraint. Economic dispatch is the optimal control paradigm for
finding the power output
levels for multiple units while minimizing cost. This is described by two equations
known as the coordination
equations. One of them is Eq. (4) and the other is
There are obviously many constraints that could be used in finding the optimal
dispatch for a power system
(e.g., environmental and hydro-thermal). References 3, 4, 9, and 10 are good
sources.
Deregulation
As a regulated monopoly the electric utility industry has been the focus of a
variety of regulators at all levels
of government—the Federal Energy Regulatory Commission (FERC ), the Rural
Electrification Administration
(REA ), state public utility commissions, and even itself in the form of NERC.
However, over the past two decades,
in an attempt to introduce competitiveness into the electric utility industry, the
US Congress and FERC have
introduced a series of legislation and rules. This legislation and rule-making was
aimed at opening the customer
base within a utility’s franchise area to competition. The Congress, in the Public
Utility Regulatory Policies
Act of 1978 (PURPA ), began this attempt by confirming that the electric industry is
an interstate industry
subject to federal control in almost all dimensions of its regulated activities
(11). The next major legislation
was the Energy Policy Act of 1992. This act laid the legal groundwork for
competitive wholesale markets and
led to what has been referred to as the FERC’s Mega-NOPR (Notice of
Proposed Rule-Making). From the
Mega-NOPR, two final orders were issued in 1996.
Order Number 888 stated that it was the goal of FERC to remove impediments to
competition in the
wholesale bulk power marketplace and to bring more efficient, lower cost
power to the nation’s electricity
consumers. This was primarily to be accomplished by guaranteeing all
generating systems open access to
utility transmission networks. An important aspect of the order was that the
reliability of the interconnected
system was not to be compromised by the wholesale wheeling of power. To
ensure this did not occur, the
transmission provider was permitted to charge an access tariff to the
transmission network that included
ancillary services needed to maintain reliability and power quality at the
transmission level.
Order Number 889 contained rules for establishing and governing an Open
Access Same-time Informa-
tion System (OASIS). This order requires each public utility that owns, controls,
or operates facilities used for
the transmission of electric energy in interstate commerce to create or participate
in an OASIS. The OASIS
will provide open access transmission customers and potential open access
transmission customers with in-
formation, provided by electronic means, about available transmission capacity,
prices, and other information
that will enable them to obtain open access nondiscriminatory transmission service.
The latest information on
FERC orders may be found at the NERC homepage (http://www.nerc.com/).
Facts
As the electric power transmission network grows in size and complexity the system
response is more difficult to
predict and control. The problem is aggravated as an increasing number of
independent power producers (IPP)
and non-utility generators (NUG) tie into the transmission system, and the
seemingly inevitable ruling of "open
access of transmission systems" by the Federal Government takes effect. A method of
accurate, dependable
control of the transmission system is needed if utilities are to provide reliable
service in the future. To answer
this challenge, the Electric Power Research Institute (EPRI) has launched a
research project called Flexible
AC Transmission Systems (FACTS).
First put forward in 1988, the FACTS program would utilize advances in
silicon science to develop FACTS
devices that can allow for compensation and control of system parameters much
faster and more accurately
than the mechanical versions in place today. These devices make use of thyristors
as solid state switches to
replace the slower, bulkier mechanical switches. FACTS devices would prove very
useful in controlling the three
parameters that determine the flow of power in a transmission system: the line
impedance, the magnitude of
the bus voltages, and the phase angle between the bus voltages.
The ideal result of the FACTS program is a power transmission system that
employs many different
FACTS compensators to eliminate problems such as loop flows, inefficient use of lines
to thermal limits, and
poor voltage regulation; all while improving power quality and system stability to
transients and disturbances.
Loop Flows. One problem that the FACTS program deals with is that of
loop flow on power systems.
Loop flow is defined as the difference between scheduled and actual power flows on the
interface between two
systems. It is the flow of scheduled power from a generation area to a load through
a transmission path where
it is not intended. In other words, one utility ends up using another utility’s
system to transmit power without
permission or payment. Loop flow occurs when there are many parallel paths from the
generation site to the
load center. The problem of loop flow from one utility to another is not currently
addressed except by after the
fact examination of scheduled transactions versus system conditions to see if in
fact a loop flow situation was
present during a given period. This is done because of the limited number of
transactions that occur between
utilities and the reasonably rare occurrence of serious loop flows that accompany
them. However, when open
access of transmission systems becomes a reality, the number of transactions on the
system, and therefore the
number of opportunities for loop flow to occur, will increase.
The problem of loop flows may be resolved by using FACTS devices to control
the power flow between
the possible paths to the load. The phase shifter appears to be the most effective
compensator to avoid loop
flows. The phase shifter is a device used to inject a series voltage into the
transmission line at a controllable
phase angle. This variable voltage source is used to inject a voltage of the
appropriate magnitude and phase
to improve the power transfer capability and the transient stability. Phase
shifters can be used to decrease
the flow of power in the nonscheduled path and thereby increase the flow in the
scheduled primary path.
Conventional phase shifters based on series transformers with mechanical tap
changers are already in use
in power systems but their operation is much slower in comparison to the expected
response of FACTS-based
phase shifters. The thyristor controlled phase shifter is expected to be able to
control the phase angle quickly
enough to increase the power transfer before the natural power swing occurs.
Although FACTS devices do have the potential to allow much finer control of
power flows on the network
than is currently available, they have not yet been widely utilized.
This is largely due to their cost and
unfamiliarity to system design and operations personnel. Another problem with FACTS
devices is that there
is no widely recognized methodology available at this time to determine
in real time how they should be
coordinated on a regional basis to optimize the network. In fact, there is no
consensus as to what such an
POWER
SYSTEM CONTROL 9
overall optimization function should be for the network. Much additional work needs
to be done in this area
before widespread use of FACTS devices will become evident.
Line Loading to Thermal Limits. One application of series capacitors
is to use them to adjust line
impedance to make full use of a line’s thermal limits. For example, suppose there
are two parallel transmission
lines that compose the primary scheduled power path, or corridor, to a given load
area. One of the lines is
loaded to its thermal limits while the other has spare capacity. The power transfer
of the two lines cannot be
increased without exceeding the limit of the loaded line. Thyristor switched series
capacitors may be used to
decrease the impedance of the less loaded line, increasing the power flow on the
line. In this manner the total
power transfer over the two lines may be increased until both lines are at thermal
limits. Phase shifters could
also be used to control the power flow between the two lines by changing the phase
angle. However, series
capacitors appear to be better suited for this case because they serve the dual
purpose of controlling power
flow and generating substantial amounts of vars required by long transmission line
corridors operating near
thermal limits. Phase shifters, on the other hand, do not supply vars but consume
them (which in turn must
be supplied by additional shunt compensation). The thyristor switched series
capacitors, then, are the more
cost-effective choice. However, it must be remembered that NERC requires utilities
to operate their systems
such that all lines will be below their thermal limit after the first contingency.
The use of series capacitors to change the effective impedance of a
transmission line brings up a topic in
need of much further research—the effect of FACTS on protection schemes. The
present relaying and protection
schemes would certainly have to be modified to accommodate the presence of the
system changing FACTS
devices. One possible solution would make use of intelligent protection schemes
capable of receiving information
from the FACTS controllers about the status of the system and adapting the relay
settings accordingly.
Voltage Regulation on Long Lines. The static var compensator (SVC) is
a thyristor controlled shunt
device primarily used to raise the voltage on long transmission lines. The SVC
supports the voltage on these
lines and, therefore, increases the power transfer. Studies have been done to
determine the most effective
placement of SVCs on a long line. The studies have shown that midpoint compensation
has the most positive
effect on voltage regulation and power transfer with the least negative effects on
system dynamic performance
(12). Although industry practice for voltage control is still dominated by switched
capacitor banks, SVCs have
shown potential for more widespread use in the future.
Power systems are equipped with numerous voltage controlling devices to ensure the
system’s voltage profile
remains within predetermined limits for a variety of topological and loading
conditions. As system load con-
tinues to increase without the construction of new sources of generation, the fear
of a voltage collapse also
increases. To aid in controlling the voltage profile of systems it will become more
important in the future to not
only have AGC control active power flow, but also reactive power flow.
Federal legislation will continue to impact the control of generation units
primarily from an environmental
viewpoint. Environmental dispatch will take on more of a supervisory role to
classical AGC to ensure that the
proper mix of units are operating to maintain a utility within environmental
regulations (clean air and clean
water) and such that the most judicious use of allowances is made. However, some
federal legislation that
mandates the use of alternative fuels for US Government automobile fleets and
alternative fuel suppliers may
have the effect of load leveling (due to the use of chargers at night) that the
utility industry has sought for
many years.
Demand side management (DMS ) is another tool to assist in the balancing of
load and generation. A smart
DMS administered by AGC is an issue that needs to be further explored. This would
permit AGC the option
to drop load at the distribution level, in lieu of requesting additional expensive
(economic or environmentally)
generation, if it determines this is the best economical solution. Traditional DSM
programs can also be rapidly
supplemented with the use of market-driven programs such as real time pricing
(RTP ). Incorporating RTP into
the unit commitment and AGC programs is another interesting problem.
The control of distributed generation is another emerging issue. As
additional nonutility generators are
added to the system (at both the transmission and distribution level) AGC will need
to be able to properly
dispatch utility owned units to meet load, provide voltage and frequency support,
and overall system security.
This will become a highly complex communication, control, and protection problem.
In today’s rapidly changing electric utility industry, mergers are becoming
commonplace, energy futures
are being traded on the commodity exchange, independent power producers (IPPs) are
coming on-line wherever
the market looks promising, and power marketers are acting as the go-between for
those with energy to sell and
with those looking to buy. As these things transpire the traditional utility is
looking to realign itself to remain
competitive in this new market. To accomplish this, most utilities are
restructuring. Separate generation
companies, transmission companies, and distribution companies will become more
prevalent in the future. The
distribution company (i.e., load) will drive the market by seeking low-cost sources
of energy from a variety of
sources. This energy may be delivered via one or more transmission companies or it
may be generated locally
at the distribution level (therefore, avoiding transmission charges). The idea of
building generation at the
distribution level is attracting several entrepreneurs. This dispersed or
distributed generation, as it is referred
to, is normally smaller capacity generation, but it has the potential of becoming
very important in system
reliability.
Even the control structures are presently being debated. One scenario is a
regional control area (RCA)
consisting of a regional transmission group (RTG ) and an associated
regional system operator (RSO ) in a
control area whose boundaries may be based on existing NERC regions. The RSO, which
may be nonutility
owned (an independent system operator or ISO ), would perform some or all of the
functions of the existing pool
control centers, depending on specific implementation. The primary function of the
RSO would be to ensure
comparable access and fair play among all transmission system stakeholders while
maintaining the integrity
and reliability of the network.
It is also important to consider the human side of power system operations.
In the past, system operators
were highly experienced and had intimate knowledge about the flow patterns and
operational characteristics of
their system. The operators of today, however, are confronted with significantly
more complex systems charac-
terized by rapid changes in stakeholders and flow patterns. Traditional
relationships between stakeholders are
changing and new stakeholders are being added at a breathtaking pace. For example,
pool control center oper-
ators have traditionally been generation and transaction coordinators. They match
power needs with available
generation. This matching has been based primarily on cost and often ignored the
use of security contingency
tools (or utilized precalculated limits from transmission or operational planning
studies) in decision making.
These operators typically handled 10 to 15 bulk power transactions per hour and
their primary concerns were
those of pricing mechanisms and peculiarities of the generation resources and
transactions.
However, now system operators must be prepared to operate in a more complex
system, highly impacted
(if not actively constrained) by the complexities of transmission system loading,
open transmission access, IPPs,
wholesale marketers, OASIS (real-time information networks), etc. In such a system
most, if not all, generation
control and transaction scheduling decisions will have to be screened for potential
system reliability impacts.
These will include both the traditional static constraints (thermal and
voltage limits) as well as dynamic
impacts (generation ramp rates and stability) and impacts on import and export
capabilities.
BIBLIOGRAPHY
3. A. S. Debs Modern Power Systems Control and Operation , Boston: Kluwer, 1988.
4. A. J. Wood B. F. Wollenberg Power Generation, Operation, & Control , 2nd ed.,
New York: Wiley, 1996.
5. NERC, Glossary of Terms, http://www.nerc.com/glossary/index.html, August 1996.
ROGER L. KING
ROGELIO LUCK
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Primary Cells
● Email this page Enter words or
Standard Article
Abstract
Electrochemical Thermodynamics
Electrochemical Kinetics
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3032.htm17.06.2008 16:44:25
PRIMARY CELLS
Batteries are reactors that convert chemical energy into electric energy
on demand. Battery reactions are
oxidation–reduction reactions similar to those that occur in ordinary chemical
reactors. However, in an electro-
chemical cell, the basic component of a battery, the charge-transfer aspect of the
oxidation–reduction reaction
occurs on a surface rather than in the bulk. The electron transfer is isolated and
is available as a direct current.
The reduction and oxidation intermediates are transported through electrolytically
conductive media. The elec-
trode referred to as an anode is the site of the oxidation reaction. The reduction
reaction of the electrochemical
couple occurs at the electrode referred to as the cathode. Oxidation is a chemical
process in which an electron
is given up by the reactant. Reduction is a chemical process in which an electron
is accepted by the reactant.
The oxidation–reduction reaction proceeds when the exchange electron from the
reaction is conducted from
one electrode to the other through an external wire (or electrically conducting
circuit) and intermediate ions
formed in the process are internally transported through the electrolyte to
complete the reaction. The current
of an electrochemical cell results from this chemical reaction in which electrons
are released at one electrode
and taken up at the other. Equations (1) and (2) show the oxidation half-cell
reaction and the reduction half-cell
reaction. Equation (3) shows the overall cell reaction:
where M is the species that will be oxidized, R is the species that will be
reduced, e − represents an electron,
2 PRIMARY CELLS
is manifested as a voltage and current flow through a wire (conducting lead). Figure
1 depicted schematically
the energy transfer in electrochemical reactions. In generating electric current,
the electrons involved in the
electrochemical reaction are transferred from the active materials undergoing
oxidation to the oxidizing agent
by means of an external circuit. The passage of electrons through the external
circuit provide a direct means for
utilizing energy without heat as an intermediate step. Figure 2 gives a graphical
depiction of a cell or battery
(1).
Electrochemical Thermodynamics
PRIMARY CELLS 3
Equation (4) shows the relation between the change in Gibbs energy and the
electromotive force.
where # G0 is the standard molar Gibbs energy, R is the gas constant (8.31433
J/mol·K), a is the activities of
the reactants (A, B) and products (C, D), a and b are the stoichiometry of the
reactants A and B, and c and d
are the stoichiometry of the products C and D.
Equation (6) is the expression of the relation between the standard molar
Gibbs energy and the standard
electrode potential:
4 PRIMARY CELLS
measured values of cell voltage. When the reactants and product of an electrode
reaction share a single phase
or crystal habitat or form an intercalate, the voltage of that electrode depends on
the ratio of concentration
of the species. This results in a sloping discharge curve that is typical of cells
with MnO , TiS , MoS , etc.,
2 2 2
electrodes. When the reactants and product of an electrode reaction form separate
phases, the voltage of the
electrode is constant during discharge in the multiphase region. Typical
examples are cells with Hg/HgO,
Cd/Cd(OH) , and PbO /PbSO electrodes. These are often used as reference
electrodes in experimental studies
2 2 4
because of their invariant voltages.
Equations (8) and (9) are fundamental thermodynamic equations relating
entropy with the Gibbs energy
and enthalpy with entropy and Gibbs energy:
+ − 1
trode potentials are determined with respect to the standard electrode potential of
hydrogen, H + e → 2
H2 at 1 atm, which has been assigned by international convention a zero value.
(The basis for the standard
hydrogen electrode potential is a weak acid.) An electrode that is more negative
than the hydrogen electrode
PRIMARY CELLS 5
will have a negative standard potential, and an electrode that is more positive
than the hydrogen electrode will
have a positive standard potential. An electrochemical cell comprising two half-
cell reactions will result in one
electrode being reduced and the other oxidized. The electrode that has a more
negative standard potential will
be oxidized and the electrode with the more positive standard potential will be
reduced. This type of analysis
allows for the determination of the cell reaction.
The combination of standard electrode potentials for two half-cell reactions
can be used to determine the
cell voltage. Equation (12) shows the relation between the standard potential for
oxidation and the standard
potential for reduction, and Eq. (13) shows how the cell voltage can be determined
from the standard electrode
potential data.
6 PRIMARY CELLS
where µ is the chemical potential. Combining Eqs. (16) and (17) gives Eq. (18):
Combining Eqs. (10) and (18) as shown in Eq. (10) gives the relation between
the standard electrochemical
potential and the thermodynamic equilibrium constant. The thermodynamic equilibrium
constant K is defined
in Eq. (20) by rearranging this equation:
Electrochemical Kinetics
The reaction rates of electrochemical reactions follow the usual kinetic laws. The
rate of reaction is expressed
as a current flow in coulombs per second (amperes) rather than a change
in concentration per unit time.
Operational battery systems exhibit a kinetic hindrance when producing current,
which manifests itself as a
loss in voltage. Thus the voltage during use (discharge) is lower than the voltage
measured under open-circuit
(reversible) conditions. This lowering of voltage is called polarization and is
usually represented by η. There are
three types of polarization: (1) activation or reaction rate polarization, (2)
concentration polarization, and (3)
resistance or ohmic polarization. Activation polarization is related to the charge-
transfer mechanism process,
concentration polarization is related to the availability of reacting species at
the electrode–solution interface
(often a diffusion- or mass-transfer-controlled process usually in the electrolyte
but occasionally in the solid
crystalline lattice of the active material), and resistance polarization
is related to the ohmic hindrance to
electron or ion flow. The same kinetic factors take place during the recharging of
secondary cells, and the
energy loss is reflected as higher voltage necessary to recharge.
In almost all cases, the positive electrode in a battery system is a metal
oxide, sulfide, or halide. In general,
these are relatively poor electric conductors and would give extremely high ohmic
polarization (impedance) if
not blended or reinforced with supporting conductors. An exception is pyrite
(FeS2 ), which is a semiconductor.
Conductors added to positive electrodes include graphite, lead, silver, copper,
nickel, and other materials in
rod, mesh, wire, grid, and other shaped configurations. In almost all cases the
negative electrode is a metallic
element of sufficient conductivity to require only minimal supporting conditions. An
exception is the hydrogen
electrode, which requires a substantial conductive surface support, which also
serves for electron transfer.
PRIMARY CELLS 7
2
2
where i is the current density (mA/cm ), i is the exchange current density
(mA/cm ), α is the charge-transfer
p 0
p
0 and the rate of the forward and reverse reactions are equal. The forward and
reverse action rates can be
expressed as a current flow i0 across the activation energy barrier at equilibrium
(reversible potential) and is
termed exchange current. The exchange current is directly related to the reaction-
rate constant, and the usual
kinetic expressions apply directly to battery reactions. A low rate constant is
associated with a high degree of
irreversibility, so that the smaller the i0 the more irreversible the reaction and
hence the greater the activation
polarization for a given net current flow. This implies that high exchange currents
are favored for battery
reactions. However, high exchange currents often lead to changes in electrode
morphology with a consequent
and sometimes deleterious change in surface structure.
When the electrode reaction occurs with polarization larger than
about 0.05 V, then Eq. (21) can be
simplified. Equation (21) reduces to Eq. (22) at large positive overpotentials and
reduces to Eq. (23) at large
negative overpotentials.
8 PRIMARY CELLS
Equations (24) and (25) are the result of taking the logarithm of Eqs. (22)
and (23), respectively:
Equation (26) relates the voltage deviation of a cell from the rest potential
caused by a pulse of current
(this also depends on surface area of the electrode and the pulse should be defined
in terms of A/cm2 ). This
The Tafel relation holds true for most of the electrochemical reactions over a wide
range of overpotentials.
At low overpotentials the relationship fails to fit the data. By plotting the data
of overpotential against the
logarithm of current, a linear relation is clear at large overpotentials. The slope
of the line is the Tafel slope
and the x-axis intercept, after extrapolating the curve, is the exchange current
density. Figure 4 shows a Tafel
plot demonstrating the relationship (3).
Whenever the local concentration of a reacting component in a battery is low
or nears exhaustion, the rate
of reaction becomes controlled by the transport of that component to the reaction
surface site on the surface
or within the electrode. At this point diffusion equations, for example, Fick’s
laws, apply. Equation (28) shows
this relation:
PRIMARY CELLS 9
Figure 3 shows the performance of a typical cell. This plot, showing cell
voltage versus discharge current,
is known as a power curve. This curve shows the regions controlled by activation,
concentration, and resistance
polarization and the change as the battery is discharged (1).
Equation (30) shows the relation between discharge capacity and the mass of
the electroactive species.
Equation (30) is a form of Faraday’s law, and Eqs. (31) and (32) define Q and Gew .
where Q is the discharge capacity (C), W is the mass of the electroactive species,
and Gew is the gram equivalent
mass of the electroactive species.
The activation energy and reduced diffusion are the major contributions to
the reduced voltage during
discharge at lower temperatures. Thus, the energy must be initially introduced to a
system in order for the
reaction to proceed. This energy is called the activation energy. Equation (34)
shows a chemical reaction and
10 PRIMARY CELLS
where A are the reactants, B are the products, and X ∗ is the activated species.
The activated species is a molecule in which a bond is in the process of
being formed or broken. Equation
(35) shows the rate constant defined as a function of both the activation energy and
temperature:
2
23
where D is the diffusion coefficient(cm /s), N 0 is Avogadro’s
number(6.02252 × 10 molecules/mol), r is the
radius of the sphere (cm), and n is the diffusing media viscosity (P).
where E is the electromotive force (V), i is the current (A), and R is the
resistance(#).
PRIMARY CELLS 11
spectroscopy. These techniques provide data regarding the processes governing and
limiting electrochemical
cells.
Sweep voltammetry is a technique in which a linear voltage sweep
is imposed across a cell and the
resultant current is observed. Cyclic voltammetry is a form of sweep voltammetry in
which the voltage scan
is reversed after the initial sweep. Qualitative and limited quantitative electrode
data can be obtained from
sweep voltammetry. Reversibility of an electrode, voltage stability of an
interface, charge transfer versus
diffusion limitations, rate constants, and reaction mechanisms are examples of
physical properties that can be
determined by sweep voltammetry. The data for evaluating the Tafel relation for a
cell can also be obtained
through sweep voltammetry.
Chronopotentiometry is a technique that studies voltage transients at
an electrode during constant
current discharge. Chronoamperometry is a technique that studies current transients
at an electrode when
a potential change is imposed across the electrode. Chronoamperometry can
aid in determining diffusion
coefficients and for investigating kinetic mechanisms. Pulse voltammetry is similar
to chronoamperometry,
but the voltage input to the electrode being studied is in the form of a square
wave. This allows for many
variations depending on the setup of the wave amplitude and sign. This includes
reverse current through
positive and negative amplitudes.
Impedance spectroscopy is a technique in which a cell is perturbed from
steady state by a frequency
sweep of alternating current of low magnitude. The potential of the cell due to the
signal gives insight to the
electrochemical processes at the interface. Impedance spectroscopy will be
discussed in more detail in the next
section.
12 PRIMARY CELLS
PRIMARY CELLS 13
Fig. 9. Equivalent circuit for a multilayer electrode with two interfaces. This
network describes the equivalent circuit for
a multilayer electrode.
14 PRIMARY CELLS
Fig. 10. Prismatic cell. This cell design typically has a high power density
and low-to-moderate energy densities.
A knowledge of current distribution and factors that affect current densities are
essential in battery design
considerations. The distribution of current on electrodes may not be uniform, even
when the electrodes are
vertically opposed and uniformly spaced. If the current distribution across the
surface and/or through the
electrode is nonuniform, undesirable effects in cell operation occur, including
nonuniform chemistry, excessive
heat evolution, and low utilization of active materials (4).
Most battery electrodes are porous structures, often with a mixture of solid
nonconducting active mass and
solid electronically conductive inert materials, filled with an electrolyte
permeating the structure. The porous
PRIMARY CELLS 15
Fig. 11. Button cell. These are characterized as high energy density but
low power density, and they are simple to
manufacture.
16 PRIMARY CELLS
´
electrolytes,ZnCl –NH Cl, ZnCl , and KOH. These electrolytes represent LeClanche,
heavy duty (zinc chlo-
2 4 2
ride), and alkaline batteries, respectively, and are used largely in flashlights,
portable electronic devices such
as radio, cassette players, and toys. They are available in a wide variety of sizes
and shapes. They have a
sloping discharge voltage, which is a function of the thermodynamics of the
manganese dioxide (MnO2 ) crystal
habitat.
The silver and mercury cells are used to power smaller electronic
devices that do not have moving
parts, such as hearing aids, cameras, and calculators. They are usually
manufactured in small, round, thin
PRIMARY CELLS 17
Fig. 13. Bobbin Cell. Bobbin cells, like button cells, are simple to manufacture,
provide high energy with low-to-moderate
power capability.
configurations, often called button or miniature cells. They have a flat or constant
discharge voltage. The zinc
air cell is used for hearing aids. It has a short (one month) life after being
activated but can deliver twice the
energy of the same size silver and mercury cells. It also has a flat discharge
curve. Mercury cells have largely
been phased out, since 1990, for environmental reasons.
Lithium–copper oxide Li/CuO and lithium–iron sulfide Li/FeS systems are 1.5 V
replacements for the
mercury and silver button cells. The Li/CuO system also has potential as a long-
life, higher-energy replacement
for the Zn/MnO2 systems. They find application for which high-current pulse demands
are not required. They
have excellent shelf or storage life and operate over a very wide temperature
range. A higher-voltage cell
(∼3 V) can be made using FeS22 − with lithium.
18 PRIMARY CELLS
a lithium battery to power the pacing systems. It has a 7 to 10 year life in these
applications. For higher-current
pulse devices, such as defibrillators, lithium vanadium oxide and lithium silver
vanadium oxide systems are
used.
Li/SO2 and Li/SOCl2 systems are used, almost exclusively, for military
applications. They have reasonably
good high-current capability, have a flat discharge profile, have outstanding long
shelf or storage life, and
◦ ◦
operate over a wide temperature range (−40 C to +70 C).
There are many other battery systems, not listed here, that have
been developed for specialty uses.
Thermally activated (“thermal”) batteries are one example. These cells employ an
electrolyte (typically, an alkali
halide mixture) that is solid at room temperature but that becomes very conductive
at operating temperatures
◦ ◦
of 400 C to 600 C. An internal pyrotechnic is used to bring the cell
to its operating temperature (6). The
conductivities of the molten salt electrolytes can be orders of magnitude
greater than aqueous or organic
electrolytes, which allows for large power and current densities to be attained.
Many battery chemistries are capable of a single discharge but are
incapable of having the original
chemistry restored by electric means internal to a cell. This lack of
reversibility choccurs from charges in
morphology, crystal habitat, conductivity, and inability to be chemically
formed in the battery electrolyte.
7 characterized the material irreversibility as arising from (1) irreversibility in
phase changes and chemical
interconversion, (2) isolation, electrolytically or electronically, of active
materials, and (3) local poor conductivity
of materials in the discharged state.
BIBLIOGRAPHY
TERRILL B.
ATWATER
US Army
Communications Electronics Command
ALVIN J.
SALKIND
Rutgers,
the State University of New Jersey
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W6827.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Quantum Storage Devices
● Email this page Enter words or
Standard Article
Abstract
Summary
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W6827.htm17.06.2008 16:44:48
The semiconductor industry has used the ability to develop smaller and faster
devices to fuel the explosive
growth in productivity and functionality of electronic products. However, as
devices shrink below 0.1 µm in
size, physical phenomena must be identified that can produce devices that
work as well or better than the
larger devices utilized over the last 40 years. In particular, quantum devices that
work because of their small
size, rather than in spite of it, become more attractive. It is critical to explore
the potential of this new class of
ultra-small devices.
The continued trend in the integrated circuit industry toward smaller
individual devices (transistors) has
actually become part of the financial structure of the industry. There is a focus in
every area of device fabrication
toward next generation technologies that scale (shrink) in a favorable way.
Unfortunately, the resources and
technology required to decrease minimum feature size in each generation of
semiconductor devices continue
to increase. Industry is now wrestling with fundamental physical device size
limitations that threaten to limit
further scaling. This suggests that new device building blocks need to be developed
whose ideal size is in the
new operating regimes. Many quantum phenomena evidence themselves most strongly in
this sub-100 nm size
range. This gives quantum devices, which may have no intrinsic scaling
limitations, a potential role in the
future of the industry.
Quantum storage cells are composed of a device or of several devices which depend
on quantum confinement,
tunneling, or an interaction between these phenomena and space charge or other
effects. Preferably, quantum
storage cells should be composed of a single device that has multiple, electrically
distinguishable, stable states
accessed through one line and set through another. To be useful these
memory cells should have terminal
characteristics which allow them to drive variable-length interconnects with
acceptable noise margins and fan
out.
Quantum storage-based memory may be volatile or nonvolatile, static or
dynamic, as in other types of
memory. Nonvolatile memory retains its memory state when power is turned off.
Static random access memory
(SRAM) is an example of volatile static memory. Dynamic random access
memory (DRAM ) is volatile and
dynamic, requiring regular refresh cycles that consume significant amounts of power
that must be dissipated
as heat. DRAM is significantly slower than SRAM but it is often used
because of its low cost and smaller
memory cell size. There would be an enormous amount of interest in a device with
SRAM performance at cost
and densities of conventional DRAM. Quantum storage-based memory has the potential
to accomplish this.
There are two methods of achieving quantum storage. One is to use logic
devices whose operation is based
on quantum phenomena to make memory circuits. The other is to use a single device
that exhibits memory
characteristics using quantum phenomena. It is difficult to imagine a single device
that is robust enough to
perform all of the functionality of a memory cell so most cases are a hybrid of
these two methods. In either
¨
nm across and 10 nm from its nearest neighbor. Interaction between cells is
governed by Schrodinger’s wave
equation
where H is the Hamiltonian, ψis the wave function solution, E is the energy
eigenvalue, E0 is the on-site energy
which is the same for all sites, ni,σ is the number operator generating an
electron at site i with spin σ, ti,j is
the tunneling energy from site i to site j , and EQ is the coulombic
energy required to have two electrons of
opposite spin on the same site. The interactions between cells are predominantly
controlled by the coulombic
perturbation between neighboring cells so that the Hamiltonian at cell 1 due to
cell 2 is given by
where
where ρ is some assumed fixed positive charge to achieve space charge neutrality.
The polarity of a cell may be
defined as
where ρn is the charge at location n. This is the numbering scheme for calculating
the polarization of each cell,
and it has values in the range of −1 to 1. The lowest energy and preferred solution
occurs when neighboring
cells align themselves with the same polarization. If an initial polarization is
supplied by a driver to a QCA
quantum interconnect, within a couple of cells alignment to a polarization of 1 or
−1 occurs and continues to
subsequent cells (1).
Fig. 1. Numbering scheme used to determine the polarity of a cell. In a QCA wire’s
lowest energy configuration, each cell
has the same polarity.
A range of logic circuits has been designed using QCA including fundamental
AND and OR gates as well
as more complicated circuits like a full adder. It may be dangerous to make
assumptions about the state of a
cell when the polarizing effects of the driver are removed, say to infinity. Since
work is done in reading the
state of a cell the very act of determining its state may change the polarity of a
cell or an entire QCA wire in
an unpredictable way. In order to reliably achieve memory functions, a memory cell
constructed of basic QCA
logic elements must function without making any assumptions about the volatility of
the data in each cell.
Figure 2 shows an SR latch composed of two NOR gates constructed from QCA
logic. One NOR gate is
outlined for reference. The cell size is about 0.3 µm by about 0.4 µm. This quantum
dot-based SRAM cell is
comparable in size to current conventional DRAM cells.
A more critical assessment of this logic family should, however, be
considered. For instance, the thought
experiment used to determine how the QCA operates is carried out at T = 0 K. The
devices could be tested
at 4 K but would probably have to work at least at 77 K (liquid nitrogen) to be
useful. In addition, for space
charge neutrality there must be a charge of +2q (where q is the magnitude of the
charge of an electron) on
each cell. This suggests precise placement of exactly two donor dopant
ions in each cell would be required.
Unfortunately, current doping processes always produces a distribution of dopant
atoms. Expected statistical
variations in doping would significantly change the electrical
characteristics of each cell. Finally, although
direct write methods of photolithography do have the resolution required to create
these patterns of dots, these
methods have not been used for production. At the time of this writing QCAs have
not been demonstrated in
the laboratory.
Resonant Tunneling Diode-Based Memory. Another example of a
novel family of logic that may be
used to create memory is based on resonant tunneling diodes (RTD s). RTDs are
ultrahigh frequency, generally
two terminal, majority carrier devices composed of a heterostructure quantum well
usually built from III/V
materials using MBE (molecular beam epitaxy). The quantum well is usually on the
order of 5 nm in width,
formed by sandwiching a low conduction/valence band offset material (e.g.,
GaAs or Inx Ga1 −x As) between
barriers made of high conduction/valence band offset materials (e.g., AlAs). These
devices are characterized
by an “N”-shaped current–voltage (I/V ) curve with a peak in current
when the bias aligns a large carrier
population at the contacts with a resonance in the well (6). A valley in the I/V
curve occurs when the applied
bias aligns a very small carrier population in the contacts with resonances in the
quantum well. Hence, the I/V
curve of these devices may have negative differential resistance (NDR ); an example
of the I/V curve of a typical
AlAs/InGaAs RTD is shown in Fig. 3. Since transport is ballistic, to the first
order, carriers do not scatter from
----------------------- Page 455-----------------------
essentially no current would flow at all. Operating by ballistic transport they may
be assumed to be, and are,
high-speed devices.
The switching characteristic used to achieve memory operation from
RTDs can be divided into three
categories. The first category is load line switching, where more than one
intersection may occur between the
I/V characteristics of a nonlinear device and the load line. The history of the
operation of the device governs
which of these intersections is the operating point. This may be used
to make logic. The second category
is intrinsic bistability which occurs in some devices where charge is
stored in the device during rising bias
and removed during decreasing bias. This induces a hysteresis loop in the I/V
curve, producing two possible
operating voltages at a given current. Both load line switching and intrinsic
bistability produce volatile memory.
A third category, memory switching, refers to a case in which there are two stable
states maintained with no
power dissipation and without input, and hence, is nonvolatile. Any of the
discussed approaches might be used
to create memory devices since they provide multiple operating voltages at the same
current dependent on
prior history.
Load Line Switching and Intrinsic Bistability Devices. A
number of attempts to make fast RTD-
based logic devices have been made (7) that are based on load line
switching. Load line switching RTDs
maintain their state only while under load. Unfortunately, as isolated devices they
generally have poor terminal
characteristics. Various schemes have been used to overcome these shortcomings by
adding conventional devices
to the memory cell. A representative approach is shown in Fig. 4 (8).
This design from Texas Instruments
employs “ultra-low” current density RTDs that dissipate 50 nW of standby power.
These devices are constructed
in the InGaAs/InAlAs and InP materials system. HFETs (heterostructure field effect
transistors) are used for
the switching transistors. This cell does not require a refresh cycle, giving the
functionality of an SRAM cell
Fig. 3. RTD I/V curve. The peak current is about ±0.9 V and the valley is about
±1.2 V. An example load line is shown
with drive voltage V and resistance R . The intersection (large dot) between this
load line and the I/V curve of the RTD is
the operating point. Multiple intersections between the two curves occur for a
range of values of V (and R ).
while occupying an area comparable to a DRAM cell. Access times below 0.5 ns have
been demonstrated. This
architecture lends itself to vertical integration that aids in achieving high
density. Multiple valued logic can
also be achieved by using cascaded RTD structures. The device count to perform
basic logic functions is much
lower than a standard SRAM cell (typically six transistors). Variations on this
scheme use cascaded RTDs to
achieve multilevel logic and intrinsic bistability (9,10).
Memory Switching. Memory switching is less developed in RTD
research for making memory cells.
A number of nonquantum memory switching devices have been reported in
the literature. Since they are
nonvolatile and static the device count to create functional memory cells
is very low. Langmuir–Blodgett
film/metal sandwiches, for instance, show conductivity changes due to light
irradiation or applied voltages
(11). For electrically switchable metal/film/metal structures a preliminary high
voltage is applied as a forming
procedure; these devices then display memory switching characteristics. A film
thickness of 18 nm has been
used and the off-state resistance is 106 # while the on-state
resistance is 20 to 100 #. The structure of
the film may actually change during switching to cause this phenomenon. This device
is simple to fabricate
and is not necessarily incompatible with standard silicon fabrication procedures.
One negative possibility is
that switching is due to filamentary pathways. Sandwiches of
metal/SiO2/metal have also been observed to
show memory effects after a similar forming procedure, but these have been
attributed to highly conductive
filamentary pathways of heavily doped silicon which form and are destroyed during
switching (12).
Typically, observations of memory switching are difficult to explain. Charge
trapped in interface layers at a
metal/semiconductor boundary can result in memory switching. Such devices are
generally not useful because
they are difficult to reproduce. In any case, two terminal devices that
change resistance when a voltage is
applied are difficult to use without supporting active electronics. For instance,
on/off state resistances must not
be ambiguous when observed through an interconnect of varying length/resistance to
any particular memory
cell.
Flash memory is a conventional type of memory that attempts to use trapped
charge in a reproducible
way. In this device a floating gate accumulates charge that affects the source drain
current characteristics of
the device. In the EEPROM (electrically erasable programmable read-only memory) hot
electrons are used to
reset the memory state of the device. They do have comparatively very long write
times and some long-term
deterioration with the number of write cycles. Even so, for a variety
of applications these are very useful
devices.
A flash memory cell which accumulates charge on a floating quantum dot above
the channel has been
built by IBM (13). This device is an excellent example of the concept
of a quantum device whose operating
principles become more favorable as it becomes smaller. The change in threshold
voltage (#Vt ) resulting from
charge trapped in the quantum dot increases as the quantum dot size
decreases. This device is interesting
because it has very small off currents in the 10 pA range, small operating
voltages, and potentially excellent
scaling characteristics. It is not clear whether write times, which have always
been the major drawback for
electrically erasable flash memory, will improve in this design.
Quantum Storage Device. Gullapalli and Neikirk have proposed a
method to engineer charge trap-
ping in an attempt to build a memory switching device. The QSD (quantum
storage device) is a modified
quantum well diode that relies on the interaction of a quantum well region with N −
/N +/N − doped layers to
achieve multiple conduction states (14,15,16). Unlike other multiple state quantum
structures, the QSD has
different current versus voltage curves corresponding to the different conduction
states. Preliminary experi-
mental findings indicate that these states remain stable even under short circuit
conditions and can only be
switched from one state to another with the application of bias in excess of
certain threshold voltages. Further-
more, calculations using a self-consistent coherent tunneling model indicate that
it is possible to design QSD
cells with more than two states, creating the possibility of multistate logic and
multibit storage (15,16,17).
Fig. 5. I/V curve for a QSD showing four cycles of the memory switching operation.
The two curves are multiple solutions
for the same device. These solutions are accessed by applying a switching voltage
of about ±2.0 V to the terminals.
a thin, lightly doped n-type GaAs layer surrounded by thin AlAs barriers has been
used. The double barrier
quantum well structure is then placed within a few nanometers of an N
− /N +/N − doped GaAs structure
that itself forms a shallow quantum well. The doping sequence is crucial
to device operation. Only devices
with quantum interference between a barrier structure and a well with a highly
variable charge distribution
show evidence of multiple conduction curves. In particular, conventional DBRTDs
(double barrier RTDs) with
monotonically doped layers show no evidence of multistate behavior.
The device possesses memory, in that once the device is placed in one state
it will remain in that state
over a wide range of bias voltages, including zero bias (i.e., the multiple states
exist even at zero bias). Once
the device is placed on one branch of its I–V characteristic, it will
remain on that branch at zero bias. The
state of the device at zero bias (which can be sensed via the value of its
differential resistance), is determined
by whether the device was last switched to the high current curve or to the low
current curve. Even when the
device is completely disconnected from the bias supply, or its terminals are short
circuited, upon reconnection
the differential resistance is unchanged from its preset value. In this
sense, the device possesses memory
of its state, that can be retained without requiring any bias or
dissipating any power. Simulations suggest
that multistate behavior occurs in other structures including single and
triple barrier devices as long as a
N − /N +/N − doped layers are in near proximity to the heterobarrier.
at room temperature. Furthermore, there is a possibility that the QSD can serve as
static, nonvolatile memory
element or logic device with zero holding power, since the multiple conduction
states are stable for extended
periods of time even when completely disconnected from any power supply. Finally,
since the QSD is a simple
two-dimensional structure, memory cells may be stacked on top of each
other. This fact, coupled with the
possibility of more than two conduction states per cell, offers other possibilities
of achieving very high densities.
There are several problems that must be overcome before the QSD can be a
viable alternative to con-
ventional memory. First the switching characteristics of the two terminal devices
built to date are poor: write
currents are high, and on/off resistances are at best differ by a factor of two.
Resistive memory elements require
supporting electronics to make viable memory cells, as in Fig. 4. The devices that
have been made so far have
been in the GaAs/AlAs materials system which is costly and difficult to
integrate with conventional silicon
technology. Attempts have been made to develop a three terminal version by making
direct contact to the N +
layer (18) that may overcome some of the deficiencies in present devices.
Summary
BIBLIOGRAPHY
OLIN L. HARTIN
DEAN P. NEIKIRK
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W6117.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Wiley Encyclopedia of Electrical and
● Recommend to Browse this title
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Short-Term Load Forecasting
● Email this page Enter words or
Standard Article
Abstract
Stlf Methods
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W6117.htm17.06.2008 16:45:10
1600 The
second application of STLF is for predictive assess-
1400 ment of
the power system security. System load forecasts are
1200 an
essential data requirement for the off-line transmission
system
analysis for the detection of future conditions under
d 1000
n which
the power system may be vulnerable. This information
a
m 800 permits
the operating personnel to prepare the corrective ac-
e
D 600 Jan. 9 tions so
that the system can be operated securely.
400 Jan. 10 The
third application of STLF is to provide system op-
200 erating
personnel with timely information, that is, the most
0 recent
load forecast. They need this information to operate
1 3 5 7 9 11 13 15 17 19 21 23 the
system economically and reliably.
Hour The
technical literature displays a wide range of methodol-
model
parameters are estimated by using multiple linear re-
1. Very short-term load forecasting (minutes or hours
gression
or exponential smoothing techniques. These models
ahead)
are
structurally simple and require relatively low computa-
2. Short-term load forecasting (days to several weeks tional
effort. Dynamic models are based on a time series de-
ahead)
scription, or its equivalent state space description (10), of the
3. Medium- and long-term load forecasting (months to load
behavior. These models treat the load pattern as a time
years) series
signal with known seasonal, weekly, and daily periodic-
ities.
This gives a rough prediction of the load at the given
Very short-term load forecasting is mainly intended to pro-
season,
day of the week, and time of the day. The difference
vide load demand information in the very near future so that
between
the predicted and the actual load is considered as a
automatic generation control functions are appropriately ad-
1 2 p
y (t), Eq.
to a changing forecasting environment through the concept of
(2) can be re-written in the following form:
self-learning. ANN application in STLF has been an ongoing
research area and promising results have been achieved.
φ(B )y (t) = #(t) (3)
2 p
where #(B) # 1 # # B # # B # . . . # B .
STLF METHODS
1 2 p
where #(B) # 1 # # B # # B # # B .
STLF can be described in the following form:
1 2 p
as:
variables with the load variable to forecast. The estimation of
the regression coefficients is usually computed with the least-
y (t) = φ y (t − 1) + · · · + φ y (t − p ) + #(t)
square estimate technique. Statistical tests, such as the F-
1 p (6)
+ θ #(t − 1) − θ #(t − q)
statistic test, are applied to determine the significance of
1 q
these regression coefficients. The t-ratios resulting from these
tests determine the significance of each of the coefficients, and By
using the back-shift operator previously defined, Eq. (6)
correspondingly the significance of the associated variables
can be re-written as follows:
with these coefficients.
uate the response lag time, b, along with the orders of r and
φ(B )#(BS )∇d ∇Dy (t) = θ(B )#(BS )#(t) (9) s
of the polynomials #(B) and #(B).
S
where f (t) is the fitting function vector for the process, c(t) is
to be expressed in terms of its history and a white noise. If
the
coefficient vector, #(t) is a white noise, and T is vector
other variables affect the value of y (t), the effect of these vari-
transpose operator.
ables should be accounted for in the model. This can be
N −1
# # #
y (t) = ω(B )x (t − b) + #(t) (14)
wj y (N − j ) − f T (−j )c 2 0 < w < 1 (18)
σ (B )
j =0
where
σ (B ) = 1 − σ B − σ B2 − · · · − σ Bs (16)
cˆ(N ) = F −1 (N )h (N ) (19)
1 2 s
where
At any time t , there will be an estimate for the process
j =0
Eq. (32):
yˆ (N + l ) = f T (l )cˆ(N ) (22)
# T #
# #
yˆ (N + 1 + l ) = f T (l )cˆ(N + 1) (24)
y (k/k) = y (k/k − 1) + K (k) y (k) − H (k)y (k/k − 1) (33)
where F #
lim F(N ). The L matrix is called the transition
N# #
where y (k/k) is the updated estimate and K(k) is the blend-
matrix and is constructed on the basis that the model will
ing factor.
have a fitting function satisfying the following relationship:
The error associated with the actual and the posteriori es-
# T #
efficient models. In this method, the load is modeled as a state
E e (k/k)e (k/k) = P (k/k) (35)
variable by using state space formulation. The state space for-
mulation is designated by two sets of equations.
The blending factor K(k) is computed such that y (k/k) is
# #
measurement error that is a white noise with a know covari-
P (k/k) = 1 − K (k)H (k) P (k/(k − 1)) (37)
ance R(k).
The covariance matrices for vectors W(k) and V(k) have the
4. Calculate the apriori estimate y (k # 1/k) and the error
following properties:
covariance matrix P(k # 1/k) associated with it
# T #
Q(k) i = k y ((k + 1)/k) = #(k)y (k/k)
(38)
two hidden layers are sufficient for the STLF problem. The
a human expert in the field. The knowledge engineer extracts
from the input layer to the hidden layers and from the hidden
are acquiring expert knowledge in the STLF domain and or-
within the same layer or from the input layer directly to the
ables that affect load values need to be identified first. The
the same as the neuron input. For each neuron in the hidden
variable identification. Relating the variables and the electric
ship between the variables and the electric is not static, but
i
is dynamic. This means that the same magnitude of tempera-
ture changes affect load values to different degrees for differ-
where i is a neuron in the preceding layer, oi is the output of
ent day types and seasons. Numerical values are expected
node i, and w is the connection weight from neuron i to neu-
ji
outputs of an expert system based STLF. Therefore, some pro-
ron j . The neuron output is given by Eq. (41):
duction rules may be qualitative and others are quantitative.
As was mentioned previously, the expert system based
1
o = = f (net , θ ) (41)
model is used to emulate the knowledge, experience, and ana-
j 1 + e−(netj +θj ) j j j
logical thinking of an experienced human expert. An advan-
tage of a human expert is his/her capability to learn from
where net is the input signal to node j and # is a bias. In the
j j
new cases and to upgrade his/her knowledge and reasoning.
training process, # can be regarded as a connection weight
j
Therefore, self-learning capability is a desired feature of an
between node j and a fictitious node whose output always re-
experts system.
mains at unity.
#ω (p ) = ηδ o + α#ω (p − 1) (42)
The nodes in the input layer take the input signal, and the
ji pj pi ji
δ = (t − o )o (1 − o ) (43)
pj pj pj pj pj
⋅⋅⋅⋅⋅ Output layer
pj pk k j pj pj
M
#
Ep = (t − o )2 (45)
2 pj pj
Figure 2. A multilayer feedforward artificial neural network.
j = 1
i i i
1 1 n n 0
y (t) = yb (t) + yt (t) + yr (t) + #(t) (46)
T
a
b . . . a b ] .
1 1 n n
n i
where RYY(k) is the auto-covariance of #y (t)#, #Y is the variance
cates that the relative significance of the harmonic in the
of #y (t)#, J is determined such that the R (J ) calculated us-
1 YY 1
GES model is above average compared to other harmonics.
ing the available sample data is statistically reliable. Usu-
ally, J1 # N/M with M greater than 10 (M # 13 is found to
GES model does not take into account the effect of weather
number of data used for modeling.
1 j 2
are defined by:
yt (t) = c0 + # # # cijk T (t − i)k (52)
j =l k =l i=0
P (ωi )
P (ω ) = (50)
n i Pav
where L 1 is the number of weather variables to be modeled,
ijk j
j =l k =l i=l
#
where tilde parameters refer to the estimated values. The
random series, #yr (t)#, can be considered a stationary time se-
where # is the estimated variance of the residuals of
#
ries and can generally be modeled by an ARMA model.
the p th order AR model. If AIC (p ) # AIC (p # 1), go to
ARMA modeling methods have been used in STLF
step 2; otherwise, the optimal parsimonious AR model
(1,9,26,27). It has been shown in (9) that AR models have
is obtained.
forecasting quality equivalent to ARMA models. Promising
forecasting was obtained using AR models (26). In addition,
WRLSE Algorithm. For a model expressed by:
parameter estimation of AR models is computationally much
less demanding than that of ARMA models. This is an attrac-
y (t) = H (t)TX (t) + #(t) (57)
tive feature for the consideration of developing AR models
with an enhanced modeling algorithm for use in this example.
set of equations:
by one. This process continues until the sum of squares of
# #
the model residuals reaches minimum. Fixed AR models were
T
R (t − 1) x (t)x (t) R (t − 1)
utilized in (26,27). An improved adaptive AR modeling algo-
R (t) = I − T (58)
t
general form of an AR model of order p can be described by:
S (t) = # λ(j )t −j #(ˆ j )2 (62)
!
p
1 #(ˆ t)2 #(ˆ t)2
qi
λ(t) = 1 − Q(t) − + 1 − Q(t) − + 4Q(t)
where #(B) # 1 # #i #1 # B , # is coefficient of the AR process,
qi qi
2 S (t) S (t)
B is back-shift operator, qi refers to the ith lagging term with
q hours preceding the current hour in the AR(p ) model. It is
(63)
determined by partial autocorrelation analysis.
An outline of the proposed adaptive AR modeling method
Test Results
follows.
of their magnitudes.
30.6#C (#23#F). Utility A is a summer peak system with one
----------------------- Page 471-----------------------
{1, 2, 3, 6, 7, 8, 13, 14, 15, 20, 21, 22, 28, 35, 49}
racy from the proposed method ranges from 10 to 30% for the
Nonlinear temperature sensitive model:
four time periods on average. This is a result of using the
23 to
Dec. are larger than those for the other three periods. This
yr (t + 1) = (θ + θ B + θ B )yr (t)
0 1 23 is
because of holidays such as Thanksgiving Day, Christmas
Day, and the New Year Eve. When compared to the fixed
Table 1 shows absolute mean error (AME) and standard
modeling approach, the ratios of improvement in accuracy
deviation error (SDE) statistics of 24 h for Utility A by the
reach up to 34% on average. Considering the importance of
proposed method. Highly accurate forecasting was achieved
peak load forecasting, this improvement is quite significant.
with seasonal average errors less than 1.5%. Compared with
By using the proposed adaptive modeling techniques, the
the fixed modeling method, the ratios of improvement in accu-
identified models for Utility A in two consecutive weeks of
April follow.
Hour AME SDE AME SDE AME SDE AME SDE {1,
2, 3, 5, 6, 7, 8, 9, 10, 11, 12,
9 24
ing. They change over time. Similarly, various updating tech-
Wed.: yr (t) = # θ yr (t − i) + # θ yr (t − i)
niques can be used to alleviate this error. While each of the
i i
i= 1 i=21
statistical STLF approaches demonstrates considerable suc-
8 24
Thu.: yr (t) = # θ yr (t − i) + # θ yr (t − i)
cess in forecasting accuracy, they are all subject to the risk
i i
i= 1 i=21
that the future cannot be fully reflected by history. This is
8 23
# #
i i
i= 1 i=23
Expert systems belong to this class of methods. Expert sys-
5 24
# #
M. KADDOURA
9. S. Vemuri, W. L. Huang, and D. J. Nelson, On-line algorithm for
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3036.htm
} { { } }
HOME
●
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Solar Energy Conversion
● Email this page Enter words or
Standard Article
Abstract
Heterogeneous Photocatalysis
The
fossil fuels such as coal, oil, and natural gas, which main-
tain
our industrialized world, will be surely running out
2 2 x
ing
and acid rain problems. The development of alternative
these are the low cost performance, the low energy conversion
possible.
2 2
the
conversion efficiency is not high, the plant photosynthesis
3.
If solar energy conversion is aimed, the visible light
Product
y
should be
used for photochemical reactions. In most
g
r ν
cases, however, photodissociation (a) and photoioniza-
e h
n
tion (d) proceed only with ultraviolet light. In such a
E U
Photosynthesis
Photosynthesis and Solar Energy Conversion. Green plants
are converting CO from the atmosphere and H O from soils
2 2
into organic substances, utilizing solar energy. This is what
Stroma
we call photosynthesis of green plants (2). The energy stored
in organic substances produced by plants is about 0.1% of the
Chloroplast
total annual solar energy reaching the Earth and is over 10
times the total amount of energy consumed by human activi-
ties per year. Organic substances produced by plants serve as
Thylakoid
food for human beings and animals, which is the energy
membrane
source for activity and growth, organic raw materials for in-
dustrial production, and fuels. Fossil fuels can also be said to
have resulted from the photosynthesis activity of ancient
plants.
Thylakoid
and CO2 fixation takes place in stroma, the substrate part of the chlo-
where C H O is hexose. Thermodynamically, the right side
roplast.
6 12 6
+ +
NADP + H NADPH
ADP + P ATP
PSII
PSI FD
Outside
QA
QB
ν
h
Pheo
PQ
ν
Chl h Tylakoid
membrane
P680
Mn Cyt f
P700
Z
H+ H+
Inside
H O 1/2 O + 2H+ H+ PC
H+ H+
2 2
A B Q
PSI. P680 is recovered from photooxidized P680# by
accepting an electron from an H O molecule,
ATP # ADP # Pi .
PSI
– 1.0
Chl
– 0.8
PSII
Several
– 0.6 Pheo acceptors
ν
h
FD
– 0.4
QA ADP + Pi
– 0.2
QB NADP
NADPH+
ATP
V
/ 0
E PQ acceptor
ADP
ν
h
0.2 + Pi 1/12(C H O )
1/2 CO
6 12 6
2
0.4 ATP
Cyt f PC P700
0.6
0.8 H2O
Mn
Figure 4. Schematic
illustration of electron flow and
Z
1.0 1/2 O2 + 2H+ P680
chemical reactions in redox potential scale. See Fig. 3
legend.
diates.
CH3 C2 H5 As
illustrated in Fig. 4, ATP is also synthesized in chloro-
plasts
by phosphorylation of ADP, namely, the reverse pro-
N N
cess
of Eq. (14). Since this ATP synthesis is achieved by utiliz-
H Mg H ing
energies released in downhill transfer of electrons that
have
photoexcited in PSI and PSII, this process is called pho-
N N
tophosphorylation. Solar energy is converted to chemical en-
ergy
here again and stored in ATP.
CH3 CH3
H
Artificial Photosynthesis. A reaction scheme of photosynthe-
H
sis
can be represented by the following equation:
CH2 H
hν
CH2 COOCH3
# ∗
COOC20 H39 O2
+ 4H+ C− T2 PS → PS
2
Figure 5. Molecular structure of chlorophyll a.
2H O C T− PS+
2 2 2 (16)
T1 C− CO2 + 4H+
T− C 1/6(C H O ) + H O
Q , Q , PQ), cytochrome f (a heme protein, Cyt f), plastocya-
1 1 6 12 6 2
A B
photoionization quantum
yield of PTH was obtained in the
+ SDS
micelles than in homogeneous methanolic solutions, and
CO + 3ATP + 2NADPH → 1/6(C H O ) + 2NADP
2 6 12 6
electron transfer from PTH in the micelles to Cu2# and Eu3#
+ 3ADP + 3P + H O (15)
i 2
adsorbed on the micellar surface occurred at higher rates
h
Ru(bpy)2# were placed on the inner and outer surfaces of di-
CO
3
HCHO
hexadecyl (DHP) vesicles, respectively.
CH3OH Eg
Intramolecular photoinduced charge separation and elec-
CH4
tron transfer is another challenge. Electron transfer has been
+ + + + O2
recognized in a synthesized triad consisting of a tetraaryl-
porphyrin (P) covalently linked to both a quinone (Q) and a
+4H+
carotenoid polyene (C). Excitation of C-P-Q by a picosecond
Semiconductor 2H2O
Oxidation
laser pulse resulted in a rapid formation of charge-separated
process
species, C# # -P-Q# # , with a quantum yield of 0.25 and a life-
time of several microseconds. Very recently, photoinduced
Figure 6. Schematic diagram of a photocatalysis process for reduc-
electron transfer with a quantum yield of 0.83 and a lifetime
tion of CO2 to methane and others by a semiconductor photocatalyst
of 55 ms has been achieved in a molecular pentad consisting
loaded with a metal or metal oxide.
Photosplitting of Water
mechanism in the photosynthesis involving water oxidation
and CO reduction is explained on the basis of electron trans-
Figure 7 shows the positions of a redox potential of H O and
2
2
2 2
2
O
P
r
a e
Z
G S
–1.0
e d i
S T C S
3
d d
O
C C 5 3
i
2 3 2
T
O 3 O
r O
2 O 2 O O
S i
b O 2
T
V V V V n e i n
e e e e N Z W F B S
0.0
H /H O
2 4 7 1 2 2
. . . .
2 1 1 1
V V V
V Conduction band
e e e
e
V V V V V V
0 2 0
5 #e e e e e e
V . . .
.
e +1.0 5 3 3
2 4. 2. 5. 2. 6. 5.
/
3 3 2 2 2 3
E
H
O /H O
2 2
N
s
v
V ####
+2.0 ##
####
+3.0
##
###
pose water into H2 and O2 in the stoichiometric ratio since the ria
killer, and their applications in organic synthesis and even
products of H2 and O2 are separated between different layers in
cancer treatment are currently very attracting.
to restrain their recombination. Ion-exchangeable layered
Perovskite type niobates show interesting photocatalytic ac-
tivity with visible response (8). Niobates and titanates with WET-
TYPE SOLAR CELLS
R
ate water splitting for various oxide semiconductor catalysts.
A
The formed carbonate radicals in the system produce peroxy-
carbonates by self-coupling or reacting with OH # radicals, re-
V
sulting in the irreversible evolution of O2 gas and the easy
detachment from the surface of the catalysts. Decomposition
3
of water by an AM1 solar UV light irradiation has been tried
+ 1 2 –
for the first time and found to take place efficiently as well as
stoichiometrically in the presence of NiOx -TiO2 catalyst (11).
Dye 4
Dye
The present problems in photocatalysts for the complete wa-
h
ter splitting are the low efficiency and its difficulty in utilizing
– Diffusion
the visible light.
e Ox
Dye* Dye e–
Other Applications
+
Photocatalysts, especially TiO2 due to its oxidizing power and
Diffusion
photochemical stability, have been utilized for various pur-
Ox Red Red
Photogalvanic Cells
V
ν
and a redox system either as an electron acceptor or as an
3 h
–
electron donor. In such a cell, light is absorbed by the dye.
e
+
ing products. The formed products then give or accept elec-
VB p
trons to or from electrodes, generating photovoltage. Using
this cell, one can only convert solar energy into electrical
energy.
Taking an electrolyte containing a redox couple Fe3#/Fe2#
(a)
and dye Ru(bpy)2# as an example, the following equilibrium
3
is established in the dark:
R
A
Fe3+ + Ru(bpy)2+ ⇔ Fe2+ + Ru(bpy)3+ (17)
3 3
V
2+ ∗ 2+
3
Ru(bpy)3 + hν → Ru(bpy)3 (18)
e– Ox
h
Fe3+ + ∗Ru(bpy)2+ → Fe2+ + Ru(bpy)3+ (19)
3 3
Red
+
The equilibrium in the dark is then broken. The resultant
VB p
increase of the concentration of Fe2# and Ru(bpy)3# changes
3
the redox potential: that of Fe2#/Fe3# is shifted negatively,
while that of Ru(bpy)2#/Ru(bpy)3# is shifted positively. As a
3 3
2#
(b)
result, Fe gives an electron to one of the metal electrodes,
and Ru(bpy)3# accepts an electron from the rest metal elec-
Figure 9. (a) n-type and (b) p -type semiconductor photoelectrochemi-
3
ph
E ' E C
F, m
eU E
E C
ph F, S Ox e–
E E E E
E E
F, m F, S F, red ox F, m
F, red ox F, m
EV
Red
p
EV
(a)
(b)
E C
E C e–
Ox Red h+
E E
E F, m E F, S E F, red ox E F, m
F, red ox F, m
EV
eU
ph
(b)
E
F, S
E 'F, m EV p+
(c)
(d)
ph
0 0
On the other hand, the holes created by light are transferred
bulk, #p is the increment of p 0 on illumination, and n0 is the
across the semiconductor/electrolyte interface to the reduced
majority carrier density. The highest possible photopotential
species:
depends on the band bending below the semiconductor sur-
ph. max g
time.
Improvement of the Conversion Efficiency. In order to
Doping of Noble Metal Nanoparticles in the Semiconductor
achieve a high conversion efficiency for the semiconductor
Electrode (18,19). One can obtain visible photoresponse by in-
PEC solar cells, the semiconductor electrode should meet the
corporation of noble metal nanoparticles or co-incorporation
following four requirements: (1) an optimum bandgap of 1.5 of
metal nanoparticles and dye molecules in the wide bandgap
to 2.0 eV, which responses to the solar spectrum, (2) a conduc-
semiconductor electrodes (18,19). The mechanism is based on
tion band edge more negative of the reduction potential of
the enhancement of the dye sensitization or visible-light-exci-
redox species in an electrolyte combined with a valence band
tation of electrons at the surface states within the bandgap of
edge more positive of the oxidation potential, (3) rapid charge
semiconductor by the surface plasma resonance of the metal
transfer kinetics to the electrolyte, and (4) strong resistance
nanoparticles. For example, visible photoresponse was found
to corrosion of the semiconductor surface. No semiconductors
for Au or Ag nanoparticles doped TiO2 electrodes, which was
investigated so far satisfy all of the requirements. As men-
ascribed to the electron excition from the surface states en-
GAOLING ZHAO
Zhejiang University
HIROMITSU KOZUKA
HONG LIN
TOSHINOBU YOKO
Kyoto University
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3037.htm
} { { } }
HOME
●
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Solar Heating
● Email this page Enter words or
Standard Article
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3037.htm17.06.2008 16:46:07
Issues
relevant to each of these are discussed separately in
the
following sections.
SOLAR
RADIATION AND ITS DISTRIBUTION
Background
The
solar constant G , defined as the energy flux (or irradi-
ance)
incident on a surface perpendicular to the incoming so-
lar
radiation outside the atmosphere of the earth at the mean
earth–
sun distance, has been measured by numerous re-
the
elliptical orbit of the earth are on the order of 3%. Addi-
The world’s supply of fossil fuels is finite and is expected to tional
fluctuations caused by sunspot activities over its ap-
be depleted in the foreseeable future. Though nuclear energy
proximately 11 year cycle are believed to be less than 1.5%.
systems are not similarly constrained, particularly since the
For engineering applications, the solar constant is as-
development of breeder reactors, nuclear waste problems still sumed
to be fixed even though small changes have been re-
remain unresolved. The future of fusion technology, where the corded
in addition to the normal variations described. This
fuel is essentially unlimited, is also uncertain at the present
assumption has no impact on solar system design because the
time. As a result, there is a great potential for alternative energy
incident on the surface of the earth differs signifi-
renewable energy sources, such as solar energy. cantly
from the solar constant. Scattering effects are caused
Humankind has a very long history of solar energy utiliza- by air
molecules (Rayleigh scattering, mostly at wavelengths
tion dating back to prehistoric times. The total solar radiation below
0.6 #m) and by dust/water particles (Mie scattering).
incident on the surface of the earth is on the order of 80 tril- In
addition, absorption by ozone, water vapor, and carbon di-
lion kW, which is more than 10,000 times the annual global oxide
has a strong impact on the overall transmission of radi-
energy consumption for all human activities at present (1). ation
through the atmosphere. As a result, the wavelengths
Although energy availability is not an issue, its direct use is of
interest for terrestrial applications are limited to the region
constrained by three factors: (1) the low energy flux (a maxi-
between 0.29 #m and 2.5 #m (Fig. 1, see Ref. 3).
mum of about 1 kW/m2); (2) the large variations and unpre-
Terrestrial solar radiation consists of beam (i.e., direct)
Wavelength ( m)
• the availability of solar energy at a given location
Figure
1. Extraterrestrial solar radiation and the effects of Rayleigh
• the energy collection and storage requirements
Preliminaries
correction terms on the order of 5% to 20% are usually neces-
sary because the ring shades the detector from some of the Collectors
are classified into two broad categories, nonconcen-
diffuse radiation (in addition to the direct component). Mea- trating or
flat-plate collectors and concentrator-receiver-type
surements of radiation on inclined surfaces are also made collectors.
In either case, it is first necessary to estimate the
with conventional pyranometers. However, the measure- solar
radiation incident on their surfaces before designing a
ments may need to be corrected by as much as 10%, de- solar
thermal system. For a horizontal surface, methods for
pending on the instrument used. estimating
the incident energy have already been discussed
in the
previous section. For inclined surfaces, the problem is
Estimates of Solar Radiation more
complicated, and the beam and diffuse components of
the incident
radiation must be considered separately.
Solar radiation data required for systems design are often un-
available. Thus, the average incident radiation is often esti- Radiation
on Inclined Surfaces. The beam component of radi-
mated from empirical equations developed for this purpose. ation is
obtained directly from the sun. Thus, it can be evalu-
These were originally of the following form (5): ated from
geometric considerations once the angles relative to
the
horizontal plane are known. Any specular reflection from
H n
= a + b (1) the ground
or the surroundings can also be similarly calcu-
Hc N lated as
long as the reflecting surface locations/configurations
are well
defined. The diffuse component is more difficult to
where H is the irradiation on a horizontal surface, (i.e., the evaluate
because a nonhorizontal surface receives radiation
energy incident per unit area integrated over a specific time by
scattering from the sky and by (diffuse) reflection from the
period e.g., year, month) where the subscript c refers to a ground.
Further complications arise because the diffuse com-
‘‘clear sky’’ terrestrial value, n is the daily number of hours of ponent from
the sky consists of three separate components,
Q = A [S − U (T − T )] (3)
A solar collector is a heat exchanger that collects thermal en-
c L p a
A [S − U (T − T )]
and low emittances above 3 #m (for a surface at 200
C, 99% c L i a
of the emitted blackbody radiation is at wavelengths greater
than 3 #m). This ensures that a very high fraction of the inci- where Ti
is the inlet fluid temperature. Because the tempera-
dent radiation is absorbed and the energy emitted is rela- ture of
the fluid is always lower than the temperature of the
tively low. The overall enclosure is also well insulated at the absorber
plate, FR is less than (or equal to) the collector effi-
rear to further minimize any heat losses to the surroundings. ciency
factor F
. F
is defined as the ratio of the fluid-to-ambi-
ent heat
transfer coefficient Uo to the plate-to-ambient (i.e.,
the
collector loss) heat transfer coefficient UL :
Then the
mean plate and fluid temperatures are related to
the
previous three parameters as follows:
Q/A c # FR #
T = T + 1 − (6)
f i F U F #
R L
T = T + (1 − F ) (7)
P i F U R
Figure 2. Schematic of a simple liquid solar collector.
R L
C = (10)
A r
Performance Testing. Performance standards have been de-
veloped for solar collectors as their commercial market has
where A a is the aperture area (i.e., the projected area of the
increased in recent years. Three parameters are used to char-
concentrator) and A r is the receiver area. This is a measure of
acterize the collector performance: the instantaneous collector the
energy concentration whose maximum values are given
efficiency, incident angle modifier, and the time constant. The by
first is a measure of the energy absorbed by the collector rela-
tive to the energy incident on it:
1
Concentrator-Receiver Collectors
In
phase-change energy storage systems, the storage me-
dium
undergoes a change of phase during the heat
ceiver, below the absorber plate, so that the operating temper- with
risers and headers made of special materials so that they
ature of the system is more uniform. can
withstand occasional freezing.
bypass, as required.
tion, solar energy has also been used for cooking and a range
A
number of specialized solar water heating systems have
of industrial and process applications, such as distillation,
been
proposed over the years. One example is the hybrid solar
salt/brine production, drying, detoxification, power genera-
Water heating is probably the most common solar heating ap- oped
to obtain increased stratification in the storage tank.
plication. A typical configuration of a residential water heat- Such
a design provides higher performance at significantly
ing system is shown in Fig. 5. Residential water heating sys- lower
cost.
tems are of two basic types, natural circulation and forced A
second large residential application of solar energy is
circulation. In natural circulation systems, density differences
swimming pool heating. The general requirements of this ap-
due to the temperature rise of the fluid in the collector(s) in-
plication are similar to those of water heating, though there
duces a flow in the system. Because this driving force depends are a
few significant differences. The most important of these
on the heat absorbed, natural circulation systems are self- is
the set-point temperature which is relatively low for a
adjusting to a certain extent. The temperature rise in the col-
swimming pool. On the other hand, the total thermal energy
lector is about 10
C over a wide range of operating conditions. required is usually
much larger than that for a residential
In forced convection systems, a separate collector-side heat water
heating system because the mass of water is much
transfer loop is often employed, and a separate pump trans-
greater. As a result, the simplest approach is to use the pool
ports the heat transfer fluid in this loop. With this approach,
itself as the collector by covering it with a solar transparent
it is possible to use a heat transfer fluid other than water. (and
IR opaque) plastic cover. Alternatively, separate collec-
This is particularly advantageous in regions subject to freez- tors
are used in series with the pool pump to heat the water
ing because then antifreeze mixtures (ethylene glycol–water, as
required.
propylene glycol–water) or air can be used in the collector
loop. Furthermore, because the boiling point of antifreeze
Space
(Building) Heating
mixtures is higher than that of water, such systems can also
be designed to operate at higher temperatures without in-
Building (residential or commercial) heating systems are of
creasing the system pressure. three
general types: passive, active, and hybrid systems. In
Freeze protection is an important issue for solar water
passive ‘‘solar houses,’’ building elements are designed so that
heaters in many parts of the world. If antifreeze mixtures
individual components also absorb, store, and release thermal
cannot be used because of cost or other constraints, systems
energy as required. In one approach, the building may include
may be designed so that the water can be drained back to the
collector-storage walls that absorb and store incident radia-
tank or out of the system below a set temperature. Alterna- tion.
In such a design, glazing is put on part of the south wall
tively warm water can be circulated through the collectors, if and
the wall itself is painted black (i.e., with a solar radiation
freezing temperatures are rare. Systems can also be designed
absorbent paint). To maximize energy storage, the wall is
Cold water
supply
Vent
T
c
DHW
Collector
Doublewall
heat
exchanger
Antifreeze
T
s
solution
Pump
Figure 5. A typical residential water Pumps on when
Pump Storage Auxiliary
T is 10 F above T
heater
heating system. c ° s
tank
the
heat pump is then used for conventional space heating.
systems.
Solar
Cooking
Living space
Solar
cookers have been investigated for many decades, and
some
low cost commercial models are available. These cookers
Air return Auxiliary can be
classified into three basic types. Hot-box designs con-
Distribution
heater sist of
an insulated box with double glazing in which the cook-
Fan ing
vessel is placed. Therefore the basic configuration is
Cover–condenser
Power Generation. Solar thermal power generation sys-
####
ergy management.
prises only about 4% to 6% of the total terrestrial solar radia- the useful
energy based on available weather data and an
tion. The overall collector is otherwise relatively simple and estimate of
the system losses (which must be less than the
requires no thermal insulation because the reaction rates are absorbed
radiation if there is to be any useful energy). For a
not a significant function of temperature. given system
design, these are typically quantified by a pa-
rameter known
as the utilizability, which is the useable frac-
tion of the
total incident energy available over any given pe-
MODELING AND SIMULATION riod (hour,
day, etc.).
the
performance of similar systems. This is the basis of the f -
Transient Models chart method
which is used to estimate the fraction f of the
A number of generalized transient models have been devel- total energy
required for a specific application that can be
oped by different research groups around the world [e.g., supplied by
solar radiation. This parameter is usually ob-
tained as a
function of the heat loss (X ) from the collector and
Commission of European Communities (EMPG2/EURSOL),
TRNSYS (United States), WATSUN (Canada)]. These are the energy
(Y) absorbed by the collector, both of which are
built on individual numerical simulation modules describing normalized
with respect to the heating load. Separate correla-
the transient performance of components that make up a typi- tions for f
are available for air- and water-based systems, and
cal (solar) thermal system. The modules are usually based on correction
factors which account for energy storage have been
a 1-D approach because the uncertainties associated with typ- developed.
Because the functions are obtained using a curve
ical solar thermal systems do not justify the increased compu- fit, the f
-chart method can be used only for the specified range
tational cost of more complex 2-D or 3-D models. Limitations of parameters
and where the minimum load temperatures are
of current multidimensional computer models for complex about 20
C. More advanced methods combining the utilizabil-
thermal processes also make the 2-D and 3-D models imprac- ity and f
-chart concepts have also been developed to overcome
tical in most cases. this minimum
temperature limitation but are not commonly
used.
The generalized transient models are implemented in the
form of large computer programs which include subroutines The
utilizability, f -chart, and related methods are directly
for the numerical solution of differential equations and utili- applicable to
active collector-based systems. In contrast, the
ties for mathematical analysis of various thermodynamic and heat
transfer process in passive ‘‘solar buildings’’ is related
transport processes. To simulate a given process/system, the closely to
its architectural design. Nevertheless, models based
modules describing the individual components are linked to- on the
utilizability concept and correlation-based methods
gether in series and/or parallel as required. Then the system have been
developed for such applications. Specialized charts
performance is studied by numerically integrating the set of and methods
based on past experience with passive heating
time-dependent differential equations with a given set of pa- are also used
in system design. These may be strictly empiri-
rameters and initial conditions. The results are output in var- cal or may be
based on more complete analysis of the building
ious forms, and with some programs, it is possible to conduct heat transfer
process. Similar detailed analyses may also be
an economic analysis of a given system/process using the re- required for
large active systems where the simple utilizabil-
sults of the simulation. ity and f
-chart methods are not adequate.
energy.
SANJAY K. ROY
Solar radiation data are available from various sources,
particularly the national meteorological offices of individual
PhD Research Group, Inc.
countries [e.g., National Weather Service (NWS), National
HONGTAN LIU
Oceanic and Atmospheric Administration (NOAA), and Na-
University of Miami
STORAGE.
Finally, it is also important to note that a significant frac-
BIBLIOGRAPHY
2. J. Lean, The Sun’s variable radiation and its relevance for Earth,
Ann. Rev. Astron. Astrophys., 35: 33–67, 1997.
11. T. A. Reddy, The Design and Sizing of Active Solar Thermal Sys-
tems, Oxford, UK: Clarendon Press, 1987.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3041.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Steam Turbines
Turbine Types
Turbine Control
Operating Problems
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3041.htm17.06.2008 16:46:32
STEAM
TURBINES
The
steam turbine is an engine designed to convert the ther-
mal
potential energy of steam to rotational kinetic energy on
a rotor
which is free to rotate when supported by a series of
axially
aligned bearings. This rotational energy is then used
to drive
some mechanism or machine, such as an electric gen-
erator,
pump, compressor, or ship propeller. Energy conver-
sion is
achieved by allowing for the controlled expansion of
the
high-energy (pressure and temperature) steam in a num-
ber of
series arranged stages, each stage allowing a portion of
the
total thermal energy to be released and then converted to
high
kinetic energy in a fixed blade row. The resulting jet of
high-
velocity steam is then directed into a following row of
blades
which are attached to a central spindle and caused to
rotate,
thereby driving an attached mechanical or electrical
device.
There can then also be a further energy release or
reaction
in the rotating row.
The
concept of expanding steam through some form of noz-
zle and
employing the resulting kinetic energy to produce
shaft
rotational energy has been recognized for centuries, by
Hero in
120 B.C. and later by Giovanni de Brance in 1629.
However,
it was not until the late nineteenth century that the
erating
life of the unit. However, those losses experienced
within
the steam path are a function of a number of complex
TURBINE TYPES
factors
both mechanical and aerodynamic, and the efficiency
within
the blade rows tends to decrease with time. A portion
Despite their apparent similarities, there are a variety of of
these steam path losses are recoverable when the unit is
types of turbine, and they can be classified in a number of opened
for inspection and refurbishment, but it is normally
ways, the most apparent being their type of application—that not
possible to recover the full losses which have been in-
is, electrical power generation, mechanical drive, or marine. curred
and associated with operation, except at very high
They can also be classified by size, arrangement, and output. costs,
and such improvement can normally only be sustained
A factor having considerable influence on the details of any for
relatively short periods.
turbine is the type of thermal cycle in which it is to be em-
Steam path efficiency, #sl, is defined as the ratio of useful
ployed, the energy source, and the steam conditions used in energy
converted to power in the unit or section to energy
that cycle. which
is available for conversion (1). This expansion in the
A significant advance in cycle arrangement was the use of steam
path is shown as in Fig. 1. The steam enters the stage
‘‘regenerative feed heating.’’ Feed heating is an arrangement
by which a portion of the steam entering the unit is removed
from the expansion, at a number of suitable steam conditions,
P
a
and used to preheat the condensate returning from the con-
A Ta
denser to the boiler. This feed heating is applied to raise the
Ha
temperature of the condensate to close to the saturation tem-
perature of the water entering the boiler. The boiler has then
only to supply the latent heat and superheat to the working
fluid.
y
A straight condensing non-reheat turbine is one in which
g
e
the steam enters the unit and expands in a series of stages to
y n
g l
a defined exhaust pressure. As discussed above, the lower-
r u
e f
pressure stages can be paralleled, but this has no effect on
n e
e s
the non-reheat character of the unit. There is also the reheat-
e U
b
type arrangement, in which the steam, after expanding to a
a
i
suitable pressure, is removed from the turbine, returned to
a
A
the boiler or other reheating device, has its temperature level
C
raised, and is then returned to the turbine to continue its
T
expansion. Some advanced cycles have been arranged to in-
Tc c
corporate two stages of reheating.
t y
s r
The reheat cycle provides considerable advantage in terms
P b o e
L n
of cycle efficiency, but suffers to a small degree from the addi-
e
steam to enter the unit and flow axially from inlet to exhaust, Figure
1. The steam expansion line of the turbine stage, as repre-
and while the direction of axial flow may be reversed in cer- sented
on the Mollier enthalpy–entropy diagram.
2
resents energy lost to the unit.
# P1 1 −
# P
#
The steam path losses can be considered to be influenced
1
m = k ·A e (2)
# P
#
by several diverse factors including (a) the design philosophy
V N + ln 1
used to define the various components comprising the steam
s1 P2
path, (b) the manner and quality with which these compo-
nents are manufactured to meet the design specification and where
then installed in the unit, and (c) the manner in which the
unit is operated and maintained. m is
the quantity of steam leaking in unit time
The actual losses can be divided into four principal groups, k is
a constant, including a flow coefficient
which are as follows:
A e
is the circumferential leakage area, # D # C
Aerodynamic Losses. The steam expands through the steam D is
the mean diameter of the seal
path at high velocities; and it flows through the blade ele- C is
the running clearance
ments, causing a frictional interaction with the surfaces with P1
is the upstream steam pressure
which it is in contact. As the unit ages, there is a deteriora- Vs1
is the upstream steam specific volume
tion of this surface due to a number of phenomena which can- P2
is the downstream steam pressure
not be entirely avoided. These include the passage of solid N is
the number of series strips
particles of magnetite (Fe O ) through the steam path, these
3 4
particles being carried into the unit from the boiler and other
As
a unit operates there are a number of phenomena, in-
portions of the steam cycle. Also the steam carried into the
cluding starting and stopping, load changes, and other tran-
steam path can have chemical compounds suspended in the
sients
which cause the rotor to vibrate to the extent the knife
steam, which, as the pressure decays, come out of solution
edge
of the seal strips rub and therefore wear. This action
and attach themselves to the steam path surfaces, thereby
causes
the clearance to open, with a resulting increase in the
increasing roughness and therefore increasing frictional
1, s1
In addition to these surface losses, there can be losses due
Inlet
to high velocities above Mach 1 where supersonic losses are
Discharge
induced. There are also losses associated with components in-
P2
###yyy#y
D
or missing blades at row closing windows. These components
cannot be avoided, but modify the flow and cause distur- Figure
2. Parameters of steam and geometry defining the laby-
bances in the steam path. rinth
seal.
###yyy#y
1 1 c
From boiler
reheater
H m
hr, r
To boiler
reheater
H m
cr, r
High
Reheat Low Generator
pressure
pressure
kW
h
H enthalpies, water h enthalpies, and
c
Return to
flow quantities ‘‘m’’ at various locations
boiler
throughout the cycle. The unit electric
Boiler
output ‘‘kW’’ is also shown.
feed pump
in
the stationary row and has a further drop in the rotating.
m (H )
1 1
Because a pressure differential exists across row blade rows,
suring that the discharge area from each blade row is con-
Figure 4. A smaller industrial or marine type unit, showing steam
enthalpies H and flow quantities m together with the output in kilo-
trolled within close tolerances and is a function of the quan-
watts ‘‘kW’’ or horsepower ‘‘HP.’’
tity of steam flow and the steam properties at inlet to the
defined as
γ − 1
m
Steam rate (SR) = (5)
where
kW or HP
In this definition, the input, in thermal energy, is mea- m
is the quantity of steam flowing in unit row
sured as the steam flow m required for a specified time be- A
is the row discharge area
tween steam properties H 1 and He to generate an electric out- Cd
is the discharge coefficient
put kW or mechanical output HP for a specified time period. g
is the gravitational constant
P1
is the row inlet pressure
P2
is the row discharge pressure
THE TURBINE STAGE
Vs1 is the row inlet specific volume
#
is the ratio of specific heats of steam
The basic turbine stage consists of two blade rows. The first
is stationary, consisting of a number of vanes which are ar-
In the practical turbine this area is achieved by defining
ranged to produce passages between adjacent elements. the
form of the individual blade vanes forming the expansion
These vanes are attached to the stationary portions of the
passage and then controlling their spatial relationship rela-
unit; they both expand the steam by allowing for a controlled
tive to each other in the final assembly.
pressure drop, and they are oriented so as to direct the re-
sulting jet into a following row of rotating blade elements.
tity which bypasses the blade row. Again, a seal is used above
0.50 provides the optimum efficiency, whereas for a stage hav-
which would occur through the tip section of the vane. These
stage, which will require a steam velocity which is twice the
A stage can also include one, two, or, in certain older de-
100
signs, three lacing or tie wires. These wires provide a continu-
Vector
Diagrams for the Turbine Stage
C
Stationary and rotating blade vanes are
designed to produce
C a
ro a
profile which will expand the steam by providing a passage
### through
which it can flow, change direction, and exhaust at
the
correct pressure and velocity to enter the following blade
L row.
The design of profiles is one of the more complex aspects
no
L bo of
turbine engineering. This is because it requires that pro-
D D
ot et files be
defined which can be produced at realistic costs with
D it
existing machining capabilities and which will induce a mini-
mum of
energy losses into the stage. These blades must also
L have
sufficient mechanical strength so they will last for the
###
bi
L ni
predicted life of the turbine unit.
As
a preliminary step to defining the requirements of a
blade
vane, it is necessary to establish the required profile
inlet
and discharge angles (8). These angles are normally de-
angles
which are required of the profile to accept the in-
flowing
steam with a minimum of incidence loss, and they
also
provide a controllable exhaust which will ensure that the
steam
discharge area and direction from the row are correct.
Cir
Shown in Fig. 7 are velocity diagrams for both a pure im-
pulse
and a 50% reaction stage. In these diagrams the blade
##
tangential velocity U is the same on both designs as are the
D
or
stationary and rotating blade discharge angles and # . Fig-
1 2
(a) ure
7(b) shows the vane profiles required by these two velocity
Form of
the Blade Vanes
C The
two blade rows comprising each stage are designed to
ro
accept
the inflowing steam, deflect it through the desired
## angle
sufficient to allow the required degree of energy conver-
sion,
and then deflect it into the following blade row, either
elements within their own row and rows both up- and down-
stream
is correct.
## TURBINE
CONTROL
Cr L b D nr Der
D ir
The
turbine control system is designed to control both unit
Figure 6. (a) The basic impulse stage showing component arrange- the
required quantity of steam.
ment and major dimensions. (b) The basic reaction stage showing In
non-reheat units, there are valves located at the inlet to
component arrangement and major dimensions. the
steam turbine unit; these valves open and close to control
the
admission of steam to the high-pressure section. In reheat
units,
there are large quantities of steam contained between
the
cold outlet and the hot reheat return to the unit. There-
fore,
it becomes necessary to incorporate valves between the
α
1 β β 1
δ
W2 C1
W1
C2
α 1 β2 β 1
W1
C2
W2 C1
δ
U is the tangential
velocity of the rotating blade.
C1 is the absolute
velocity of the steam discharging from the stationary blade row.
α 1 is the discharge
angle from the stationary blade row,
W1 is the relative
velocity of the steam entering the rotating blade row.
β 1 is the rotating
blade vane inlet angle.
β 2 is the rotating
blade vane discharge angle.
W2 is the relative
velocity of the steam discharging from the rotating blade row.
C2 is the absolute
velocity of the steam discharging from the rotating blade row.
δ is the steam
absolute discharge angle.
(a)
This EHC system also allows remote control and remote diag-
nostic monitoring and is capable of being computer-controlled Sliding
Pressure Control
from the control room.
An
alternate method of steam admission into the control
The high-pressure oil system used on the EHC design has
stage
is ‘‘sliding pressure operation’’ in which the initial pres-
strict requirements for cleanliness and fluid stability. Because
sure
of the steam produced in the boiler is reduced with re-
of the possible leakage problems associated with a high-pres-
ducing
load to a condition that the control valves can be left
sure oil, it is necessary to use a fire-resistant fluid in this
wide
open at partial loads. This method of steam admission
system.
has
the effect of requiring less energy to be expended in the
A well-designed EHC system is capable of taking the main
boiler,
to raise the conditions of the steam to one suitable for
and intermediate stop and control valves from a fully open to
admission to the turbine unit, and does not suffer from the
a fully closed position in about 0.2 s.
The
turbine-generator unit is a high-speed machine, op-
Speed and Load Transients erating
at high direct stress levels and subject to a number of
9.
W. J. Kearton, Steam Turbine Operation, 7th ed., London: Pit-
STARTING AND STOPPING STEAM TURBINES
man, 1964.
The starting and stopping of steam turbines can be a complex
WILLIAM P. SANDERS
operation. The complexity is introduced by the effect of tem-
Turbo-Technic Services, Inc.
BIBLIOGRAPHY
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3006.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Synchronous Generators
● Email this page Enter words or
Standard Article
Abstract
The sections in this article are
Steady-State Operation
Dynamic Performance
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3006.htm17.06.2008 16:46:57
SYNCHRONOUS GENERATORS
Synchronous generators have long been used for converting mechanical to electrical
energy in ac power systems.
The great majority of these generators are three-phase machines operating
at 50 Hz or 60 Hz. In some
other applications, other frequencies are also used, such as 400 Hz in
aerospace and marine electric power
applications. For other special applications, single-phase generators and
generators with more than three
phases are built. In this article, three-phase synchronous generators operating at
power frequency, (i.e. 60 Hz
or 50 Hz) will be mainly considered.
A three-phase synchronous generator has essentially two elements: a three-
phase armature winding and
a magnetic field excitation winding or structure. For constructional
convenience and economic reasons, the
armature winding is commonly located in the stator (nonrotating part of the
machine), and the rotor (rotating
part of the machine) contains the magnetic field excitation winding or structure. In
nearly all synchronous gen-
erators, a field winding carrying a dc current is used to establish the magnetic
field. In some small synchronous
generators, permanent magnets are utilized to establish this magnetic field.
A three-phase armature winding consists essentially of sets of coils
equal to the number of pairs of
poles symmetrically distributed around the stator, and each set consists
of three groups of coils, which are
symmetrically distributed under every pair of poles. Each one of these groups
belongs to one of the phases,
and all the groups belonging to one phase are connected together, usually in series
but occasionally in parallel,
to form the phase winding. For the availability of a neutral and because of
insulation requirements, the three
phase windings are connected in wye (star) for most synchronous generators.
When the rotor of a three-phase synchronous generator is driven by a prime
mover, the magnetic flux
linking the armature winding coils alternates and an alternating voltage is
generated in them. The generated
voltages in the three phase windings are similar, but are always 120◦
apart in phase. The frequency of these
voltages will depend on the speed of the prime mover and the number of poles of the
synchronous generator
and is given by the following relationship (1,2):
2 SYNCHRONOUS GENERATORS
As mentioned before, the voltages are generated in the armature phase windings of a
synchronous generator
as a result of the relative motion between the magnetic field and the armature
winding. The waveshape of
these generated voltages will only be sinusoidal if the magnetic-flux-
density waveshape is also sinusoidal.
SYNCHRONOUS GENERATORS 3
STEADY-STATE OPERATION
When the active power demand from a synchronous generator changes, small changes of
speed cause a governor
to adjust the input power of the prime mover accordingly. In this case, the prime-
mover speed and the system
frequency will be maintained as nearly as possible at the specified
values. On the other hand, the effect of
changing the field current of the synchronous generator will depend on the nature of
the system connected at
its terminals.
4 SYNCHRONOUS GENERATORS
Fig. 3. The relationships between the air-gap flux and the magnetizing ampere-turns
(saturation curves) along the direct
and quadrature axes of a salient-pole synchronous generator.
SYNCHRONOUS GENERATORS 5
Figure 6 shows the phasor diagram for this case when the generator is
overexcited and supplying power
at a lagging power factor. In this phasor diagram, I adX d and I aqX q
are the voltage drops in the direct- and
quadrature-axis synchronous reactances calculated by using the direct- and
quadrature-axis components of
the armature current, respectively. To obtain the values of these two components, a
phasor equal to I aX q is
added to the terminal voltage to obtain the phasor Eq . The direction of Eq gives
the direction of the quadrature
axis, and the direction of the direct axis will, as a result, be known. Knowing the
directions of these two axes,
the components of the armature current along the direct and quadrature axes will be
known, and the voltage
drops I adX d and I aqX q can be calculated. It is clear from the phasor diagram
that the phasor of the generated
voltage EG will be in the same direction as Eq , that is, it will be in the
direction of the quadrature axis.
Effect of saturation on the steady-state performance. In
the analysis of the steady-state perfor-
mance of synchronous generators, the accurate calculation of their field excitation
and load angle depends to
a large extent on their saturation condition. In general, the effect of saturation
will depend not only on the
saturation curve in the axis (direction) of the resultant machine ampere-turns, but
also on the phase angle
between the resultant ampere-turns and the resultant magnetic flux. This phase angle
depends on both the
machine saliency and the different saturation levels along the different
axes of the synchronous generator
(8,9).
To take account of saturation in the synchronous-generator phasor diagrams,
empirical methods, which
in most of the cases seem to give fairly closer agreement to the measured values,
are used. In these methods,
the custom in the past was to use only one saturation factor or
increment corresponding to the saturation
along the direct axis. This has been replaced in present practice by the use of two
different saturation factors
corresponding to the saturation conditions along the direct and quadrature axes,
since the saturation levels
are very different along the two axes even in the case of the cylindrical-rotor
synchronous generators. These
saturation factors (or increments) are defined as the ratio of (or the difference
between) the actual excitation
and the excitation on the air-gap line of the saturation curves at the
operating point. The internal voltage
behind the leakage (or Potier) reactance is usually used to locate the operating
point on the saturation curves.
In using these two saturation factors, the magnetic coupling between the direct and
the quadrature axes (the
6 SYNCHRONOUS GENERATORS
SYNCHRONOUS GENERATORS 7
Use of two saturation factors. In this approach, both the direct- and
quadrature-axis saturation effects
are represented by specifically adjusting the unsaturated direct- and quadrature-
axis synchronous reactances
by corresponding saturation factors K d and K q to obtain the saturated direct-
and quadrature-axis synchronous
reactances as discussed in the previous sub-subsection. In this case, K d is
calculated from the direct-axis sat-
uration curve to give X dsat , while K q is calculated from the quadrature-
axis saturation curve to give X qsat .
The internal voltage behind the leakage or Potier reactance is used to determine
the operating point on both
saturation curves. Using these saturated reactances X dsat and X qsat in
Fig. 6, an internal voltage (≡EG in
Fig. 6) is obtained. The total excitation is then obtained by multiplying this
internal voltage by the saturation
factor K d . Empirically, the use of the two saturation factors seems to give, in
some cases, an accurate represen-
tation of the effect of saturation along the intermediate axis of the total
resultant ampere-turns, that is, the
effect of both the saturation factor due to the total resultant ampere-turns using
the saturation curve along
the intermediate axis of the total resultant ampere-turns, and the phase angle
between the resultant flux and
the resultant ampere-turns.
Cross-magnetizing phenomenon. It has been recognized recently
that the magnetic coupling between
the direct and quadrature axes of saturated synchronous machines (cross-magnetizing
phenomenon) plays an
important role in their analysis (8,9). When the flux linkages along the direct and
quadrature axes due to the
components of the ampere-turns along these axes are handled separately, this
magnetic coupling causes changes
in these flux linkages. This can be represented in the phasor diagram of saturated
synchronous generators
8 SYNCHRONOUS GENERATORS
Fig. 10. A power capability diagram that shows the loading limitations imposed by
heating, prime mover, excitation, and
stability on a nonsalient-pole synchronous generator.
If the cross-magnetizing effect [the last two terms of Eq. (4)] and saliency
[the second term of Eq. (4)] are
neglected as represented in the phasor diagram of Fig. 8, the power–load-angle
relationship becomes
a QD sat
equals 90◦ . This is the stability limit under steady-state conditions. However, if
saliency is not neglected, the
angle at which a maximum power can be delivered by the synchronous generator is
less than 90◦ , and its value
is dependent on the operating condition. In this case, the maximum power will be
larger than EaEQD/X sat .
Loading limitations. The loading limitations imposed by heating, prime
mover, excitation, and stability
on a synchronous generator connected to an infinite bus system can be evaluated by
using power capability
diagrams (11121314–15). Neglecting losses and saturation, Fig. 10 shows
such a power capability diagram
for the case of a nonsalient-pole synchronous generator connected to an infinite bus
system. In this diagram,
each point represents an operating condition. For example, point e represents the
rated loading condition, the
line bo represents the constant-phase voltage of the infinite bus system Ea , the
line oe represents the rated
stator current to a scale or the rated MVA to another scale at power-factor angle
φ, the lines og and oh are the
corresponding active and reactive powers, and the line be represents the phase
generated voltage EG .
The operation of this machine must be so controlled that the operating power
remains within the bound-
aries set by:
(1) An arc ef of center b and radius be representing the rated field current, that
is, the rotor heating limit
(2) An arc of center o and radius oe representing the rated stator current, that
is, the stator heating limit
(3) The line dge representing the rated power output of the prime mover
(4) An arc ac that represents the minimum acceptable field current
(5) The curve ci, which represents the practical stability boundary beyond which
instability at all loads occurs
when the synchronous generator absorbs excessive reactive power, (i.e., is
underexcited)
10 SYNCHRONOUS GENERATORS
Fig. 11. A power capability diagram that shows the loading limitations imposed by
heating, prime mover, excitation, and
stability on a salient-pole synchronous generator.
diameter pn . This semicircle represents the locus of the operating point e with
zero excitation. For any other
constant excitation, the locus of point e is not quite a circular arc; however,
this becomes evident only with low
excitation. In this figure, the active power represented by the line og is divided
into two components: ok, which
is the power contributed by the excitation, and kg (= bl), which is the reluctance
power due to the saliency.
The theoretical steady-state stability boundary nqr can be obtained by applying the
expression dP/dδ = 0. It
should be noted that salient-pole synchronous generators can be operated with
negative excitation (12,16).
SYNCHRONOUS GENERATORS 11
Fig. 12. Typical oscillograms of the stator and field currents of a synchronous
generator on a sudden three-phase short
circuit at its terminal.
axis short-circuit transient time constant. The magnitude of the second component,
namely the offset direct
12 SYNCHRONOUS GENERATORS
current, depends on how near the instantaneous value of the phase voltage is to
zero at the instant of the short
circuit. For the instant corresponding to zero instantaneous phase voltage, the
offset direct current will be at
its maximum magnitude, and it will be zero for the instant corresponding to the
peak of the phase voltage. The
magnitude of the offset direct current in each phase will be equal and opposite to
the instantaneous value of
the balanced alternating current component at the instant of the short circuit, and
thus it can be equal to zero
in only one of the phases. In comparison with the alternating component, the offset
direct current components
decay at a rapid rate, with the armature short-circuit time constant τa .
The balanced three-phase component of the short-circuit stator
current gives rise to a direct current
component in the rotor circuits, namely the rotor damper circuits and the field
winding. This is because the
balanced three-phase component produces a magnetomotive force rotating
synchronously with the rotor. In the
case of the initially unloaded synchronous generator, the axis of this
magnetomotive force lies along the pole
axis and the magnetomotive force tends to reduce the flux linkages with the rotor
circuits. Such changes of the
flux linkages with the rotor circuits result in induced currents in the rotor damper
circuits and field winding.
The induced currents in the damper circuits decay with the direct-axis short-
circuit subtransient time constant
τd## , while the induced current in the field winding decays after the subtransient
period with the direct-axis
#
short-circuit transient time constant τd .
On the other hand, the stator offset direct current components produce a
stationary magnetomotive force
that will induce a rotational-frequency current in the rotating rotor
damper and field circuits. These rotor
currents, like the stator offset direct current components, will decay
with the armature short-circuit time
constant τa .
Effect of sudden three-phase short-circuit on synchronous generators.
As a result of the large
currents flowing in the stator windings of a synchronous generator on a
sudden three-phase short-circuit,
electromagnetic forces much larger than normal will result on the end
windings of the stator coils (18,19).
These forces can produce severe mechanical strains on the end windings and cause
insulation failure. These
end windings must, therefore, be adequately braced.
Besides the forces on the end windings of the stator coils, a sudden three-
phase short-circuit of a syn-
chronous generator results also in pulsating torques, which can be many
times the magnitude of the rated
torque (17,19). The casing, the shaft, and the foundation of the synchronous
generator must be designed to
withstand the stress developed as a result of these torques.
DYNAMIC PERFORMANCE
(1) The phase sequence of the incoming synchronous generator should be the same as
that of the bus bars.
(2) The frequency of the incoming synchronous generator should be close to that of
the bus bars.
SYNCHRONOUS GENERATORS 13
Fig. 13. The application of the equal-area criterion to illustrate the dynamic
performance of a synchronous generator.
1 3
However, due to the existence of damping, these oscillations will damp and the
operating point will settle at
the point C at which the electric output power is again equal to the mechanical
input power. In the case that
14 SYNCHRONOUS GENERATORS
the mechanical input power is too close to the peak of the power–load-angle curve
Ea2 , the area CDF will be
less than the area ABC and the rotor can never be slowed to the synchronous speed
before reaching point F . In
this case, the generator rotor starts to accelerate beyond point F and the
generator falls out of synchronism.
If the change in the load angle (#δ) around an operating point is small, the
accelerating or decelerating
power (synchronizing power) can be expressed as follows:
SYNCHRONOUS GENERATORS 15
where EG is the internal electromotive force that represents the air-gap flux along
the quadrature axis. Mea-
suring the terminal voltage Ea and the armature current I a at the stability
limit and subtracting the armature
leakage voltage drop X lI a from Ea , the relationship between the
electromotive force EG and the armature
current I a can be obtained and can be used to represent the relationship
between the air-gap flux and the
excitation current along the quadrature axis, that is, the quadrature-axis open-
circuit characteristic curve.
Leakage reactance. For determining the value of the leakage reactance,
the Potier reactance X p mea-
sured at rated terminal voltage is usually used as an approximation. This reactance
can be determined with
the help of the open-circuit and zero-power-factor characteristic curves (7).
However, it has been found that
the value of the Potier reactance measured at rated terminal voltage may
not be a good approximation of
the value of the armature leakage reactance. To obtain a more accurate value, the
Potier reactance has to be
determined in the highly saturated region of the open-circuit characteristic curve
(24). This may however be
difficult, since the synchronous generator may not be able to stand the values of
the field current needed to
obtain the open-circuit and zero-power-factor characteristics in the highly
saturated region. Reference 24 has
proposed an alternative testing method to obtain accurate values of the armature
leakage reactance without
any risk to the generator under test.
Subtransient and transient reactances and time constants. The
direct-axis subtransient and transient
reactances and their respective time constants can be obtained from the three-phase
sudden short-circuit test.
16 SYNCHRONOUS GENERATORS
SYNCHRONOUS GENERATORS 17
AHMED M. EL-SERAFI
University of Saskatchewan
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3007.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Synchronous Motor Drives
● Email this page Enter words or
Standard Article
Abstract
The sections in this article are
Machine Models
Power Converters
Sensor Requirements
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3007.htm (1 of 2)17.06.2008 16:47:18
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3007.htm
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3007.htm (2 of 2)17.06.2008 16:47:18
ans
apart are associated with each of these three phases. The
Field main
rotor winding is the field winding (fd-fd #). In addition,
exciter
the
rotor also includes damper ‘windings’ denoted (kd-kd #)
and
(kq-kq #). These windings may not actually be windings at
all;
instead they often represent currents flowing in alumi-
num
bars inthe rotor structure which provide damping
Power Power
source converter SM
torque during electromechanical transients. The q and d axes
of
the machine denote the magnetic axes of the q-axis damper
and
d-axis field and damping windings, respectively. Figure 2
Switching Sensors also
defines the mechanical rotor position #rm of the machine,
logic
which is measured from the as axis to the q axis. The mechan-
ical
rotor speed #rm is measured in the counterclockwise direc-
Command Control
tion. For the purpose of machine analysis, it is convenient to
rotor a North magnetic pole and the rotor iron on the opposite
side
a South magnetic pole. Next, a three-phase ac source ex-
machine. Another important aspect of this system is a sensor
cites the stator windings. The a-, b-, and c-phase applied volt-
array which often includes a rotor positional sensor, current
ages
have the form
sensors, and voltage sensors. Based on the sensor output and
a torque, speed, or other input command, the drive controller
v = √2 V cos(θ ) (1)
determines the on/off status of each of the power converter
as s e
semiconductors so that the desired performance is achieved.
√ # 2π # (2)
v = 2 V cos θ −
The operation of synchronous motor drives is dictated by
bs s e 3
# #
The interior of the machine (the rotating member) is called
2π
the rotor and the exterior portion (the stationary member) is
v = 2 V cos θ + (3)
cs s e
the stator. Three sets of windings (as-as #, bs-bs#, and cs-cs #)
3
located in the interior of the stator are called phases. Physi-
where
cally, these windings are sinusoidally distributed in slots on
the inside of the stator although they are shown in lumped
θ = ω t (4)
positions in Fig. 2. Three magnetic axes, which are 2#/3 radi-
e e
In
Eq. (1–4), Vs is the rms amplitude of the applied stator
√ # 2π # (6)
ibs = 2 Is cos θe + φi −
cs θrm
3
bs
kg'
fd' and
kd kd as axis
fd
√ # 2π # (7)
nously with the stator magnetic field. Thus, the electrical ro-
Figure 2. Cross-sectional view of a wound-rotor synchronous ma-
tor
speed is # , and the mechanical (physical) rotor speed is
chine. The stationary member, labeled ‘‘stator,’’ contains three sets of
e
2# /P . The fact that the rotor speed is directly tied to the ra-
windings which are sinusoidally distributed (although they are shown e
here as lumped coils). The rotating member, labeled ‘‘rotor,’’ is a sa- dian
frequency of the stator excitation is the defining charac-
lient structure containing a field winding to produce the rotor MMF
teristic of synchronous machines.
and damper windings to produce damping torque during electrome-
To make this observation more quantitative, if the resis-
chanical transients.
tance of the stator windings is negligible, the electromagnetic
as'
3 P Lmd i# √2 Vs
fd
T = − sin δ
e 2 2 ω L
e d (8)
# #
cs bs rm
√
3 P 1 1 1
S
− − ( 2 V )2 sin 2δ
N
4 2 ω2 L L s
e q d
as axis
In Eq. (8), L , L , and L are the d-axis magnetizing induc-
md d q
bs' N cs'
inductance, respectively, i# is the referred field current, and
fd
Air
# is the torque angle of the synchronous machine defined as
δ = θr − θe (9)
as
d axis
Rotor iron
Physically, the torque angle represents the difference be-
cs axis Permanent
tween the electrical rotor position and the instantaneous
magnet
angle of the applied voltages. In Eq. (8), the first term repre-
sents the torque produced by the interaction of the field wind-
Figure 4. Cross-sectional view of a surface-mounted, permanent-
ing with the stator magnetic field (field torque), and the sec-
magnet, synchronous machine. This machine differs from the config-
rm q axis
as'
)
m
⋅
N 0
cs θrm
(
bs
e
T
S as axis
Air
as
–50
d axis
–2π – 3π –π – π 0 π π 3π 2π
Rotor iron
2 2 2 2
cs axis
Permanent
δ (rad)
magnet
material
Figure 3. Field, reluctance, and total torque as a function of delta.
The field torque is produced by the interaction of the stator magnetic
Figure 5. Cross-sectional view of a buried, permanent-magnet, syn-
field with that of the rotor field winding. The reluctance torque is
chronous machine. This permanent-magnet machine has magnetic
produced by the interaction of the salient rotor iron and the stator
saliency due to the low relative permeability of the magnetic material
magnetic field.
buried in the rotor iron.
v = r i + ω λ + p λ (13)
at higher speeds than their surface-mounted counterparts be-
qs s qs r ds qs
cause the iron surrounding the mechanically weak permanent
v = r i − ω λ + p λ (14)
ds s ds r qs ds
magnet adds the structural strength required to withstand
high operating speeds.
and
v = r i + p λ (15)
0s s 0s os
MACHINE MODELS
and
To provide a more quantitative look at synchronous-machine
drives, it is first necessary to set forth the mathematical mod-
v# = r# i# + p λ# i ∈ 1 . . . m (16)
v# = r# i# + p λ# (17)
els have three parts: the voltage equations which relate the
fd fd fd fd
applied voltage to the ohmic voltage drop to winding resis-
tance and the time rate of change of flux linking the various and
windings; the flux linkage equations which relate the flux
# # # #
linking the individual windings to the current in the windings
vkdi = rkdi ikdi + p λkdi i ∈ 1 . . . n (18)
# # # #
2 2 2
(10)
and
In Eq. (10), f is a stator voltage v, stator current i, or stator
# # #
flux linkage #. Expressing the machine model in terms of
λkqi = Llkqi ikqi + λmq i ∈ 1 . . . m (22)
dent inductances from the machine description and also re- and
sults in a model wherein the various voltages, currents, and
flux linkages are constant in the steady state. In addition to
λ# = L# i# + λ i ∈ 1 . . . n (24)
kd lkd kd md
expressing the synchronous machine model in terms of trans-
i i i
i= 1
# Ns
vfd = N vfd (11) and
fd
# n #
and
λ = L i + # i + i# (26)
md md ds kdi fd
i= 1
# 2 Nfd
ifd = 3 Ns ifd (12)
Finally, electromagnetic torque, in terms of the transformed
variables, is expressed as
where Ns and Nfd are the effective stator and field turns, re-
3 P
T = (λ i − λ i ) (27)
spectively. The damper circuits are similarly referred. Be-
e 2 2 ds qs qs ds
cause the turns ratio is not needed to use the model and can-
not be measured in the laboratory, however, these referred
Figure 6 illustrates an equivalent circuit consistent with
variables are not defined here. Eqs.
(13–26). It is worth noting that the model is valid for
iqs As
in the case of the wound-rotor machine, this model is valid
ω λ
r ds
for both steady-state and transient conditions. The model is
+ –
also valid for both surface-mounted and interior-mounted,
+ r i
s qs
permanent-magnet machines, although, in the case of sur-
L'lkq 1 L'lkq
face-mounted machines, Lq and Ld equal. Therefore, both of
v L
these symbols are normally replaced by the symbol Lss when
qs mq
d q
i'
L' L' fd
lkd1 lkqn
POWER CONVERTERS
v L v'
ds md fd
T1 T2 T3
mally negligible) and (2) treating the field current as a con-
stant. After making these changes and combining the flux
linkage equations with the voltage equations,
vdc
+ + +
v = r i + ω L i + L pi + ω λ (28)
qs s qs r d ds q qs r m
v = r i − ω L i + L pi (29)
T4 vag T5 vbg T6 vcg
ds s ds r q qs d ds
– – –
and –
v = r i + L pi (30)
0s s 0s ls 0s
ias ibs ics
vbs
inductances and #m is a constant related to the strength of the
vas vcs
permanent magnet (the fictitious, constant, field current), the
– –
number of stator winding turns per phase, and the machine
–
geometry. The torque equation becomes
3 P
T = [λ i + (L − L )i i ] (31)
Figure 7. Fully controlled, three-phase bridge inverter and wye-
e 2 2 m qs d q qs ds
connected load.
idc
voltage drops, these voltages are expressed by
+
vag = #vdc T1 on, T4 off (38)
T1 T2 T3
0 T1 off, T4 on
+ + +
v v v and
T4 ag T5 bg T6 cg
– – – –
vag = #vdc T3 on, T6 off (40)
0 T3 off, T6 on
– +
ibs
vas vbs In
Eqs. (38–40), it is assumed that one, and only one, active
i
semiconductor of each phase leg is conducting. Operation with
as
two
devices simultaneously conducting is normally not al-
+ –
ics lowed
because this would short circuit the dc source. Opera-
tion
with neither device on is occasional. However, analyzing
– v +
cs this
condition is quite involved and so the reader is referred
possess considerable
harmonic content which results in in-
1 2 1
creased machine losses. Nevertheless, this strategy is often
vbg = − 3 vag + 3 vbg − 3 vcg (33) used
for permanent-magnet synchronous motor drives be-
cause
it is one of the least expensive strategies. For the pur-
and poses
of analysis, machines operated from the three-phase
2
8,
v = v cos(θ ) (41)
as ,f und dc c
π
vas = vag − vbg (35)
2 v # 2π # (42)
and
and
2 v # 2π # (43)
vcs = vcg − vag (37)
vcs ,f und = π dc cos θc + 3
S1
T1
T1,T4
T4
S2
T2,T5
T2
T5
T3,T6
S3
T3
T6
d +
c
vag vdc
t –
π 3
v
1/3 dc
The
advantage of this control is that the amplitudes of the
v
vng 2/3 dc
applied voltages are readily controlled. The low-frequency
da +
nected Eqs. (41–43) are adjusted by multiplying the ampli-
– T1
tude by #3 and adding #/6 to each of the cosine terms.
The next level of sophistication in the control of the fully
T4
controlled bridge converter is applying duty-cycle modulation
to the switching signals, as illustrated in Figure 10. There,
db +
T2
the signals S1, S2, and S3 are identical to T1, T2, and T3 in
–
the case of a six-step operation. The commanded duty cycle d
is compared to a high-frequency triangular wave tr, which
T5
varies between zero and one. If the duty cycle is greater than
dc +
the triangular wave, the resulting line-to-neutral voltages are
T3
–
the same as in a six-step operation. Otherwise, they are zero.
The net effect is that the fundamental components of the ap-
T6
plied voltages become
t
2
Figure 11.
Sine-triangle modulating switch-control algorithm. The
v = v d cos(θ ) (44)
as ,f und π dc c
fundamental component of the resulting a- b- and c-phase line-to-
# #
2 2π
neutral voltages using this control strategy are in phase with and
vbs,f und = vdcd cos θc − (45)
proportional to the magnitudes of the instantaneous phase duty cy-
π 3 cles
d , d , and d provided that the signals constitute a balanced set.
a b c
i <i * – h
as as
v = 1d v (47)
Figure 12. Hysteresis
current-regulated, switch-control algorithm.
as ,f und a dc
2
Using this algorithm, the converter switching is based on comparing
the
error between the commanded currents and the actual currents
1
vbs,f und = db vdc (48) to
a threshold level h.
2
and
by
the hysteresis level h, the lower transistor of the leg (T4)
1 is
turned on, and the upper transistor (T1) is turned off, so
v = d v (49)
cs ,f und 2 c dc
that the current decreases. Conversely, if the actual current
as as as cs as cs
d = d cos #θc − 2π # (51)
respectively, in the hysteresis control. The b- and c-phases are
b 3
controlled similarly. The net result of this switching is that,
3
First, a step change in current command necessarily causes
the
actual current to deviate from the commanded current by
where d is the duty cycle (which is a constant as long as the an
amount exceeding the hysteresis level because the actual
amplitudes of the applied voltages are constant) the line-to-
current does not change instantaneously under an inductive
neutral voltages become or
motor load. Furthermore, even for steady-state conditions,
if
the actual currents are within h of the commanded cur-
1
v = dv cos(θ ) (53)
rents, the rms value of the fundamental component of the mo-
as ,f und 2 dc c tor
phase voltages must satisfy
1dv # 2π # (54)
vbs,f und = dc cos θc −
1
2 3
v < v (56)
s √6 dc
and
if
the machine is wye-connected or
1dv # 2π # (55)
vcs ,f und = 2 dc cos θc + 3
1
v < v (57)
s √2 dc
Ldc
eas ebs ecs
– v + id
as
e
as Lac i + +
as
– + T1 T3 T5
– v +
as
ebs L
ac ibs
– +
vr vdc
T2 T3 T4 T5 T6 T1
– v +
as
e
cs L
ac i
cs
– + T4 T6 T2
– –
i
as
cs
other to the point at which the individual thyristors are fired.
For example, valve three is fired at a firing angle # after ebs
becomes greater than eas , or, more formally, when
π
θc = α + (58)
3
r
thyristor, which was turned on 2#/3 radians previously, to
pass through zero whereupon it turns off. This process is
called commutation, and the angle between the point where
one device begins to conduct and the device gated on 2#/3
π π π 2π 5π π 7π 4π 3π 5π 11π 2π
radians previously ceases to conduct is called the commuta-
6 3 2 3 6 6 3 2 3 6
tion angle u. For overly large dc currents or firing delays, it
θc (rad)
is possible that commutation does not occur (a commutation
failure) which prevents the converter from operating satisfac-
Figure 14. Operation of a three-phase semicontrolled converter in
torily. The final trace illustrates the rectified output voltage.
rectifier mode. The three-phase voltages represent the machine’s
back-emf. The arrows beneath these voltages indicate the rotor posi-
The most pronounced feature in the figure is the notch which
tion at which the various thyristors are turned on. The phase cur-
occurs during the commutation process.
rents are the result of the firing sequence assuming that the dc cur-
Analyzing the semicontrolled, three-phase, bridge con-
rent is constant. The notch in the rectifier voltage results from the
verter is normally quite involved, because a synchronous ma-
phase current commutation.
chine cannot be modeled accurately in a voltage-behind-in-
ductance form (unless the subtransient inductances are
equal, but this is actually never the case). However, neglect-
ing commutation, the average rectifier voltage is given by
√
√
3 3
vd = 2 E cos α (59)
π
2√3
bs axis
ias = id cos(θc − α + π ) (60)
ha
π
√
as'
ibs = 2 3 id cos #θc − α + π − 2π # (61)
ωrm
π 3
S bs qs axis
cs
and
θ
rm
bs' N cs'
From Eq. (59), for firing angles between 0 and # radians, the
average rectifier voltage is positive. Because the dc current
must be positive, power must be flowing out of the machine,
as hc
and so the converter acts as a rectifier. However, as the firing
cs axis
delay is increased past # radians, the average rectifier voltage
ds axis
is negative, and so power flows from the dc side of the con-
close to the permanent magnet on the end cap or buried in the stator
pear that the optimal phase delay is 2# radians, which maxi-
iron. The main purpose of this figure is to show the position of the
mizes power transfer as an inverter, and in practice the firing
sensors and the phase shift #hm .
delay angle is maximized for inverter operation. The possibil-
ity of commutation failure, however, limits the maximum
phase delay achievable in this type of converter (12).
control loop. This latter approach typically mitigates stability
SENSOR REQUIREMENTS
Regardless of whether the machine is excited from a fully con-
trolled or semicontrolled inverter, the synchronous machine The
closed-loop control of synchronous motor drives requires
always operates at a speed corresponding to the frequency of a
rotor position sensor to # . For the 180# VSI switching strat-
r
the applied voltage and current providing a very easy means egy,
rotor position is determined by inexpensive Hall-effect
of speed control. In particular, by simply operating the in-
sensors mounted on the stator as shown in Fig. 15 where
verter at a frequency corresponding to the desired speed, the #hm
denotes the angle of mechanical shift of the sensors from
torque angle automatically adjusts itself so that the electro- the
reference position shown. These sensors put out a logic
magnetic torque is equal to the load torque, subject to the high
when under a South magnetic pole and a logic how when
conditions that (1) the load torque is greater than the mini-
under a North magnetic pole to produce the logic signals as a
mum and less than the maximum electromagnetic torque pro-
function of rotor position as shown in Fig. 16 where the elec-
duced in accordance with the torque versus torque angle char-
trical phase shift angle #h is related to the mechanical phase
acteristic, (2) the system is dynamically stable for the given
shift by the number of pole pairs or #h # (P/2)#hm . Note that
load torque, and (3) the rate at which the frequency is varied
and the way the load torque varies with speed are such that
transient stability is maintained during startup. Because the
torque angle automatically adjusts itself to the correct value
ha
to satisfy the load torque and because the speed must match
the applied frequency, this open-loop type of speed control is
attractive in that no rotor position or speed sensors are re-
quired. In practice, however, conditions (2) and (3) limit the
hb
use of this type of control. The alternative is closed-loop con-
trol in which rotor position #r is measured and #c is calculated
by
θc = θr + φv (63)
6 2 6 6 2 6
sired torque angle (in fact, #v #
#; the difference in nomen-
θ +φ (rad)
clature is that #v is traditionally used by drive engineers,
r h
eff = (68)
the drive. Although the encoder or resolver can be eliminated
P
in
by using a full observer to determine rotor position, these
methods usually require an elaborate scheme for start-up If
the power electronic converter losses are neglected, another
(15,16), and cannot guarantee a bound on the initial error.
useful relationship applicable to any drive system is that the
Furthermore, these methods generally require knowledge of
average dc current is found from the input power by
the motor parameters. Recently a hybrid observer was devel-
oped which senses rotor position continuously with inexpen-
P
sive Hall-effect sensors (17). This observer does not require
idc = in (69)
v
dc
knowledge of motor parameters or an elaborate start-up
scheme. Furthermore, the error in estimating the rotor posi-
For
this discussion, it is assumed that the dc voltage is con-
tion is bounded by a limit which is independent of operating
stant.
conditions. Besides rotor position, some synchronous-machine
dressed. These drives are used widely in industry and are re-
ment, are often used for this purpose.
as e
φs = angle(f qs −j f ds ) (65)
lation. As can be seen, the a-phase voltage waveform is a
300
300
4 ms
4 ms
)
V
) (
0
V s
a
( 0 v
s
a
v
–
300
–300
15
15
)
) A
A 0 (
0
( s
s a
a i
i
–15 –
15
3
3
)
) m
m ⋅
N⋅ 1.5 N
1.5
(
( e
Te T
0
0
the same as in Fig. 17. In this case, the voltage, current, and The
phase advance #v is physically determined in software
torque waveforms contain high-frequency instead of low-fre-
(and
considered an input) for continuous type of rotor position
quency harmonics. This mode of operation is advantageous in
sensing.
If Hall-effect devices are utilized, the phase advance
terms of both current and torque ripple. Furthermore, the
is
dictated by the physical position of the sensors. If the ma-
duty cycle is readily controlled to control the speed. Disadvan-
chine
windings are connected in a delta configuration, Eqs.
tages of this control are the expense of the rotor position sen- (70–71)
must be modified by multiplying the voltages by #3
sor and increased semiconductor switching losses.
and
adding #/6 to the cosine and sine terms. The fundamental
Analyzing this type of drive system begins with determin-
v = mv (73)
v − mv cos(φ ) (70)
s √2 dc
qs dc v
and The
machine q- and d-axis currents are determined by aver-
aging
the machine voltage Eqs. (28–29) and solving for cur-
v − mv sin(φ ) (71)
ds dc v rents.
This yields
qs 2 2 s qs r d ds s r m
r + ω L L
s r q d
2 ◦
for 180 operation
π and
m = 2 d for duty-cycle modulation (72)
1
π
2
i = [ω L v + r v − ω L λ ] (75)
r q qs s q m
ds 2 2 ds r
r + ω L L
1
s r q d
d for sine-triangular modulation
2
The
average torque is calculated by averaging Eq. (31) which
yields
Table 1. Permanent-Magnet Synchronous Machine
Parameters
3 P
T = [λ i + (L − L )i i ] (76)
e 2 2 m qs d q qs ds
rs # 2.985 # Lls # 1.84 mH
Lmq # 9.51 mH Lmd # 9.51 mH
In
obtaining Eq. (76), the average of the product of the q- and
P # 4 # # 0.156 V · s
m d-axis
currents is approximated. Given the average machine
25
25
φv = 0
20
20
m = 0.47
φ = π
m = 0.35
v
) 15
) 15 6
m
m
⋅
⋅
N
N
(
(
e m = 0.24
e
T 10
T 10 π
φv =
3
m = 0.12
5
5
φv = π
0
0
0 100 200 300 400 500 600
0 100 200200 300 400 500 600
ω (rad/s)
ω (rad/s)
rm
rm
mum stall torque whereas a phase advance of #/2 provides the maxi-
dance with Fig. 22. (Note that current control is also possible
)
A
#r , the current command synthesizer determines the q- and
(
s
i m = 0.24
d-axis current command (i* and i*) so that the desired torque
qs ds
10
is obtained. In the case of a nonsalient machine, a simple cur-
i∗ = e (77)
qs 3 P
λm
2 2
0
0 100 200 300 400 500 600
∗
ids = 0 (78)
ω (rad/s)
rm
⋅
set forth in (19,20) for salient and nonsalient machines, re-
N 1.5
* e 8 ms
spectively. Once the q- and d-axis current command is estab-
T
lished, the abc variable current command i* is established
abcs
using the inverse of Park’s transformation given by Eq. (10).
0
A
magnet, synchronous motor drive during a step change in
( 3
* q
torque command. The machine parameters are identical to
v
those used to produce Figs. 17 and 18, the dc bus voltage is
225 V, the hysteresis level is 0.6 A, and the current command
0
A 0
ables depicted include the torque command T*, the q-axis cur-
(
e
s
a
rent command i*, the a-phase current i , and the electromag-
i
qs as
⋅
To analyze current-regulated drives, for normal operating
N 1.5
e
conditions, it is assumed that the actual q- and d-axis cur-
T
rents are equal to the commanded currents:
0
iqs = i∗ (79)
qs
Figure 23. Operation of a hysteretic, current-regulated, inverter-fed,
command. When the current command is stepped up, the phase cur-
i = i∗ (80)
rent increases. This increases the machine torque which closely
ds ds
v = r i + ω L i + ω λ (81)
The machine torque is calculated from Eq. (76). Other quanti-
qs s qs r d ds r m
ds s ds r q qs
for this type of drive system, torque is independent of speed
i*
VOLTAGE-SOURCE INVERTER-FED, SYNCHRONOUS-
abcs
i*
T* qs
RELUCTANCE, MACHINE DRIVES
e Current
— 1
command Kr
i* s
ds
synthesizer
Although not as common as permanent-magnet, synchronous
ω
motor drives, synchronous-reluctance motor drives have ad-
r
kq 1 lkq1
r# # 0.923 mH L # # 3.4 mH
kq2 lkq2
r# # 40.47 mH r# # 4.73 mH
kd 1 lkd1
)
T* = 2 N⋅m
r# # 1.31 mH L # # 3.68 mH
m e
kd2 lkd2
⋅
N 2
Ns/Nfd # 0.02711 P # 4
(
e
T
T* = 1 N⋅m
e
1
alent to a delta angle of
#/4, produces maximum torque as
)
m
Voltage
⋅
N
regulator/
(
Te φv = – π
exciter
6
4000
vfd
φv = 0
π
Semi
2000 φv =
Current
4
controlled SM
source
π
bridge
φv =
6
T1-T6
0
0 10 20 30 40 50
θ
Gate r
ω (rad/s)
(Firing delay) β
rm
sequencer
150
levels. One possible configuration is illustrated in Figure 26
) This
current source is synthesized by a voltage source coupled
V
( 75
d with
an inductor and a closed-loop, current-control system. A
f
v
6.6 ms
voltage regulator/exciter monitors the terminal voltages of
the
machine and adjusts the field voltage of the synchronous
0
machine so that the flux is at an appropriate level. To this
2 end,
a constant-volts-per-hertz scheme is employed. The firing
of
the inverter thyristors is based on rotor position. In this
π
0
θr = + β (83)
3
0
In
Eq. (83), # denotes the firing delay instead of # as in the
)
defined relative to the back-emf of the machine, not the rotor
V –200
(
r
position. Firing relative to the rotor position is robust in that
V
the
firing sequence is unaffected by the harmonics in the
s
a
v tor
speed is 377 rad/s, and the firing delay relative to the
as e
( puter
simulation of a variety of operating points (the analyti-
e
T
cal
solution is quite involved; the reader is referred to
r# # 31.8 mH L # # 6.13 mH
SEMICONTROLLED, INVERTER-FED, WOUND-ROTOR,
kq 1 lkq1
r# # 0.923 mH L # # 3.4 mH
SYNCHRONOUS MACHINE DRIVES
kq2 lkq2
r# # 40.47 mH L # # 4.73 mH
kd 1 lkd1
r# # 1.31 mH L # # 3.68 mH
Because of the high voltage and current capabilities of thyris-
kd2 lkd2
r# # 0.122 mH L # # 2.54 mH
tors, synchronous machine drives are often based on the semi-
fd lkd2
Ns/Nfd # 0.02711 P # 4
controlled rather than fully controlled bridge at high power
30
8. R. R. Nucera, S. D. Sudhoff, and P. C. Krause, Computation of
modeling of
line-commutated converter-synchronous machine
0
systems, IEEE Trans. Energy Conversion, 8: 92–98, 1993.
1.48 1.66 1.83 2.01 2.18
13. DC Motors, Speed Controls, Servo Systems: An Engineering
β (rad)
Handbook, Eden Prairie, MN, Robbins & Myers/Electro-Craft,
1989.
Figure 28. Torque vs # of a semicontrolled, converter-fed, wound-
454, 1995.
BIBLIOGRAPHY
21. S. D. Sudhoff et al., Transient and dynamic average-value model-
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3042.htm
} { { } }
HOME
●
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Thermal Energy Storage
● Email this page Enter words or
Standard Article
Abstract
Fundamentals
Applications
Design
Operation
Conclusion
About Wiley InterScience | About Wiley | Privacy | Terms & Conditions
Copyright © 1999-2008John Wiley & Sons, Inc. All Rights Reserved.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3042.htm17.06.2008 16:47:40
THERMAL
ENERGY STORAGE
Thermal
storage is a method of storing heat for future use.
Most often,
thermal storage is used when future heat sources
or sinks
are either not available at a sufficient rate or will be
more
expensive. A common application of thermal storage is
the
domestic hot water heater. Most household water heaters
use an
electric resistance element or a natural gas burner
as a heat
source. A resistance element may supply heat at
a rate of 6
kW, whereas a hot shower requires 20 kW to heat
6 L/min
of water from 16#C to 40#C. Some amount of hot
water must
be stored in the tank because the element cannot
heat the
water at a rate sufficient for a shower or most other
uses.
Heat is
stored in the hot water tank to meet the demand
for future
use. The number or duration of showers possible
depends
upon the quantity of water stored and the tempera-
Thermal
storage in
Internal
heat gains
Thermal
storage
system
Rotating
Rotating
Vapor
Coal Steam shaft
Electricity Electrical Electricity Electric shaft compressior
Boiler Turbine Generator
distribution
motor refrigeration
network
system
Heat rejected
to the
environment
e d
l a
Without any intervening energy storage along the path
i o
h
shown in Fig. 1, the rate of energy transferred from the
C Storage L
To To From
outdoor air temperature, incident solar radiation) and the q (t)
storage storage storage
building’s internal air temperature. The interior temperature
can vary over a relatively small range of comfortable interior
#
##yy##yy##
Time
and peak electrical demand. Morris et al. (3) present results
from experimental and simulated optimal control of building Figure
3. An example building cooling load profile over a 48-h
##yy##yy##
sized only to meet the difference between the building’s peak cient
rate or at relatively low cost) and discharge it over an-
cooling load and the undersized chiller’s capacity. other.
However, in most thermal storage systems, the rate
Storage size can be determined by assuming a load profile and
duration of charging and discharging the store can be
that must be met and the capacity of the thermal energy
controlled. To maximize the benefits of thermal storage, an
source. An example cooling load profile over a 48-h period is optimal
control strategy that minimizes cost or some other
shown in Fig. 3. One possible profile for operating the chiller desired
criterion should be employed (15).
is also shown. The quantity of energy storage required is indi- In
some cases, when the heat source or sink either will not
cated by the integrated area below the required load curve be
available in the future or will be available only at great
and above the available chiller load curve. In order to charge cost,
the optimal strategy is obvious. In a solar heating sys-
the store, the integrated area where the chiller capacity ex- tem,
the incident solar energy is free, and, if the heating re-
ceeds the required load must exceed the area indicating
quirement exceeds the maximum amount of heat storage, the
storage. store
should be charged as much as possible. If the on- and
With thermal storage, the chiller is decoupled from the off-peak
rates of an electrical rate structure differ by orders
load and can have almost any profile of operation so long as of
magnitude, a chilled water system should be charged fully
the integrated area associated with charging the store ex- (or at
least to some amount greater than the expected next
ceeds the area associated with storage discharge. However, day’s
cooling load) when the lower rate is available.
there is at least one profile of chiller operation that minimizes In
most applications, the best or optimal control strategy
energy costs over the subject time period. The storage indi- is not
entirely obvious, but may be determined through simu-
cated by this optimal chiller operation profile is the optimal lation
or other analysis. A general outline of an approach us-
energy storage requirement over the subject period. But for a ing
simulation follows:
valid economic analysis, operation over a longer period, such
as the anticipated life of the system, along with equipment
Develop a model of the system including the storage unit
costs must be considered.
and the load served by the store. Often weather-depen-
Models of the thermal storage system, chiller system, and
dent or otherwise time-varying loads can be predicted
predicted load profiles can be used to estimate operating costs
using statistical models.
over a given period of time. Weather data used in the simula-
Identify controlled variables such as the mass flow rate of
tion can be real measured data or meteorological data which
the fluid entering the store. Identify the external time
simulates a typical year. Various storage charging and dis-
varying uncontrolled variables such as the predicted
charging schedules along with various storage sizes and
heat-transfer load to be served by the storage system.
chiller capacities can be simulated to determine an optimal
In some cases the thermal loads may be calculated from
storage size. a
building model where the external variables will in-
Electrical rates that differ depending on the time of day
clude the outdoor weather conditions.
are usually the primary influence in determining the opti- The
model should reasonably agree with the actual pro-
mum tank size as well as in determining the best operating
cess. Model parameters, if present (such as rate param-
schedule (or profile) for charging and discharging thermal
eters), can be determined using measured data for pro-
storage systems. Before deregulation of electrical utility sys-
cesses already operational. If measured data are not
tems in the United States, a common electrical rate structure
available, manufacturers frequently provide perfor-
applied to commercial or industrial customers included time
mance data in tabular or graphical form.
of day rates (usually one ‘‘on-peak’’ rate and one ‘‘off-peak’’
Define the objective function to be minimized, such as elec-
rate) for energy use (kW-hr) and a demand charge for the
THERMAL MAGNETORESISTANCE 25
CONCLUSION
sign, Analysis and Improvement of Energy Systems, New York:
ASME, 1991.
127–139, 1992.
on the installation of thermal storage systems is considered
by Caldwell and Bahnfleth (16).
BARRETT A. FLAKE
This article has presented the basic uses and operation of
BIBLIOGRAPHY
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3043.htm
} { { } }
HOME
ABOUT US
●
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Thermionic Conversion
Abstract
Background
Work Function
Modes of Operation
Conversion Efficiency
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3043.htm17.06.2008 16:48:12
THERMIONIC CONVERSION 49
THERMIONIC
CONVERSION
Thermionic
(TI) energy conversion relies on the emission of
electrons by
solids heated to a sufficiently high temperature.
50 THERMIONIC CONVERSION
external load are ignored, then the work done by the electrons
voltage drop due to the flow of the electrons through the inter-
Electron
plasma-filled gap)
tion into space nuclear reactor power systems. In 1987 and
1988, the USSR announced the successful operation of two
2. Minimizing the voltage drop in the interelectrode gap
thermionic power systems in space for 6 and 12 months,
3. Minimizing joule losses in the electrodes and the leads
respectively. These power systems, which used in-core
to the load
multicell thermionic fuel elements (TFEs), became known
4. Minimizing the collector work function
later in the West as TOPAZ-I. A number of Soviet TOPAZ-
II systems, which employed in-core single-cell thermionic
WORK
FUNCTION
fuel elements, had been built and ground tested successfully
in late 1970s and in the 1980s, using tungsten electric
duction band are free to move inside a metal, for them to es-
deployed in orbit.
cape
the metal surface they must have sufficient energy to
Following the end of the Cold War, a joint technology pro-
electron’s positive image charge inside the metal and the un-
dom and France. The objectives of the program were to (1)
2 2 −(eϕ/kT )
tured by the collector. They then return to the emitter
J (A/cm ) = AT e (1)
through an external load, hence producing electrical power
(Fig. 1). In
Eq. (1), e and me are the electron charge and mass, respec-
The thermodynamic cycle that describes the operation of a
tively, k is the Boltzmann constant, and T is the absolute tem-
thermionic converter resembles that of Carnot, in which ther-
perature of the metal surface. For an ideal, isothermal sur-
face,
the constant, A (A/cm2 2
mal energy is ideally supplied to the emitter and removed
#K ), derived theoretically by
THERMIONIC CONVERSION 51
5
Table 2. Bare Work Functions of Typical Electrode Materials
n W
(110) 5.25 Pt (111) 5.7
o
i
t W
(111) 4.47 Re polycrystal 5.0
c 3
n
Ir polycrystal 5.27 Re (1011) 5.75
u
f Collector
k
Ir (110) 5.42 Ta polycrystal 4.12
r (typical)
o
Ir (111) 5.76 Ta (100) 4.15
W Polycrystal W
2 3
The emitter temperature and the operating cesium reser-
A = (4πem k /h ) = 120.4 (2)
e
voir temperature (or vapor pressure) are, therefore, selected
to
reduce the effect of the space charge near the emitter and
In Eq. (2), h is Planck’s constant. The actual value of A ,
at
the same time increase electron emission. For a given emit-
however, is determined experimentally and depends on the
tron emission and increasing the voltage drop in the gap. The
and the constant, A , for a number of typical electrode materi-
cesium at the proper vapor pressure into the gap. The Cs ions
[Eq. (2)].
for the plasma in the gap are produced mainly by surface ion-
All refractory metals, capable of prolonged operation at
n g i i
controlling their respective temperatures and the cesium va-
= e kT (3)
por pressure in the interelectrode gap (Fig. 2). When Cs
ntot g o
(3)
Electrode Bare Work A Constant
Material Function, 2 2
# (eV) (A/cm # K )
ni (x ) # n2 (x ) #
Molybdenum (Mo) 4.2 55
#(x ) = 1 − i (4)
τ n2 (T )
Nickel (Ni) 4.61 30
i e
Tantalum (Ta) 4.19 55
Tungsten (W) 4.52 60
where # is the effective ionization time, n (x) is the local
i i
4.54 46
W
Ba 1.6 1.5 plasma
concentration, and n(Te) is the equilibrium plasma
52 THERMIONIC CONVERSION
MODES OF OPERATION
sen discharge, can be written as
# #
o 1 1 2AT 2 r me
those that operate in the Knudsen and ignited modes of dis-
ϕE = Ei + kTE ln E (7)
charge. The major features of the different discharge mode
2 2 eno mCs
E i e e Cs
Vacuum and Quasi-Vacuum Modes of Discharge
excess ions and the converter is said to operate in the over-
− e ϕ E e (ϕE −V )
High-temperature operation in vacuum converters, however,
2 kT 2 − kT
j
E = ATE e E and j C = ATC e c for V ≤ ϕE − ϕC
is not possible because it risks closing the gap, due to the
thermal expansion of the electrodes, and shorting the con-
(8a)
verter.
e (ϕ +V ) e ϕ
C C
In the quasi-vacuum mode converters, Cs is introduced j
= AT 2e− kTE and j = AT 2e− kTC for V > ϕ − ϕ
E E C C E C
into the interelectrode gap, but its pressure is kept quite low
(8b)
(a few millitorrs), just enough to adjust the electrodes’ work
function, as discussed earlier, but not sufficient to neutralize
and
the space charge near the emitter surface. Very little or no
surface or impact ionization of cesium takes place in the gap,
j = j E − j C (8c)
since the electron mean free path, # , is much larger than the
e
tor, respectively.
function by a certain amount, #, that depends on the extent
The voltampere characteristics of an ideal, Knudsen mode
of the space-charge region near the emitter surface. In this
THERMIONIC CONVERSION 53
j
characterized by the ion generation in the gap by volume ion-
Saturation region Obstructed region
ization of Cs atom due to the collision with emitted electrons.
In
this mode of discharge, the interelectrode voltage is low
)
2 enough
and the Cs pressure is high enough that the emitted
m
c Transition point
electrons gain sufficient kinetic energy, as they are acceler-
/
A
( Saturation ated
in the sheath near the emitter surface, to cause volume
y
t
i current
ionization of Cs atoms in the gap.
s
n
When the load voltage is slightly lower than the ignition
e
d j E
t
voltage, Vign , the overcompensated plasma in the interelec-
n
e trode
gap does not occupy the entire gap and a small region,
r
r
u j C with
a negative space charge, V*, exists near the emitter.
C
This
space charge is called virtual cathode because the effec-
tive
emitter work function now becomes #eff # #E
V*. The
0
corresponding portion of the
voltampere characteristics is
ϕ E – ϕ C V
called the obstructed region [Figs. 4(a) and (b)]. As the con-
Open circuit
voltage verter
voltage is reduced further, V* is also reduced, until
Load voltage (V) the
transition point or the ‘‘knee’’ is reached, at which the ion
ϕ C
electrons reach the collector. In addition, the high concentra-
tion
of positive Cs ions near the emitter creates a strong posi-
ϕ E ϕ E ϕ C ϕ E
tive
electric field on the emitter surface, increasing the satu-
V V ration
current, due to the Schottky effect, above the emission
V
current given by Eq. (1). The voltampere characteristics of an
j = −eu n − eD e − u n (1 + β )k e (9a)
The diffusion mode of discharge occurs when the emission
e e e dx e dx e e e dx
dV n T
mode voltampere characteristics and the electron motive dia-
j = −eu n + eD e + u n (1 + β )k e − R (9b)
i i i dx i dx i i i dx i
grams are qualitatively similar to those of the ignited mode.
Because of the significant losses of electron energy, diffusion
where
#e and #i are the electron and ion collision coefficients,
mode discharge is not desirable in thermionic conversion for
and Ri
is the friction force due to ion-electron collisions. The
electric power generation, due to the low conversion effi-
dx dx
density.
In diffusion mode discharge, as the load voltage is lowered, The
equations for the electron energy flux, Se, and its energy
the emitted electrons gain more kinetic energy as they are
balance are
accelerated by the interelectrode voltage drop, V. When
their energy becomes sufficiently high to cause volume ioniza-
## # #
5 kT T
tion, the operation mode of the converter transitions to the
Se = −j e + βe e + V − λe e (11a)
2 e dx
ignition mode discharge. The ignited mode discharge is char-
acterized by overcompensated plasma in the interelectrode and
gap, due to the high ionization rate of Cs atoms and the low
ion diffusion rate as compared with that of the electrons. The
dSe
emitter, V , and collector, V , plasma voltage drops change
= −Ei #(x ) − #Srad − #Sei − #Sea (11b)
E C
dx
their sign as compared with the diffusion mode (i.e., the emis-
sion electrons are accelerated near the emitter and deceler- In
these equations, #e is the thermal conductivity of electrons,
ated near the collector) (Fig. 4). and
S , S , and S are the losses of electron energy by
rad ei ea
respectively.
This is the operation mode of choice in the state-of-the art
When supplemented by the appropriate boundary condi-
thermionic converters for electric power applications. It is tions,
electron energy fluxes to and from plasma at both elec-
----------------------- Page 563-----------------------
54 THERMIONIC CONVERSION
)
2
m
Transition point
c
/
A
(
,
Saturation
y
t
i
current
s
n
e
Ignition
d
t
n
e
Obstructed Unignited
r
r
region (diffusion)
u
C
mode
V V
ign L
∆V ∆VE ∆V
Vd ∆VE
Vd ∆VE Vd ∆VC
∆VE
∆VC ϕ C
ϕ E ∆VC ϕ E
∆VC ϕ E ϕ C ϕ E
ϕ C
ϕ C
trode surfaces, Eqs. (9) to (11) could be solved for the distribu-
trodes. The latter depends on the emitter temperature and
tions of T , n, V, j , j , and S across the interelectrode gap, the
electrodes geometric mean temperature, T # (T
e e i e
m E
and the net current discharge as functions of the interelec-
TC)0.5, and is given as
trode voltage.
eff E E C m
CONVERSION EFFICIENCY
cond Cs E C E C Cs
∗ 3 3 2
in the leads by joule effects. The electron cooling heat flux,
T = 2/3{(T − T )/(T − T ) } (17)
E C E C
qec, is given as
For
optimized leads, q and q are given in terms of the con-
Iead j
J 2 −e ϕ
version efficiency of the converter, as follows (18):
qec = (2kT + eϕ)T e kT (13)
e
lead o E C
The thermal radiation heat flux from the emitter to the collec-
tor through the interelectrode gap, qrad , is given as and
q = σ # (T4 − T4 ) (14)
q = J [L (T2 − T2 )(η/2 − η)]0.5 (18b)
rad eff E C
j o E C
THERMIONIC CONVERSION 55
Emitter
η = (V − V )J /q (20)
lead o in
where V is the measured terminal voltage of the converter.
This equation is solved for the efficiency after substituting for
the input heat flux to the emitter, qin [Eqs. (13) to (19)]. The
Electric load
total heat rejection to the heat sink from the collector and the
leads, qrej , is given as
56 THERMIONIC CONVERSION
stainless steel has been used as the structure material for the
diator.
Nb collar
Sc O spacer
2 3
W/cm2 2
Nb collar
Emitter connection
Many investigations have attempted to increase the conver-
THERMIONIC CONVERSION 57
o
l
tor surface when covered with tungsten oxide deposits m
1 torr Qth = 3500 W
from the emitter surface, which increases the load u
m
i
voltage. x 150
a
M
In recent laboratory tests (20), conversion efficiencies of
0.75 torr
2 120
10 10 10 10 10
been reported. These tests were performed using a planar
O/Cs
TIC. Oxygen vapor pressure up to 8 10#6 Pa was intro-
high-
temperature Cs converters). In addition to neutralizing
the
space charge, Cs vapor serves as the transport media for
15 the
electrons from the emitter to the collector. The Ba vapor
y
t tions
of the electrodes. Consequently, it is possible to operate
i
s
n Cs–Ba
converters in the Knudsen mode, where scattering
e
d 5 Westinghouse losses
are almost nil and the maximum electric power output
r
e Bettis could
be derived at a terminal voltage that is equal to the
w
o contact
potential difference, (#Eo # #co).
P
The
optimum current density from the emitter surface,
0 which
corresponds to the transition from the unignited to the
0 0.2 0.4 0.6 0.8 1.0
ignited
mode of discharge, Jo, is independent of the type of
Interelectrode gap (mm)
2
the
emitter material ( 1.5 A/cm ) but increases with the ce-
Figure 7. Reported electric power densities for TICs with oxygen- sium
pressure, raised to the one-half power [i.e., Jo # (PCs)1/2,
ated electrodes (20). up to
2 10#2 torr] (25). These investigators have demon-
58 THERMIONIC CONVERSION
c
because of the scattering of electrons by cesium atoms in the
/
W
interelectrode gap.
( 10 1.7 eV
t
The emitter temperature strongly affects the value of Jo .
i
s
#2
n
At PCs # 10 torr, Jo almost doubles in value with increasing
e 5
d Gap size — 1 mm
e
1900 K. For example, at TE # 1900, 2000, and 2100 K, the
w Collector work
2
o 2
measured J was 2, 4, and 8 A/cm , respectively (25). The
P function — 2.2 eV
o
1
emitter temperature increases (e.g., #Eo , # 3.27, 3.65, and 4.04
1900 2000 2100 2200 2300 2400
eV at emitter temperatures, TE , of 2000, 2500, and 3000 K, re-
Emitter temperature (K)
spectively).
The only effect the emitter material has on Jo is to change
Figure 9. Electric power densities versus emitter temperature for
the value of the barium pressure at which Jo occurs. This Cs–Ba
thermionic converters (28).
change is due to the difference in the sorption energies of the
barium atoms onto the different electrode materials; the
higher the sorption energy of barium, the lower is the opti-
THERMIONIC CONVERSION 59
e– setup
and procedures, reported results were inconclusive and
described by Shimada.
1. electrons for cesium ionization in the nongrooved por-
Tskhakaya et al. (36) results also showed that the opti-
tion of the emitter surface, which operates in the diffu- mum
gap size for their grooved emitter converter (1.6 mm)
sion mode was
higher than that for a smooth electrode converter (1.0
mm).
They have also tested a converter with a grooved emit-
2. Cesium ions, which diffuse from the grooved to the non-
ter
and a zirconium oxide collector that had the same grooves
grooved portion of the emitter surface, contributing to
as
the emitter. In these experiments, the collector’s grooves
the neutralization of the space charge, hence lowering
were
laid in the perpendicular direction to the emitter’s
the optimum cesium pressure (32,33) [Fig. 10(a)].
grooves.
60 THERMIONIC CONVERSION
sure (120 Pa) than the peak electric power for the latter (0.31 onic
converter operating at an emitter temperature of 1900 K
W/cm2 at 150 Pa). and a
collector temperature of 800 K would have to deliver a
They also reported that at cesium pressure less than 110 ther
enhance the performance of grooved electrodes convert-
Pa, the electric power output of the grooved collector con- ers, in
terms of both higher efficiency and electric power den-
verter was also higher than that of the smooth electrode con- sity.
Further research on grooved electrode converters could
verter, but lower than that for the grooved emitter converter be
rewarded handsomely by opening a broad range of poten-
(32). At higher cesium pressure, the electric power of the tial
application for commercial electric power generation.
grooved collector converter was the highest, and peaked at
the same cesium pressure as the smooth electrodes converter
(150 Pa). The electric power output of the grooved collector Effect
of Inert Gases on the Performance of Cesium TICs
converter was 1.4 times that of the smooth electrodes con-
There
was a lot of interest in the 1960s and 1970s in investi-
verter, which is the same as the increase in the grooved col- gating
the effect of inert gases, such as helium and xenon,
lector surface area. Since the use of a grooved collector does on the
performance of Cs converters. Inert gases affect the
not affect space-charge neutralization at the emitter surface,
THERMIONIC CONVERSION 61
He gap
of Cs ions by inert gases and negligible scattering losses. At
high cesium pressure, there should be no effect if scattering
Spacer
Collector
the introduction of inert gases would improve the perfor-
mance of Cs converters due to two effects: (1) an increase in
Emitter
the plasma density and enhanced plasma distribution in the
gap, and (2) a change in the potential distribution, resulting
Figure 12. A schematic of a microgap converter.
62 THERMIONIC CONVERSION
35
Quasi-vacuum
mode (d = 10 m)(Fitzpatrick et al. 1993 & 1998)µ
µ
Quasi-vacuum
mode (d = 6 m)(Nikolaev et al. 1996)
µ
Knudsen mode (d
= 20 m, eff = 0.05)(Nikolaev et al. 1996)
30
µ
Quasi-vacuum
mode (d = 4 m)(Nikolaev et al. 1996)
µ
Knudsen mode (d
= 20 m, eff = 0.17)(Nikolaev et al. 1996)
)
% 25
(
y
c
n
e
i
c
i
f
f 20
e
n
o
i
s
r
e
v
n
o 15
C
10
THERMIONIC CONVERSION 63
Open Short
Cause
Circuit Circuit Degradation
Emitter:
Change in position, emission coating
Separation
Damage or deformation
Loss of vacuum tightness, release of fission products
Contamination and mass transfer
Interelectrode gap:
Lack of cesium
Penetration of contaminants (from outside or from
nuclear fuel)
Increase in heat losses
Spaces:
Destruction or damage
Decrease of contact thermal resistance
Deposition of conducting layers
Collector assembly:
Deposition of contaminants and reaction products
Change of cooling conditions
Increase in thermal resistance deformation
Insulation failure
Emitter-collector commutation:
Destruction
Cross-section reduction
The most commonly used ceramic materials in TI converters
Characterization
are oxides of Al, Be, Zr, Mg, Th, and Sc.
j A (RA , ϕ E, ϕ C)
taminants or ambient air into the interelectrode gap; (2)
j exp
changes in heat rejection, increasing the coolant and the col-
A A
64 THERMIONIC CONVERSION
THERMIONIC CONVERSION 65
Fuel
BIBLIOGRAPHY
Combustor
of the bare work functions, Proc. 27th Annu. Conf. Phys. Electron.,
Recent advances in natural-gas-fired gas turbines, such as
Cambridge, MA, 1967, pp. 160–168.
integrated gasification combined cycle (IGCC) and pressur-
9. G. N. Hastopoulos and F. Rufeh, Thermodynamic correlation of
ized fluidized bed combustion (PFBC) technologies, have
work function, Energy Convers., 10: 135, 1970.
made using a thermionic topping cycle that employs state-of-
10. G. N. Hastopoulos and E. P. Gystopoulos, On the relation be-
the-art or near-term advanced TI technology even less attrac-
tween work function and other thermophysical properties, Proc.
tive. Recent studies of using thermionic topping cycles have
Thermionic Convers. Spec. Conf., Miami Beach, FL, 1970, pp.
shown that, even using advanced thermionic technology (for
203–209.
example, microgap converters), the efficiency of combined cy-
11. T. Alleau and M. Bacal, Influence of oxygen adsorption and oxy-
cle power plants that employ advanced GTs, such as Westing-
gen-cesium coadsorption on (100) tungsten work function, Proc.
house 501ATS, can be increased only by less than 1.5 percent-
Thermionic Convers. Spec. Conf., Miami Beach, FL, 1970, pp.
434–440.
age points (71). Such a small performance gain does not
justify using a thermionic topping cycle, considering the tech-
12. J.-L. Denisot, Electron Emission Experimental Study of a (100)
Tungsten Single Crystal Covered wtih Cesium and Oxygen, CEA-
nology development required and the risks involved. As far as
`
Hill, 1997.
culminated in manufacturing and testing a thermionic mod-
66 THERMIONIC CONVERSION
1996.
onic converter at current saturation, Sov. Phys.—Tech. Phys., 17:
1695–1700, 1973.
48. V. V. Sinyavskii et al., Design and Testing of Thermionic Fuel
Systems, Rep. SPI-52-1, San Jose, CA: Space Power Inc., 1990.
Proc. 12th Intersoc. Energy Convers. Eng. Conf. (IECEC), Soc. Au-
tomot. Eng., 1977, pp. 1568–1574.
56. J. G. Jacox et al., Small ex-core heat pipe thermionic reactor con-
ics, 1994.
vers. Manage., 38: 533–549, 1996.
onic diode, Proc. 10th Annu. Power Sources Conf., Atlantic City,
1966 (in Russian).
NJ, 1966.
42. F. V. Kondrat’ev, G. V. Sinyutin, and V. F. Tikhonov, Effect of
45. F. Rufeh and D. Lieb, The influence of inert gases on the I-V
IECEC, 1978, pp. 1893–1897.
characteristics of a cesium diode, J. Energy Convers., 10: 149–
67. G. Miscolczy et al., Thermionic combustor application to com-
153, 1970.
bined gas and steam turbine power plants, Proc. 16th IECEC,
46. R. Ya. Kucherov et al., J. Sov. Phys.—Tech. Phys., 15: 1937, 1971.
1981, pp. 1956–1961.
THERMISTORS 67
MOHAMED S. EL-GENK
University of New Mexico
DMITRY V. PARAMONOV
Westinghouse Science and
Technology Center
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3044.htm
} { { } }
●
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Thermoelectric Conversion
Abstract
The sections in this article are
Thermoelectric Effects
Thermoelectric Refrigeration
Concluding Remarks
Acknowledgments
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3044.htm (1 of 2)17.06.2008 16:48:32
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3044.htm
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3044.htm (2 of 2)17.06.2008 16:48:32
80 THERMOELECTRIC CONVERSION
THERMOELECTRIC CONVERSION
THERMOELECTRIC EFFECTS
B B
mann’s
constant. A thermoelectric effect occurs when the
most
energetic (or ‘‘hot’’) electrons and holes are transported
THERMOELECTRIC CONVERSION 81
Heat load/source
leave in the same way as the evaporation of water has a cool-
ing effect.
R
There are several distinct but related physical effects that
arise from this heat capacity of electrons and holes. When an
n Q p
Heat sink
effect, which is the basis of thermoelectric refrigeration.
When one end of a piece of thermoelectric material is hotter
(a)
than the other, a voltage develops across the material. This
is called the Seebeck effect, which is used in thermoelectric
Heat load/source
power generation. Finally, when a temperature gradient ex-
ists inside a thermoelectric material, a gradient in the electric
potential develops due to the Thomson effect. We will not dis-
Q Q Q
cuss the Thomson effect here, since it is beyond the scope of
n p n p n p
this article, but more details may be found in Refs. 1–3.
These effects are known in the science of thermodynamics
I I I I
as reversible processes (1), in the sense that they may operate
Heat sink
in the forward or in reverse directions. For instance, by re-
versing the direction of current flow in a thermoelectric refrig-
(b)
erator, the direction of heat flow is reversed, thereby causing
heating instead of cooling. The efficiency of thermoelectric re-
frigerators and generators is limited by certain irreversible
phenomena (1). The two main irreversible effects are thermal
kBT
conduction and electrical resistance.
When a temperature gradient exists in any material, heat
flows through the material to minimize the temperature gra-
(ii) E F
dient. This is clearly an irreversible process, otherwise objects
(i) (iii)
would spontaneously develop cold and hot spots until they
(iii) Eg E F (i)
82 THERMOELECTRIC CONVERSION
2 2
To understand the microscopic operation of a thermoelec-
the amount of heat I R is generated in n and I R is generated
n p
n p n
Q = −# I − K #T − 1I 2R (2)
to traverse the circuit. It is also important to note where the
n n n 2 n
different energy-exchange mechanisms are taking place. In
Step (i), the electron and hole absorb thermal energy from the
and that flowing from R to Ep is
Q = # I − K #T − I R (3)
sorbs heat from the heat source in the case of thermoelectric
p p p 2 p
Finally, in Step (iii), the electron and hole release their excita-
Q = Q + Q − Q = (# − # )I − (K + K )#T
cooling n p L p n n p
tion energy to the electrodes E and E , which then release
(4)
n p
1 2
the heat to the heat sink. The heat leak due to the thermal
− 2 I (Rn + Rp ) − QL
THERMOELECTRIC CONVERSION 83
(9)
perature difference #T, then R will cool until it sits at a tem-
−# = # ≡ #
n p
perature below the ‘‘ambient’’ temperature T.
The coefficient of performance of a thermoelectric refrig-
so
that the coefficient of performance simplifies to
erator is defined to be the ratio of the cooling power to the
electrical power P required to drive the current I through the
#I − K #T − 1I 2R
refrigerator
ϕ = 2 2 (5## )
Iα#T + I R
Qcooling
ϕ = P (5) We can
further simplify Eq. (5
) using the Kelvin relationship
T − #T T − #T
the text in curly brackets indicates the direction of positive
ϕmax = #
contact potential, in this case from thermoelectric material
#T 1 + ZT + 1 ###
(5 )
potential is called the Seebeck coefficient. Because the junc-
#1 + ZT − ϕ −1
Carnot
tions at R and those at En and Ep sit at different tempera-
= ϕCarnot #1 + ZT + 1
V = V {E , n} + V {n, R}
n contact n contact where
T # 1/2(T # (T # #T)) is the mean temperature in the
= −V {n, E } + V {n, R}
thermoelements and Carnot # (T # #T)/#T is the Carnot ther-
contact n contact
= −(αn T + V ) + (αn (T − #T ) + V )
modynamic maximum efficiency of a refrigerator (1–3). The
0 0
quantity
= −αn #T ;
(7)
Vp = Vcontact{R, p} + Vcontact{p, E }
α2
p
Z = (11)
= −Vcontact{p, R} + Vcontact{p, E }
RK
p
= −(α (T − #T ) + V ) + (α T + V )
p 0 p 0
is
known as thethermoelectric figure of merit, and has the
= α #T
p
dimensions of inverse temperature. Thus, the quantity
α
is also an Ohmic voltage drop I(R # R ) in the circuit; thus
ZT = T (12)
n p
RK
the total power supplied to the thermoelectric refrigerator is
P = IV is
known as the dimensionless figure of merit, which summa-
2 (8) rizes
the thermoelectric performance of a material at temper-
= I (αp − αn )#T + I (Rn + Rp ) ature
T. More analytical details of the efficiency of a thermo-
status
of thermoelectric technology.
Qcooling
ϕ = The
material parameters of the thermoelements enter Eq.
P
1 2 (5# ) (5
) only through Z . In this sense, the figure of merit summa-
(# − # )I − (K + K )#T − I (R + R ) − Q
p n n p 2 n p L rizes
the thermoelectric performance of a thermoelectric ma-
=
I (αp − αn )#T + I 2 (Rn + Rp )
terial. Although Eq. (11) includes the electrical parameters R
and
K , we can see that it may also be defined in terms of
Generally, one would like to know the best coefficient of
intensive material parameters. For instance, suppose that the
performance that a given material system can supply. For
thermoelements are of uniform cross-sectional area A and
this case, we consider the limit with no load (QL # 0). Also, length
L . Then the thermal conductance and electrical resis-
84 THERMOELECTRIC CONVERSION
A L
K = κ ; R = ρ (13)
L A
2
α
Z = (11# )
ρκ
2 3
2
ZT = α T (12# )
literature from Marlow Industries, Inc.
ρκ
2 3
heat rejected for a given amount of cooling. Thus, thermoelec-
available thermoelectric refrigerators (such as the one shown
tric materials have not yet offered a strong competitor in the in
Fig. 3), and may attain ZT values approaching 1. In fact,
area of large-scale refrigeration or air conditioning.
K
THERMOELECTRIC POWER GENERATION
The
circuit of Fig. 1(a) may be used as a thermoelectric gener-
ator
by heating R. Heat QS is supplied from a heat source to
the
hot electrode, which causes a flow of thermoelectric cur-
rent
I . The thermal conductance K of the thermoelements
P IV
η = = (14)
Qs Qs
Carrier concentration
where
P is the electrical power supplied to a load through the
Figure 2. Schematic graph of the properties of thermoelectric mate-
ambient-temperature electrodes and V is the total voltage
rials as a function of electron concentration (1). drop
across the generator. By following a line of reasoning
THERMOELECTRIC CONVERSION 85
η = #
talline quality and purity by using methods such as molecu-
T
T + #T 1 + ZT + T + #T lar-beam
epitaxy (11). With this capability, traditional semi-
# conducting
materials may be formed into structures not
# #
(14 )
1 + ZT − 1 available
to the original designers of thermoelectric devices in
#
= ηCarnot the 1950s
and 1960s.
1 + ZT + (1 − ηCarnot)
86 THERMOELECTRIC CONVERSION
ergy
as an electron excited an equal amount above EF . If this
T > T (ii) Fermi
sea of electrons is at temperature T, the average exci-
0
0
tation energy of its electrons and holes from EF is roughly
T
c kBT
(3). In addition, we use the term ‘‘hole’’ hereafter to refer
i
n
o to a
vacant electronic state in a metal, which is more general
r
t
c than
the definition used for p -type semiconductor material
e (iii)
l
E (3).
This nomenclature simplifies the understanding of a ther-
Phonon E
moelectric device.
heating
The voltage differences between different pieces of metal,
O K g (E ) such
as the electrodes and the reservoir of Fig. 4(a), are re-
(b)
structures marked ‘‘electron tunneling channel’’ and ‘‘hole
electrons and holes are excited from the Fermi levels, remov-
ing
the excited electrons and holes should cool the remaining
to thermodynamic properties, which are exploited in the dilu-
electrons. The alternative to this selective process is to allow
tion refrigerator, the most common means of performing ex- both
electrons and holes to move arbitrarily through both in-
perimental work at temperatures between 0.01 and 1 K.
terfaces, resulting in a loss of control over the direction of
The low-temperature properties of matter of interest here heat
flow. The preferential control over the direction of heat
are related to phonons, the quantum-mechanical manifesta- flow
during electrical transport is the basis of thermoelectric
tion of vibrations in the crystalline lattice. Phonons are the
refrigeration.
predominant repository of heat in solids at temperatures
Thus, for thermoelectric refrigeration to be possible, it is
more than a few kelvins above absolute zero. Near absolute
required to preferentially remove excited charge carriers from
zero, however, electrons in a metallic solid hold most of the near
the Fermi level. The conventional way of doing this is to
heat, and hence thermal energy. Thus, cooling the electrons use
a negative thermoelement, which carries thermally ex-
in a metal at these low temperatures should provide an effec- cited
electrons away from the reservoir, and a positive ther-
tive means of refrigerating the entire metal.
thermoelement.
acts with electrons in the solid, analogous to the way that
broad swells on the sea do not overturn a small boat as do
As was mentioned in the foregoing, an important differ-
abrupt breakers. ence
between the thermal properties of matter at room tem-
This ‘‘decoupling’’ of the electrons and the phonons in a
perature and cryogenic temperatures is the fact that the elec-
solid enables the operation of certain devices based upon ‘‘bal- trons
hold most of the heat within 1 K of absolute zero.
listic’’ electron transport (12). It also allows the temperature
Combined with the fact that electrons and phonons and other
of the electrons in small devices to be controlled indepen-
degrees of freedom of a metal tend to decouple at cryogenic
dently of the phonon temperature. This is the basis of the
temperatures, this gives rise to some exotic possibilities for
first experimental demonstrations of cryogenic thermoelectric
thermoelectric refrigeration. Figure 4(b) shows a diagram of
refrigeration, to be discussed in the following. the
ambient temperature T versus the electronic tempera-
The structure and operation regime of a cryogenic thermo- ture
To for a ‘‘generic’’ thermoelectric refrigerator as shown in
electric refrigerator are depicted in Fig. 4 (5). In Fig. 4(a), the Fig.
4(a), in which we assume that the electrons are cooled
88 THERMOELECTRIC CONVERSION
membrane
collection of electrons is not strongly coupled to the crystal
A1 electrode Pb electrode
lattice and hence is not heated by the absorption of phonons.
Thus, quantum-effect thermoelectric refrigeration might be a
viable means of cooling microelectronic components that rely
upon sub-1 K material properties (e.g., superconductivity or
ballistic transport) in which a temperature difference between
electrons and phonons is acceptable.
As it turns out, a thermoelectric refrigerator with quantum
dots as thermoelements may be capable of bulk cooling at
temperatures a thousand times lower (1 mK) or below (5).
Due to the fine scale of the energy levels of quantum dots,
r2 r1
Substrate at
this might be an effective means of microkelvin or nanokelvin
temperature T
cooling of metals. Such refrigerators have not been tested,
however, thus their true performance is an open question.
(a)
However, the thermoelectric transport properties of various
quantum structures have been measured and found to be con-
sistent with the assumptions of the theoretical simulations of
Ref. 5. The greatest experimental challenge in the realization
of a quantum-dot thermoelectric refrigerator remains the ac-
2∆ Pb
curate measurement of the temperature of the electrons in R .
2∆Al
(b)
is simple to model and may be capable of refrigeration near
the Carnot limit, there are other low-temperature phenomena
Figure 6. (a) Schematic and (b) energy-level diagram of the normal-
which may be exploited in the lab to perform refrigeration.
insulator-superconductor (NIS) thermoelectric refrigerator. Hot elec-
One such system is a tunneling junction between a supercon-
trons and holes in the Cu reservoir R tunnel into quasi-particle states
ducting metal and one in the ‘‘normal’’ state. A normal-insu- in
the Al superconducting thermoelement, cooling R . The Pb contact
lating-superconductor (NIS) refrigerator was demonstrated
electrode is used to inject charge into R , but has no net heating or
THERMOELECTRIC CONVERSION 89
90 THERMOPILES
SENSORS.
7. M. Nahum, T. M. Eiles, and J. M. Martinis, Electronic microre-
THERMOMETERS. See TEMPERATURE SENSORS.
frigerator based on a normal-insulator-superconductor tunnel
junction, Appl. Phys. Lett., 65 (24): 3123–3125, 1994.
Reading List
HAL EDWARDS
Texas Instruments, Inc.
QIAN NIU
ALEX dE LOZANNE
The University of Texas at Austin
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3009.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Turbogenerators
● Email this page Enter words or
Standard Article
Abstract
Fundamentals
Excitation Source
Construction Features
Thermal Design
Turbogenerator Applications
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3009.htm (1 of 2)17.06.2008 16:49:12
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3009.htm
Operation of Turbogenerators
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20...ENGINEERING/22.%20Energy
%20Conversion/W3009.htm (2 of 2)17.06.2008 16:49:12
TURBOGENERATORS 595
A
B C
C B
A A
B C
TURBOGENERATORS
C B
FUNDAMENTALS
tric and mechanical losses. In addition to these basic machin-
ery problems, there are various other issues to consider, such
as designing the generator to be suitable for the turbine Modern
turbogenerators usually operate at one of four rota-
which drives it, meeting special terminal voltage require- tional
speeds:
120f
1903 to the 1500 MW generators available in 1998. Figure
ns =
1 shows a typical installation of a 400 MW hydrogen-cooled
P
596 TURBOGENERATORS
tor. As the rotor turns, the voltage in each conductor peaks,
goes to zero and reverses, creating approximately a sine wave.
For the two-pole generator shown, one complete rotor revolu-
tion generates one complete electric cycle.
All large alternating current generators have the stator
conductors connected as three separate windings (or phases)
spaced 120# apart around the periphery of the stator. Thus
the voltages generated in each of these windings is 120# out
of phase with the others. The cross section of Fig. 2 shows a
typical configuration with two conductors per stator slot, each
phase labeled A, B, or C. The top coil bars are connected to
bottom coil bars of the same phase in the ends of the machine
forming a coil-type winding. The ends of each winding are
brought out to terminals, usually at one end of the generator.
Although all six terminals are usually accessible from outside
Figure
4. Two-pole hydrogen inner cooled generator axial cross
the generator housing, three are usually connected together, section.
p T
rents that flow are also 120# out phase and thus create an
additional rotating flux wave. The wave travels at the speed where Ip
is the current in one phase (A) and VT is the phase-
of the rotor (synchronously) and generally opposes that of the to-phase
voltage (kV).
rotor flux at some angle depending on the operating phase
angle of voltage and current at the terminals of the generator.
EXCITATION SOURCE
Thus, to maintain a regulated voltage at the machine termi-
nals, the dc field current in the rotor windings (also called
There
are several common approaches for providing dc cur-
excitation current) must be adjusted depending on the stator
rent to
the generator rotor. One is to transfer dc current from
current and voltage magnitude and the phase angle differ-
a
stationary source to the rotor via stationary carbon brushes
ence. A common parameter describing this voltage and cur-
riding
on rotating ‘‘collector’’ or ‘‘slip rings.’’ Another means is
rent phase angle is called power factor (PF), which is
with a
‘‘brushless excitation system,’’ which eliminates the
PF = cos θ need
for brushes or slip rings. Excitation systems are dis-
cussed
in a later section.
where is the phase difference between stator voltage and
current.
CONSTRUCTION FEATURES
The real power output of a three-phase electric generator
in watts is given by
A
typical hydrogen-cooled generator axial cross section is il-
Power Output = 3IP ×VP × PF = 3IPVP cos θ
lustrated in Fig. 4. The rotor is made of a solid high-strength
magnetic
forging with slots machined to contain the field
winding.
When thestator phase currents are balanced, a
(a)
measurement on the rotor would observe no time-varying
magnetic
flux because the rotor and flux rotate together. The
A B stator
core is made of laminated ‘‘punchings’’ typically about
0.5 mm
thick that are insulated from each other. The stator
core
carries a time-varying flux that creates the voltages and
currents
in the stator winding. The electric conduction path
C in the
stator core is broken by laminating the core steel to
minimize
eddy currents in the stator core, which only gener-
(b) ate heat
and losses. The rotor and stator windings are usually
made of
copper, but aluminum is occasionally used. Bearings
B and
bearing supports required at each end of the generator
need
bearing lubricating oil, usually from the same source as
A C
for the
turbine bearings. For generators that use hydrogen
cooling,
a hydrogen-tight frame is required and hydrogen
gland
seals at each end of the generator where the rotor shaft
Figure 3. (a) ‘‘Y’’ connected generator. (b) ‘‘Delta’’ connected gen- must
extend through the frame to minimize hydrogen leak-
erator. age.
The stator core must be mounted to the frame and in
TURBOGENERATORS 597
turn to a foundation. For large generators, the core is often slots.
One of the most substantial loads imposed on the rotor
spring-mounted to the frame to isolate core vibration induced is the
force needed to retain the windings in the slots at op-
by the rotating flux field from the frame and foundation. Ro- erating
speed. This is accomplished by placing a metal wedge
tor shaft lateral and torsional dynamics must be carefully cal- at the
top of the rotor winding in each slot. Loads from the
culated to avoid resonant modes at operating speed.
windings and the wedge are transferred to the ‘‘tooth’’ of the
r 1500 the
core to assume an oval shape. As the rotor turns, the oval
o
t Direct-cooled with hydrogen,
a shape
of the core moves to align itself with the magnetic poles
r water-cooled stator, 2-pole
e
n 1000 of the
rotor body. This rotating oval induces vibrations in the
e
g Direct-cooled with
frame
at the attachment to the core.
m hydrogen, 2-pole
u 500 If
the vibrations are significant, they could damage the
m
i Air-cooled, 2-pole
foundation supporting the generator and cause metal fatigue
x
a
M 0 of the
frame support. The noise from uncontrolled vibrations
0 1 2 3 4 5 6
2 can
also become excessive. Small generators directly connect
Per unit active core volume, D L
the
core to the frame because the magnitude of the core defor-
Figure 5. Generator maximum megavoltamperes versus active vol- mation
is small. Large generators with high power densities
ume for different types of cooling. must
provide flexible mounting for the core so that vibration
598 TURBOGENERATORS
from the core is not transferred to the frame and the foun- ducted
through the core steel to the vents where the heat is
dation.
transferred to the gas. The frame is designed to direct the
Some frame designs include the bearings that support the
circulation of the cooling gas, and it is cooled by direct contact
rotor. Then the weight of the rotor must be carried by the with the
gas.
frame and transferred to the foundation. Other generator de- The
windings may be cooled by one of three methods: con-
signs have pedestal bearings mounted outside the generator ventional
cooling, direct cooling, or liquid cooling.
frame. The pedestals support the rotor and transfer the loads In
conventional cooling the heat generated in the copper is
to the foundation, allowing a smaller frame. conducted
through the insulation that surrounds the wind-
Hydrogen-cooled generators use pressurized hydrogen gas ings
into the core where it is conducted through the steel to
to remove the heat generated in the copper conductors and the
vents. This is the simplest cooling method, and it has been
the core. The frame must be designed as a pressure vessel to used
since the beginning of the electrical age. Heat conduc-
retain the hydrogen within the generator and prevent its es- tion
through insulating materials is not very efficient, so this
cape to the atmosphere. Because hydrogen and oxygen com- method
of cooling seriously limits the power density of the
bine to form explosive mixtures, the frame must be designed
generator.
to contain any explosion that might occur. The explosion pres- Direct
cooling is achieved by providing cooling passages
sure may be many times the hydrogen pressure, so large gen- within
the windings. This is normally accomplished by em-
erator frames are typically made from steel plate several bedding
nonmagnetic metal tubes within the winding. The
inches thick. The frame of hydrogen-cooled generators must
circulating gas is forced through the cooling tubes. Conduc-
also support coolers to remove the heat picked up by the hy- tors
within the windings are electrically isolated from each
drogen gas. The coolers are usually finned-tube heat ex- other
and the cooling tubes by a thin covering of insulation.
changers. Hydrogen gas flows across the fins mounted on The
cooling tubes are in direct contact with the thin layer of
banks of tubes through which a liquid coolant passes. The
insulation resulting in much improved heat transfer.
generator frame must support the weight of the cooler, and it Liquid
cooling is the most efficient method of removing
must also provide gas passages for the hydrogen to circulate. heat from
the copper conductors. The winding is designed so
The hydrogen gas is circulated through the generator by ei- that
approximately every third copper conductor is hollow. At
ther an external motor driven blower or a rotor-mounted fan each end
of the winding, all of the hollow conductors within a
or blower.
half-coil
(or bar) are joined together into a header. Liquid
coolant
is pumped into a manifold at one end of the generator.
Insulated
tubes connect the manifold to the header at one end
THERMAL DESIGN
of the
winding, and coolant passes through each of the hollow
TURBOGENERATORS 599
tors
Torsional vibrational problems are generally more serious
than lateral vibrational problems for two reasons. First, there •
ANSI C50.15 ANS for Rotating Electrical Machinery—
is no way to ‘‘balance’’ torsional vibrations. The magnitude of
Hydrogen-Cooled, Combustion-Gas-Turbine-Driven, Cy-
the lateral vibration is controlled by how close the critical
lindrical-Rotor Synchronous Generators
speed is to the operating speed and by the magnitude of the •
IEEE Std 115 IEEE Guide: Test Procedures for Synchro-
residual imbalance. If the magnitude of a lateral vibration is
nous Machines
too large, the system can be balanced to reduce the amplitude •
IEC 34-1 Rotating Electrical Machines Part 1: Rating
of vibration. Torsional vibrations can arise in turbogenerator
and Performance
shaft systems that have natural frequencies very near single
•
IEC 34-2 Rotating Electrical Machines Part 2: Methods
or double line frequency. This is not something that can be
600 TURBOGENERATORS
P
jX
P E · E S
P = δ
E · sin( )
X
I
a
P
max
E F ET
PT B
A
δA δ
A max T
d d
tor’s current and its rate of decay immediately after the short
The primary purpose of a turbogenerator is to deliver electric
Electrical
point A in Fig. 6. The equilibrium at point B is unstable be-
output
cause an increase in angle # causes a reduction of delivered
power. The generator’s operation is more stable the smaller
Figure 8. Simple cycle steam turbine drive.
TURBOGENERATORS 601
mature winding
generator’s ability to supply reactive power to the system, and •
Stray load loss: load-dependent electric and magnetic
contribute to the local area voltage support.
loss caused by load in various members of the machine
are
specified by standards. Development testing is much more
Fuel
detailed and is usually performed only on the prototype for a
Aux.
Clutch Static-start power new
design or if a substantially new design feature is intro-
power and
duced. The types of tests and measured variables depend on
controller the
type of development that is being investigated.
drive mechanism and electric load. Such tests are done only duces power
in a steam turbine. This configuration is referred
in the power plant after the generator is installed. Because a to as a
combined-cycle plant. Normally, the CT and the ST
power plant environment poses substantial limitations on drive their
own independent generators. If the CT and the ST
such testing, most manufacturers test generators in their own of the
combined-cycle plant are attached to opposite ends of
test facilities, using testing methods that validate the perfor- the same
generator, the configuration is called a single-shaft,
mance of the generator. These test facilities require a drive combined-
cycle arrangement.
mechanism sufficient to bring the generator to the required A few
fundamental issues must be addressed when select-
speed and compensate for losses. These tests are the steady- ing an
electric generator to match a particular prime mover.
state, open-circuit test; steady-state, short-circuit test; and First, the
generator’s real power output rating must meet or
sudden short-circuit test. exceed that
of the prime mover(s). Additionally, the generator
In a steady-state, open-circuit test the generator is driven must
provide reactive power to the power system as needed,
at constant rated speed without load (terminals open-cir- which is
usually specified as a power factor. The power output
cuited). Several consecutive values of excitation are applied, of many
combustion turbines varies according to their inlet
and the resulting terminal voltage is measured. The shaft in- air
density, and thus is higher on cold days and lower on hot
put power and excitation power are also measured. This test days. One
way to assure that the generator capability meets
determines the open-circuit saturation curve (correlation be- or exceeds
that of the CT is to map both CT output and gener-
tween induced voltage and applied excitation current) that ator
capability against ambient air temperature and com-
provides information for calculating the excitation required at pare them.
load and, to a significant extent, validates the magnetic cir- A second
important consideration is to determine the type
cuit design. The measured power supplies information for de- of
generator needed. Most synchronous generators rated over
termining the losses that are independent of the load (core 50 MVA use
water cooling, hydrogen cooling, or air cooling for
loss, friction and windage, ventilation). the stator
winding. The rotor cooling method can be specified
The steady-state, short-circuit test is performed with the
independently, but because water-cooled rotors are relatively
generator’s terminals short-circuited (zero voltage) and the rare, most
machines that use water cooling for the stator
field excited until a desired armature current is established. winding use
hydrogen cooling for the rotor. Consequently, a
The rotor is driven at rated speed. Excitation current, arma- ‘‘water-
cooled’’ generator, as the term is often used, generally
ture current, drive power, and excitation power are measured. describes a
generator with a water-cooled stator winding and
This is repeated at several current values. This test supplies hydrogen
cooling for the rotor winding, core, and other active
information about load-dependent losses, mainly the stray parts.
Water- and hydrogen-cooled generators have the ad-
load loss. vantages of
being more efficient and less susceptible to envi-
Both open-circuit and short-circuit, steady-state tests are ronmental
contamination. They are also capable of higher
usually performed at a few selected voltage or current levels, power
densities. Their disadvantages are cost, complexity,
respectively, for as long as it takes to reach thermal steady and
maintenance. Because of advances in technology, today’s
state. Then the temperatures reached by various machine air cooled
generators obtain relatively high power densities,
parts are recorded. This supplies information about the ade- making them
a very inexpensive and attractive alternative to
quacy of the thermal design. water- and
hydrogen-cooled generators for sizes under 150
The adequacy of the ventilation design, including pres- MW to 200
MW.
sures at various key points in the generator and gas flows
Generator capability is defined by the output achieved
through the machine’s components, is determined in a sepa- while
keeping temperatures inside the generator within lim-
rate ventilation test when the machine is run unexcited at iting
values listed in industry standards. Therefore the ‘best’
constant speed while the measurements are taken. generator
is defined as the least expensive generator that
Sudden short-circuit tests are performed to measure the meets or
exceeds its design requirements. With that in mind,
machine’s transient parameters (transient and subtransient the
following concepts can be utilized to maximize a genera-
reactances and time constants), and to verify its ability to tor’s
capability:
withstand such events without structural or other damage, as
specified by standards. 1. Water
provided to the generator for cooling should be
Before shipping, testing is concluded with a thorough in- as
cool as possible. This allows the lowest generator
spection and final dielectric tests of the stator and rotor wind-
temperatures and the highest generator capability.
ings, as required by standards.
2.
Electric power factor requirements for the generator
also
contribute to the generator size. As the rated power
TURBOGENERATOR APPLICATIONS
factor decreases for a given real power output, the re-
TURBOGENERATORS 603
denser operation
Optimizing a generator for efficiency is very important. For
example, an efficiency improvement of only 0.01% on a 250
Before discussing abnormal operating conditions, it is nec-
MW generator results in an additional 25 kW available for essary
to establish what normal operating conditions are.
sale. At 5 cents per kilowatt hour, this totals roughly $10,000 Normal
operating conditions are generally recognized as
per year, assuming a 90% availability factor. Amortized at those
specified in industry standards.
10%, the resulting energy savings are roughly $100,000 in The
generator terminal voltage is directly proportional to
present worth. Additionally, each additional kilowatt repre- the flux
density inside the generator core. One of the concerns
sents capacity that normally costs in excess of $500/kW. The of
operating with a continuous overvoltage is that the genera-
corresponding savings, considering both energy and capacity tor
internal flux density exceeds design levels, possibly dam-
savings, are roughly $4500 per kW of efficiency enhancement. aging
core steel laminations. If the generator voltage is too
Before an effective efficiency study can be done, a loss eval- low and
this results in underexcited operation, core end-
uation factor in $/kW saved must be established using appro-
priate parameters for the application. This evaluation factor
is a guide in determining if changes to an existing generator
Operating
design are warranted. The costs of the changes are compared
load of
to the dollar savings achieved by the efficiency enhancement.
Generator efficiency vs. load generator
Efficiency enhancements are occasionally obtained by uti-
lizing excess generator capability if it is available. For exam-
Curve shifts to
ple, consider a generator with excess capability (defined by
right by using a
its low operating temperatures). If the generator is hydrogen-
slightly larger
cooled using 4 atm of hydrogen pressure, the generator could y
generator
c
have its hydrogen pressure reduced to 3 or 3.5 atm, as the n
e
i
c
design allows. This would raise internal temperatures be- i
f
f
cause of reduction in heat transfer capability of the hydrogen, e
Efficiency
r
change due to
o
but it would also reduce hydrogen density and decrease wind- t
a
change in static
age loss, thereby reducing generator loss. Other modifications r
Efficiency
e
H gas presure
n
2 gain by
can sometimes be effected to enhance efficiency. For example, e
a stator or rotor can sometimes be rewound with an optimized G
Changing static using a
Generator load
larly for hydrogen- and water-cooled generators, which have
relatively low fixed loss. Because the point of maximum effi- Figure
11. Variation in generator efficiency with size and gas
ciency is where fixed loss equals variable loss, a reduction in
pressure.
604 TURBOGENERATORS
Generator
rated capability:
Underexcited
Generator real
operation power
output
(reactive moves
along this line
power into
generator)
Generator maximum
(leading pf)
output at rated leading
(underexcited) power
Figure 12. Generator reactive capability
factor; typical leading
curve.
power factor is 0.95
Continous
During start-up conditions, the frequency changes rapidly
operating Frequency
as the unit is brought to operating speed. At low frequencies,
zone (per unit)
the generator terminal voltage must be kept low (or zero) to
Short-time
strained by armature reaction for corresponding increases in
operating
excitation level.
zone
Balanced three-phase current in the generator stator is
0.92
sometimes called ‘positive sequence’ current because it cre-
ates a flux wave that rotates in the same direction as the
rotor. When stator currents become unbalanced, the magni- Figure
13. Normal voltage and frequency conditions for turbogener-
tude of the current unbalance is called ‘negative sequence’ ators.
TURBOGENERATORS 605
Overexcited
operation
(reactive
power out of Operate generator as
"synchronous condenser"
generator
(lagging pf) along this line. Slight motor
action in real power axis
Generator and generator action in
reactive reactive power axis
power
output
(MVAr) Generator real output (MW)
Generator
real
Underexcited power
output is
zero
because no
operation
(reactive No prime mover
exists
power into operation
generator) region
(leading pf)
606 TURBOGENERATORS
d δ/dt = (P − P )/(2H )
m e
chines, transmission lines, transformers, etc., to reflect the
complexities of modern interconnected power systems, simpli- where #
is a measure of the angular displacement of the gen-
fied models have been developed for power system equipment, erator
rotor with respect to a continuously rotating reference
including turbogenerators. These simplified generator models frame;
Pm is the accelerating mechanical power supplied by
provide accurate simulations for specific types of power sys- the
turbine; and Pe is the restraining electric output power of
tem studies. Consequently, the model used to represent a tur- the
generator.
bogenerator depends upon what is being studied in the
During steady-state operation, the electric output power
power system. equals
the mechanical input power, and the rotor angular ac-
For typical power system applications, studies are divided
celeration (d2 2
TURBOGENERATORS 607
OPERATION OF TURBOGENERATORS
(a) (b)
generally must be unit-connected (i.e., they must be
connected to the transmission system directly via the
Figure 16. Generator installed in (a) a ‘‘breaker-and-a-half’’ and
step-up transformer, with no generator circuit breaker) (b)
‘‘ring-bus’’ configuration.
608 TURBOGENERATORS
P is the unit output (MW). The third unit has a fuel cost de- a dc
machine. The dc machine output is rectified by a multi-
scribed by L # 300 1.5P 0.15P2 . It is estimated that a
segmented commutator ring and brushes. These exciters are
unit start-up costs roughly $1500 for any of the three units,
driven by the generator shaft, either directly or through a
based on fuel costs and increased maintenance costs. The
speed reduction gear, or are driven separately by an induction
plant has a contractual obligation to supply a daily output of
motor. The exciter dc output supplies the main generator field
235 MW from 9:00 A.M. to 9:00 P.M. and 135 MW from 9:00 by a
collector assembly composed of slip rings and brushes.
P.M. to 9:00 A.M. How many units should be committed dur-
Although many dc generator-commutator exciters are still
ing each 12 h period and how should the units be loaded dur- in
service, maintenance problems and unavailability of re-
ing those periods?
placement parts make them increasingly popular candidates
The incremental operating costs of units 1 and 2 are given for
retrofitting with static exciters.
by # # dL /dP # 2 0.2P , and that for unit 3 is given by
n n n n
# # dL /dP # 1.5 0.3P . Between 9:00 A.M. and 9:00 P.M.,
3 3 3 3
Rotating Ac Exciter. Rotating ac exciters employ an ac gen-
it is required that all three units be committed to satisfy the
erator whose output is rectified and sent to the main genera-
output requirement of 235 MW. Solving the ‘‘unit dispatch tor
field. Some rotating ac exciters employ a rotor-mounted
problem’’ for equal incremental costs, #1 # #2 # #3 # $19.5/ field
winding and stationary armature. The three-phase ac
MWH when P 1 # P2 # 87.5 MW and P3 # 60 MW, so that
exciter output is rectified by one or more stationary rectifier
P 1 P2 P3 # 235 MW. Between 9:00 P.M. and 9:00 A.M.,
bridges, and the dc bridge output is sent to the generator field
only two units must be committed, but it may prove more
through brushes and slip rings. However, most rotating ac
economic to commit all three units to avoid the start-up cost.
exciters built today are brushless, so that they use rotating
Explicit evaluation of the possible alternatives (i.e., solving
rectifiers and do not require slip rings to provide generator
the ‘‘unit commitment problem’’), shows that the most eco- field
current.
nomic alternative is to commit all three units rather than
keep two units on line for 12 h, followed by a daily unit
Brushless Exciter. Figure 17 shows the arrangement of a
startup for the third unit. Then, between 9:00 P.M. and 9:00
typical brushless excitation system. The shaft-driven rotating
A.M., with all three units committed, the units should be as- ac
exciter utilizes a stationary field winding and rotating ar-
signed loadings of P 1 # P2 # 50 MW and P3 # 35 MW, on the
mature. The three-phase armature ac output is sent to a ro-
basis of equal incremental costs.
tating rectifier wheel whose dc output is sent directly to the
Obviously, a more realistic example, with unequally sized main
generator field, typically through a bore in the genera-
units, ramp rate and other operating restrictions, and provis- tor
shaft. The brushless exciter accomplishes this without
ions for complex fuel contracts, hourly power transactions,
brushes, slip rings, or commutators and thus eliminates the
and so on, would considerably complicate the problem. Com-
maintenance and carbon dust associated with them. The rec-
puter analysis is normally required for optimal economic op-
tifier wheel typically includes diodes and fuses connected in a
eration of all but the smallest power systems.
bridge rectifier configuration.
TURBOGENERATORS 609
Rotating elements
PMG Rectifier
N
CT
Ac
Generator
Exc.
S
wheel
Voltage regulator
VTs
Shunt
adjuster
Ac Dc Auto.
Firing Manual
Dc voltage
circuits control
adjuster
Man.
Figure 17. Simplified block diagram of a
which rectify the PMG output and control the level of dc volt- the
generator terminals, a means of momentarily flashing the
age to the brushless exciter field winding. The firing angle of
generator field is typically provided to initiate a buildup of
the SCRs relative to the PMG voltage, and thus the dc output
generator voltage each time the unit is started.
voltage, is determined by the voltage regulator control cir-
cuits. All excitation power in a brushless excitation system is
Compound Source Static Exciter. A compound source static
derived from shaft rotation.
exciter derives its power from both the voltage and current
Voltage Regulators
controlled by a solid-state power amplifier, which uses one
or more SCR rectifier bridges whose dc output is sent to the The
voltage regulator is the portion of the excitation system
generator field via brushes and slip rings. The voltage regula- that
provides the means of automatically controlling the out-
tor adjusts the SCR firing pulses to control the dc output. In put
of the exciter. In an excitation system with a rotating
a static excitation system that derives excitation power from
exciter, the voltage regulator controls the voltage to the ex-
Rotating elements
CT
Generator
Collector
Power
amplifier
PTT
VTs
Dc Ac
Auto.
Firing Automatic control,
circuits limiting, and
protective elements
Ac voltage
adjuster
Field Man.
flashing Manual
Dc voltage Figure 18. Simplified block diagram of a
circuit control
adjuster typical potential source static excitation
system.
610 TURBOGENERATORS
TURBOGENERATORS 611
• Overexcitation protection, which protects the generator L. K.
Kirchmayer, Economic Operation of Power Systems, New York:
field winding from overheating due to excessive excita-
Wiley, 1958.
tion current. G.
Kron, Equivalent Circuits of Electric Machinery, New York: Wi-
ley,
1951.
• Volts-per-hertz protection, which protects the core of the
generator (and connected transformers) from the heating M.
Liwschitz-Garik and C. Whipple, Alternating Current Machines,
New
York: Van Nostrand, 1961.
effects due to excessive magnetic flux.
L.
Matsch, Electromagnetic and Electromechanical Machines, Intext,
• Loss of excitation protection, which senses the condition
Scranton, 1972.
in which generator synchronism is jeopardized because G. R.
Slemon, Magnetoelectric Devices: Transducers, Transformers,
of a loss or severe reduction in excitation level. and
Machines, New York: Wiley, 1966.
• Loss of synchronism protection, in which a generator that W. D.
Stevenson, Elements of Power System Analysis, New York:
loses synchronism with the system is tripped off-line be-
McGraw-Hill, 1982.
fore damaging pole-slipping takes place. F. S.
Tse, I. E. Morse, and R. T. Hinkle, Mechanical Vibrations, Bos-
• Motoring protection, in which negative real power is ton:
Allyn and Bacon, 1968.
sensed which would indicate the loss of turbine output D. C.
White and H. H. Woodson, Electromechanical Energy Conver-
ROSS GUTTROMSON
THEORY.
T. Baumeister and L. S. Marks, Standard Handbook for Mechanical
Engineers, New York: McGraw-Hill, 1967. TV,
INTERACTIVE. See INTERACTIVE VIDEO.
TWIN
LEAD. See ANTENNA ACCESSORIES.
A. F. D’Souza and V. K. Garg, Advanced Dynamics, Englewood Cliffs,
TWO-
DIMENSIONAL CODES. See MARK SCANNING
NJ: Prentice-Hall, 1984.
A. E. Fitzgerald, C. Kingsley, and A. Kusko, Electric Machinery, New
EQUIPMENT.
York: McGraw-Hill, 1971. TWO-
DIMENSIONAL GRAPHICS TRANSFORMA-
C. M. Harris and C. E. Crede, Shock & Vibration Handbook, New
TIONS. See GRAPHICS TRANSFORMATIONS IN 2-D.
York: McGraw-Hill, 1976. TWO-PORT
PARAMETERS. See NETWORK PARAMETERS.
E. W. Kimbark, Power System Stability: Synchronous Machines, New TWO-
PORTS. See MULTIPOLE AND MULTIPORT ANALYSIS.
York: Wiley, 1956. TWT.
See TRAVELING WAVE TUBES.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3046.htm
} { { } }
HOME
ABOUT US
CONTACT US
●
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Waste-to-Energy Power Plants
● Email this page Enter words or
Standard Article
Abstract
The Opportunities
Conclusion
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3046.htm17.06.2008 16:49:40
States, not just for those connected with animal raising) (4).
Crop
wastes are about 8 kg/p/d. Mining wastes in the United
portunities.
tricity worldwide in 2010 (6), and the used oil (about 4.5 bil-
lion
liters in the United States in 1990) and old tires from
motor
vehicles (around 125 million per year in the United
States).
Table
1. Organic Wastes Produced in the United States
in
1971a
Readily
Manure 200 26
Urban
refuse 130 71
WASTE-TO-ENERGY POWER PLANTS
Logging and wood manufacturing residues 55 5
Miscellaneous 50 5
Nature itself produces wastes such as dead branches, leaves,
Industrial organic wastes 44 5
and sun-dried grasses. Primitive societies, once they had
Manicipal sewage solids 12 1.5
learned to master fire, used the energy in these wastes for a The
total has a potential energy equivalent to 175 billion liters of oil per year.
cooking and heating. These wastes can be considered the b Dry
ash-free basis.
however, both within large countries and from one country to city
of Cambridge, Massachusetts were officially given as 50
another. In the United States and elsewhere, the climate has cents
per tonne. In the 1990s the disposal costs (additional to
a large influence, with the municipal collections in the south
collection costs) for this urban region of the country were in
and west of the country being dominated by palm fronds at the
region of $100 per tonne.] We can afford to do much bet-
certain seasons of the year, for instance. In poorer countries ter.
The second concern has more validity: the as-yet unsolved
such as Lebanon, urban wastes have less paper and more food problem
of groundwater pollution. Modern solid wastes, even
wastes (generally disposed of in sink grinders in richer coun- those
collected from homes, are extremely heterogeneous, in-
tries), so that the moisture content is higher and the overall
corporating used batteries, containers with paints, varnishes,
quantity per person is lower (7).
cleaning solutions, pesticides, fungicides, unused medical
drugs,
and many other potentially toxic materials. Industrial
wastes
can include all types of noxious solids and liquids.
THE PROBLEMS OF SOLID-WASTE Rain
percolates through refuse dumps, dissolving some solids
PRODUCTION AND TREATMENT and
combining with some of the liquids to produce what is
known
as leachate. If this passes directly into an aquifer,
It is generally conceded that as countries move to what is
drinking-water wells in the vicinity can be rendered hazard-
generally regarded as the western model, the production of ous
within a short period of time. If there is no nearby aqui-
consumer goods, packaged food and drink, newspapers and fer, a
leachate ‘‘plume’’ spreads out underground and moves
magazines, and everything associated with automobile trans- slowly,
perhaps only a few meters per year. Soil scientists
portation increases very greatly; that the resulting solid-
estimate that in some cases these noxious plumes may not
waste flow increases enormously in quantity and variety, in- pollute
an aquifer for thousands of years. However, in most
cluding a large number of noxious and toxic materials; that a cases
the pollutants will begin to appear in regional water
large proportion of citizens are careless in disposing of these within
tens or perhaps hundreds of years. Many efforts are
wastes; and that the authorities responsible for collection and being
made to devise methods by which such plumes can be
disposal of wastes do not always use the highest standards
contained or diverted, but none has yet proven effective.
that could protect future generations from harm. The follow-
Rules have therefore been established for the safe land-
ing aspects are less-publicized aspects of the solid-waste filling
of solid wastes (so that a landfill differs from a dump
problem. in that
such rules are nominally followed). The pits are lined
with
waterproof clay or a membrane of urethane or other
Are We in Danger of Burying Ourselves in Trash? long-
lasting and supposedly impenetrable material. The lea-
We are not, in the overall sense, burying ourselves in trash. A chate
is collected at the lowest point of the liner and taken to
large proportion of the materials we use comes from quarries, a
treatment facility. However, public concern over alleged
mines, and the like, and we leave most of these sites unfilled. (and
often confirmed) serious health effects from polluted
A ‘‘satellite’s eye’s view’’ of the material-transportation net-
drinking water and perhaps from gaseous emissions from old
work of any country would show trains, barges, and road ve- dumps,
as well as public confidence that landfill safeguards
hicles taking materials from mines and quarries principally will
operate perfectly for hundreds of years, is so low that
into towns to be used in buildings and roads and industry. A it is
very difficult to get acceptance for the establishment of
new
landfills.
large proportion of these vehicles would be empty on the re-
turn trip. [Cities in the past did, however, frequently get bur-
ied in their wastes. Some medieval German cities avoided this
Concerns Over the Incineration of Solid Wastes
danger by requiring that wagons that had brought in produce
The
open burning of refuse was, in the distant past, accept-
take out wastes to be deposited in the countryside (1)]. A proj-
able in
rural areas but prohibited in towns. Controlled burn-
ect investigated in the early 1970s by the New York Central
ing in
an enclosed furnace with a high stack to disperse the
Railroad to transport compacted wastes from New York City
smoke
seemed to be an advanced solution for towns when the
to a large quarry near a small town in Ohio failed because
first
so-called incinerator was built in England in 1874. Incin-
the town residents, who had anticipated enduring about five
eration
was not practiced in the United States in the period
years of noise and dust in return for a golf course and other
up to
the 1970s to the extent that it was followed in Europe
recreation areas created as part of the agreement, discovered
and
elsewhere. However, in the late 1960s a ’’solid-waste cri-
that it would take 175 years at then-current rates of produc-
sis’’
was declared in the United States, and a period of intense
tion of New York solid wastes to fill the quarry. Even the
use of additional fuel oil to reheat the gases before discharge) companies
can indeed charge a ‘‘tipping fee’’ for the accep-
are used, a potentially noxious sludge must be disposed of. tance of
wastes; the greater the degree of pretreatment, the
lower the
fee.)
The
final degree of pretreatment is to convert the wastes
THE OPPORTUNITIES into a
liquid or gaseous fuel of desirable characteristics. Doing
this has
two beneficial side effects. One is that the treatment
Problems lead to opportunities. The opportunities are particu- process
is in effect a chemical plant, out of which there is
larly attractive in the waste-to-energy area, although not in tight
control of all effluxes. The other is that the clean fuel so
the relatively unsophisticated incineration plants of the past. produced
can be burned in processes that have much higher
These involved, in general, so-called ‘‘mass burning’’ of un- potential
thermal efficiencies, and possibly much lower capi-
treated or minimally separated wastes and had low thermal tal costs,
than incinerators. These improved processes will be
efficiencies (expressed as the output of useful heat or useful discussed
below, after a review of the baseline mass-burning
power divided by the calorific value of the wastes). Low effi- waste-to-
energy plant.
ciencies imply high-temperature high-volume discharge of ex-
haust gases, difficult and expensive to treat. The attainment
ALTERNATIVE METHODS OF RECOVERING
of high efficiency in itself results, therefore, in a reduction of ENERGY
FROM WASTES
pollution, in that the exhaust flow and temperature are re-
duced. It is easier to incorporate exhaust-cleaning systems. Mass-
Burning Incinerators Raising Steam
Nitrogen oxide emissions will probably be greatly reduced. Di-
Incineration may be carried out without any attempt to re-
oxin emissions are also likely to be reduced. cover
energy from the combustion heat. Energy recovery in
To reduce emissions from incinerators, better control sys-
association with incineration should be welcomed, although
tems are needed. Even the most highly sophisticated control the power
generated from the wastes must perhaps be looked
system cannot greatly improve the combustion process in upon only
as a byproduct. Perhaps surprisingly, the costs of
mass-burning in which, for example, a piece of dry tissue pa-
incineration with and without energy recovery are similar in
per may be close to a stack of water-soaked telephone books. many cases
examined (see below) because cooling of the gases
One is consumed in a fraction of a second, while the other reduces
the cost of the gas-cleaning equipment required and
dries and smolders for hours. Therefore there is also a need thus
compensates for the added costs of heat-recovery
for better fuels. equipment.
7000
combustion of paint and pesticide cans and the like. In the
5%
late 1980s and 1990s there has been a revival of the picking
15 ) 10%
belt, along with research and development work to automate )
m = Typical
g
b 6000
k
l solid wastes
it, to produce a more homogeneous combustible product.
This / / 15%
J
u
operation is called full-stream processing (3). M
t
( B
( 20%
A further improvement in the fuel is to use hammer mill e
e
u
5000
l
u
or other types of shredders to produce a more uniform fuel a
l
v
a 25%
v
from a stream out of which undesirable components have g 10
g Beirut
n
US
i
n
t
i
been separated. (Serious explosions have occurred in solid- a
t 4000
e
a
waste hammer mills into which partly full cans of gasoline h
e
h
e
and live ammunition, for instance, have been fed, so that sort- s
e
u
s
f
u
ing of the input is necessary.) The shredded waste, mostly e
f 3000 Noncombustibles
R
e
paper and plastic, can then be fed (normally after warming
R percentage = 30%
and moisturizing) to a briquetting or pelletizing unit. The 5
product is called refuse-derived fuel (RDF) or sometimes den-
2000
0 10 20 30 40 50 60 70
sified RDG (d-RDF) (8); it can be stored, transported, fed and
burned in the same way as coal. If the presorting is done ef-
Refuse moisture content (%)
fectively, it has much lower emissions than coal. It has been Figure 1.
Refuse heat values versus moisture content and percent-
frequently co-fired with coal in utility steam generators with age of
noncombustibles. (From Ref. 7.) Moisture content depends on
the purpose not only of reducing the emissions to below some presence
of food wastes, on climate, and on collection practices, and
regulatory limit but of reducing the cost of the fuel. (Power has a
large influence on heating value.
)
2400 $
S 350
U
Moisture /
n
2000 content o
i 300
l
l
= 50% i
m
) (
D 40% t
P 1600 s 250
T o
( c
30% l
y a
t 20%
i t 200
c i
a 1200 p
p 10% a
c
a
c ) 150
d
t e
n 800 t
a c
l e
P r 100
r
o
c
400 3
9 50
9
1
(
0 t
n 0
0 12 14 16 18 20 22 24 26 28 a
l
0 500 1000 1500 2000 2500 3000
P
Plant overall efficiency (%)
Plant capacity (TPD)
nomical,
the capacity should be at least 500 TPD. Figure 5
34 120
32
Coal 100
30
) 80
% 28 RDF )
Tipping fee
$
(
60
y S
c 26 U
n ( 40
e
i 24 Raw refuse e
c s
i 20
f u
Electrical-R
f 22 f
e e
Material-R
r
l
l 0
a 20 n
Amortization-C
r o
e t
–20
v 18 r
o e
t p –40
O & M-C
n 16 t
a s
l o –60
P 14 Raw refuse C
RDF –80
12
Coal
–100
10
0 50 100 150 200 250 300 –120
Plant gross power output (MW)
0 500 1000 1500 2000 2500
Cooling water
Intercooler
Inlet air
Power
Compressor Compressor
Turbine
Natural
gas
or
petroleum
Scrubber
Refuse
Burner
or RDF
Mechanical
link
Refuse
Hot-gas
burner
cleanup
after a scrubber.
since the 1950s, and from at least the early 1980s large land-
fills have been capped and drilled to supply fuel gas, largely
Compressor
have been strongly advocated, but have not been successfully
Gas combustor
put into practice to the end of the 1990s. Pyrolysis or starved-
is
particularly suitable to the SFEHGT cycle discussed above.
Modular regenerator
The two other processes are acid and enzymatic hydrolysis
of
organic wastes to produce ethanol. Lynd et al. (21) point
Figure 7. Conceptual sketch of a gas turbine plant. (From Ref. 7.)
out that the technology for ethanol production from cellulosic
An attempt has been made to show the size of the RDF combustor
materials is fundamentally different from that for production
(which could be an incinerator) in relation to those of other compo-
for food crops. (Fuel ethanol is successfully produced by fer-
nents.
mentation, in Brazil from sugar cane and in the United States
from corn and other starch-rich grains.) Lynd et al. state that
bines the output and efficiency are increased through the use
1990 production cost of ethanol is similar for the two forms of of
a ‘‘reheat’’ combustor. This increases the gas temperature
hydrolysis. However, the enzymatic process was at an earlier
after the gas has expanded through the first turbine stage.
stage of development and was likely to be the more cost-effec-
There is more heat in the exhaust, and therefore more power
tive after further research. Neither had favorable conversion is
delivered by a turbine running on the steam generated
economics in 1990. The authors point out that there is a con-
from this heat.
siderable quantity of solid material left after hydrolysis that,
There is sufficient oxygen in the exhaust of a simple-cycle
after dewatering, could be used as a solid fuel. gas
turbine to support additional combustion. However, most
Stack
Turbine- Shaft-power
Compressor
expander load
Hot
Air
Combustor- Process
exhaust
burner or thermal
load
Figure 8. Combined heat and power (CHP) plant.
(From Ref. 23.) As shown, this is a plant for ‘‘clean’’
fuel, oil or gas. The hot exhaust can raise steam for
process plants or for district heating.
Fuel
Stack
Turbine-
Shaft-power
Compressor
expander
load (1)
Air Combustor-
Heat-
burner
recovery
steam
Shaft-power
Hot
generator Steam- load (2)
Fuel exhaust
turbine
Condenser
Steam-Turbine
Boiler
"Bottoming" Cycle
feed pump
The prices forecast for the plants are 75% of those for the than
for the combined cycle because virtually all the water is
IGCC plants.
normally lost in the exhaust rather than being circulated
in
a closed system and because the specifications are more
Steam-Injection Gas Turbines. Steam injection in a location
stringent. Any dissolved solids that become deposited on the
where it will expand through the turbine blading with the
turbine blades or elsewhere could form corrosion sites or po-
combustion gases is a third use (besides expansion in a steam
tential blockages. However, Tuzson states that water-purifi-
turbine and, in advanced plants, cooling of the turbine blades)
cation cost is of the order of 5% of the fuel cost and is not,
of the steam generated in a HRSG. Steam may be injected
therefore, a decisive factor. The reliability of early steam-
upstream of or into the combustion chamber, or into the tur-
injected units has been high—for instance, 99.5%. Rather
bine nozzles anywhere along the expansion. The steam does
surprisingly, combustor-liner durability has been found to
less work the further along the expansion it is injected. In
increase.
comparison with the combined cycle, the steam-injected cycle
One of the advanced gas-turbine systems being developed
has the following advantages. A substantial increase in power in
Japan uses an intercooled-reheated gas turbine (the in-
can be obtained from the gas-turbine engine with no modifi-
tercooler is a water-spray direct-contact type) in which the
cation in the configuration of the expansion turbine itself. The
steam raised in the HRSG can power a conventional steam
part-load efficiency is improved. The production of NOx is re-
turbine, or the steam can be injected into the gas turbine (26).
duced. In a review of the status of steam-injected gas tur- The
output, 400 MW, and the predicted efficiency, 54.3%,
bines, Tuzson (25) stated that combined-cycle turbines have
place it outside Tuzson’s guidelines above.
demonstrated the highest power-generation efficiencies and A
gas turbine is a good candidate for steam injection if the
the lowest cost in sizes above 50 MW (although he also quotes
compressor has a wide range of operation because the in-
a study giving the power level below which steam-injection
creased flow creates a higher back pressure. A high pressure
systems become more attractive than combined cycles as 150
ratio and a high turbine-inlet temperature are also desirable.
MW). At lower power levels the steam-injected gas turbine
These conditions seem to favor the aircraft-derivative turbine.
becomes attractive because of the avoidance of the large cost
However, Tuzson points out that heavy-duty industrial tur-
of the steam turbine. A typical power gain from steam injec-
bines can accommodate concentrations of contaminants about
tion for a GE LM5000 gas-turbine engine was quoted as in- five
times higher than can the aircraft-derivative turbines.
creasing the engine output for 34 MW to 49 MW, together
There are many variations of these relatively simple forms
with an efficiency increase from 37% (simple cycle) to 41%. of
water/steam injection. El-Masri (27) proposed an in-
GE analyzed the gains that would be obtained from a combi-
tercooled-recuperative cycle in which the intercooler and an
nation of intercooling and steam injection for its LM5000 gas-
aftercooler are direct-contact water-injected evaporative units
turbine engine: a power increase from 34 MW to 110 MW and and
there is subsequent water injection into the recuperator
an efficiency improvement from 37% to 55%. The water-puri-
(Fig. 10). There is no steam generator. The results of his anal-
fication requirements are more demanding for steam injection ysis
show considerably higher efficiencies over the conven-
Stack
5
Recuperative heat exchanger
Water
Air
2B
2
a
Water Combustor- Fuel
Combustor- 4A
1 burner
burner
4
CD
3
High-pressure
High-pressure
Low-pressure compressor
turbine Low-pressure Power
turbine
compressor
Cooling flows
2A 1A
Water Evaporative
intercooler
Combustor-
burner
Fuel
Low-pressure High-pressure
Turbine-
compressor compressor
expander
Recuperator
Air
Humid air
Warm
water
Warm water
Mixer
Intercooler
Aftercooler
Economizer
Water
Pump
drain
Water
81–100, 1995.
is melted during power peaks for four hours per day, five days
per week. An increase of hot-weather capacity of 29% was
13. L. Lamarre, Tapping the tire pile, EPRI J. 20 (5): 28–34, 1995.
given for the plant quoted. The modification paid for itself in
14. H. E. McCarthy, R. L. Clayson, and G. R. Short, Solid waste/
under a year.
oil shale combustion—an environmentally benign plant, IECEC
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3049.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Wind Power
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3049.htm17.06.2008 16:50:07
ing
offshore in order to get the maximum power from the
wind.
Offshore the landscape is flat with high wind speeds.
POWER
IN THE WIND
When
the wind blows, kinetic energy is stored in it. This en-
ergy
is transformed to mechanical energy by use of a wind
Pv = 1 Av 3ρCp (1)
2
A = π(d/2)2 (2)
where
Pv is the mechanical shaft power (W), A is the total
2
area
swept by the wind turbine (m ), d is the turbine diameter
(m),
v is the wind speed (m/s), # is the mass density of air
POWER
IN A WIND TURBINE
WIND POWER
The
only unknown parameter in Eq. (1) is the power coeffi-
Energy comes from various sources. The major sources of en- cient
Cp for the turbine. This coefficient is a characteristic of
ergy for human use are fossil fuels such as oil, coal, and natu- a
specific wind turbine construction, and it is normally shown
ral gas. Another important contributor is nuclear energy. All as a
function of the ratio # of the tip speed to the wind speed:
these sources produce environmental pollution. Alternatives
to them are renewable sources such as solar, ocean waves,
λ = (d/2)ωr (3)
and wind energy. In some areas of the world, wind can be an
v
important contributor to electricity production. It gives elec-
trical power without any emission or pollution. The energy where
#r is the rotor speed and # is the tip–wind-speed ratio.
from the wind is normally extracted by a wind turbine, which
The tip speed will typically be fixed if an induction genera-
converts the kinetic power in the wind to mechanical power, tor
is used and connected directly to the grid. Measured Cp
which in many applications again is converted into electrical
characteristics for two windmills are shown in Fig. 2. The
power by use of an electrical generator. Figure 1 shows photo-
measured curves show that Cp has an experimental maximum
graphs of a wind turbine and a whole wind turbine park. of
0.48–0.5. It can be shown theoretically that the absolute
There are two basic wind power systems. In a standalone
maximum Cp is 0.59. The Cp curve is used for power control
system where wind is a dominant source of generation, a of a
windmill and is an important consideration when a wind
backup source based on diesel engines is needed because the rotor
is designed aerodynamically. The Cp curve also shows
wind is not constant. In the other system, the wind-electric that
the maximum Cp can be obtained at only one wind speed
generator directly feeds power into a well-established electri- if
the blade tip speed is fixed.
cal power grid. Here also, if wind-electric is a significant por-
tion of the overall generation, fluctuations in wind speeds POWER
CONTROL METHODS
cause undesirable power system oscillations. To overcome
this, energy storage has to be available on a large scale, which Stall
Regulation
is usually not economical. Batteries, superconducting induc-
tors, and fuel cells are obvious means, but in general, storage The
most commonly used power control method at higher
systems reduce the overall efficiency and cause the price to wind
speeds is stall regulation, where the rotor profile of the
be too high. Another solution, to keep wind-electric generation wind
turbine controls the mechanical power and limits it. The
fairly constant, requires reducing the power production of the
position of the rotor profile is always fixed in relation to the
wind turbine system by control at high wind speeds. However,
windmill, and since the rotor speed is almost fixed (varies
this solution is not beneficial for the economy of the wind only
1% to 2% due to the slip of the induction generator), the
power system. wind
speed across the rotor profile will increase, and if the
In many areas wind power will be a fractional supplement speed
exceeds a certain limit, turbulence (stall) will result.
(
10%) to future energy production because of the necessity Then
power is lost and the Cp value decreases. If the wind
of a backbone power source. In some special areas where the
turbine is properly designed, stall regulation along the whole
energy transportation cost is high, the wind power can be a wind
turbine profile can be used to control the maximum
significant portion. Typical wind power systems are today in shaft
power Pmax according to the equation
the power range of 400–600 kW, but new commercial 1.5 MW
(a)
(b)
variations.
cal power is obtained and thereby also a fixed electrical
power. This also reduces the mechanical stress in the system
compared to stall regulation. A second effect is that at lower
SIMPLE WIND POWER SYSTEMS
wind velocities the highest possible Cp value can be obtained
all the time. For
low power (
1 MW) the cheapest generator on the market
This strategy gives a more flexible control system, but it is
widely used: the three-phase squirrel-cage induction gener-
also requires a more complex system to control the pitch. This ator
(IG), which can be connected directly to the grid without
is traditionally done by a hydraulic system in the windmill any
special power and control electronics. For higher power
but some systems are also electrically based. A basic control other
systems can be preferable because of the requirement to
system is shown in Fig. 4, where a control unit adjusts the cool
the rotor. The only power electronics needed are thyristor
pitch.
silicon-controlled rectifiers (SCRs) for limiting the inrush cur-
0.8 800
700
0.6 600
) 500
W
k
(
C r 400
p 0.4 e
w
o
P 300
200
0.2
100
0
0.0
0 5 10 15 20 25
0 5 10 15 20 25
Wind speed (m/s)
0.8
Induction
generator
0.6
(4-pole)
ω ω
Wind
r g
Gearbox Grid
C
p 0.4
Pitch Control
control unit
0.2
Wind
0.0
Wind
0 5 10 15 20 25
speed
700
rent during startup. One alternative is the three-phase syn-
chronous generator (SG), but this generator is very stiff, and 600
without any special damping of wind power fluctuations,
which can give problems. The SG is also more expensive than ) 500
W
the IG. The most cost-optimal solution at present is therefore k
(
r 400
a standard four-pole squirrel-cage induction generator with a e
w
gearbox. Multipole IGs and SGs can also be used. Figure 6 o
P 300
shows the systems for fixed-speed operation.
Some systems use also an induction generator with pole 200
change, so the windmill can operate at two speeds. This gives
a better utilization of the wind, but the generator is more 100
expensive. Different techniques exist to get maximum energy
from the wind at a given place in the landscape; in particular, 0
0 5 10 15 20 25
it is not always best to have the highest Cp value of the tur-
bine and windmill at maximum wind speed. It is better to
Wind speed (m/s)
maximize Cp at the wind speed that has the highest proba- Figure
5. Measured power curve for a pitch-controlled windmill:
bility. VESTAS
V39.
Induction ADVANCED
WIND POWER SYSTEMS
generator
(4 pole) As
mentioned, the maximum Cp of the wind turbine can only
Wind ω r ωg be
reached at one point in fixed-speed operation. In order to
obtain
more energy from the wind, adjustable-speed systems
Gearbox Grid
can be
used, but they require power electronics.
A
number of different topologies are possible when ad-
justable
speed is used. The most promising alternative is a
(a) forced-
commutated, voltage-source ac–dc–ac converter with
an
induction generator, which is shown in Fig. 8. Another
ω ω
Wind r g
possibility is a current-source inverter, which has poorer dy-
Synchronous
namics,
as shown in Fig. 9.
Gearbox generator Grid
(4 pole) Up to
2 MW forced-commutated voltage-source systems are
Grid the
grid.
Some
of the advantages in this system are that ideally it is
possible
with a standard 1 MW generator to get 2 MW output
(c) without
overloading the system, and the speed can be ad-
Induction
generator
ω ω
Wind r g
Gearbox
Grid
Reactive
power
control
Induction
generator
ω ω ac/dc dc/ac
Wind r g
Gearbox
Grid
generator.
Induction
generator
ω ω
Wind r g
Gearbox
Grid
induction generator.
DIG
ω ω
Wind r g
Gearbox
Grid
DIG
ω ω
Wind r g
Gearbox
Grid
ac/dc
ac/dc dc/ac
Wind
ω ω
r g
Gearbox Synchronous
Grid
generator
Figure 12. Forced-commutated voltage-
nous generator.
Wind
ω ω
r g
Gearbox Synchronous
Grid
generator
synchronous generator.
ac/dc dc/ac
Wind
ω
r Multipole
synchronous Grid
generator
Figure
14. Multipole permanent-magnet synchronous generator
with
voltage-source forced-commutated converter.
408–413.
B. S. Borowy and Z. M. Salameh, Optimum photovoltaic array size
M. Yamamoto and O. Motoyoshi, Active and reactive power control
for a hybrid wind/pv system, IEEE Trans. Energy Convers., 9: 482–
for doubly-fed wound rotor induction generator, IEEE Trans.
488, 1994.
Power Electron., 6: 1991, pp. 624–629.
´
C. Brune, R. Spee, and A. K. Wallace, Experimental evaluation of a
variable-speed, doubly-fed wind-power generation system. IEEE
Conf. Record Ind. Applications Soc. Annu. Meet., 1993, p. 480–487.
´
H. L. Nakra and B. Dube, Slip power recovery induction generators
for large vertical axis wind turbines, IEEE Trans. Energy Convers.,
3: 1988, pp. 733–737.
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3050.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Wind Power Plants
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3050.htm17.06.2008 16:50:37
WIND
POWER PLANTS
However, with the first oil shock of the early 1970s there was
a
reawakening of interest in wind power; it is from this time
that
the development of modern electricity generating wind
r
For assessment of the annual energy that will be obtained
e
w from
a wind turbine it is usual to use hourly mean wind
o
P
i=n
E = #H W
i i
i= 1
puts of, say, 50 kW and rotor diameters of 15 to 17 m. These
small wind turbines were installed in large numbers in the where
E is the annual energy yield in kilowatts per hour, H
United States in California and in Denmark where particular is
the hours in wind speed bin i and W is the output power at
financial and administrative arrangements had been made to that
wind speed in kilowatts; n is the cut-out wind speed or
encourage electricity generation by wind power. Although
maximum wind speed of the probability distribution.
many of these small turbines also failed to work properly,
over the next 20 years, the development of wind turbine tech-
nology was such that by the end of 1997 some 7500 MW of
OPERATING EXPERIENCE OF WIND POWER PLANTS
wind turbine capacity had been successfully installed world-
wide, and a number of wind farms with turbines of rotor di- Wind
power plants, or wind farms as they are often called,
ameters up go 45 m and with outputs of 600 kW had given
operate as unmanned, automatic systems. The wind turbines
satisfactory service for several years. The first units of the
generate electricity automatically, provided that the wind
next generation of wind turbines with rotor diameters of up to speed
is within the operating limits for the wind turbines and
60 m and rated outputs of 1.5 MW had also been constructed. that
there are no equipment faults. Manual intervention is
There are two basic architectures of a wind turbine: (1) only
required for routine maintenance and fault correction op-
the horizontal axis or propeller type, and (2) the vertical axis,
erations.
Darrieus or ‘‘egg beater’’ type. Although there are some wind
Modern wind farms all have a supervisory, control and
farms of small vertical axis turbines in California and there data
acquisition (SCADA) system that allows an operator to
is limited continuing interest in the technology, this architec-
monitor the performance of the wind farm remotely and to
ture is not of commercial significance. Modern horizontal axis
identify any faults that require manual intervention. Each
wind turbines usually employ three blades as this is the most wind
turbine operates autonomously under the control of its
aesthetically pleasing arrangement. The rotor is located up- own
computer-based controller and the SCADA system inter-
wind of the tower to reduce torque pulsations and noise due
rogates each controller in turn to monitor its operation and
to the blades passing through the wind shadow of the tower. any
alarm conditions.
The power curve of a wind turbine (Fig. 1) shows the rela- A
typical example of the performance of a wind farm is
tionship between the wind speed some distance upsteam of shown
in Fig. 3. The upper trace shows the wind speed mea-
the rotor and the electrical output from the turbine generator. sured
by an anemometer located on a mast, at the hub height
At wind speeds below cut-in (5 m/s in this case) there is not of
the turbines, and in the middle of the site. However, as the
enough power in the wind to overcome the frictional losses of wind
farm was several kilometers in size and located in hilly
the turbine drive train and the rotor is kept parked. Between
terrain, each turbine would be subject to a different wind
cut-in and rated wind speed the output power rises very rap-
speed. The lower trace shows the output power of the plant
idly. The power available in the wind is proportional to the
measured at the point of connection with the distribution sys-
cube of the wind speed. However, with a fixed-speed wind tur- tem
of the public utility. With the measured wind speed
bine the aerodynamic efficiency of the rotor varies with wind
above, say, 15 m/s, the wind farm produced its rated output
speed and thus the power curve between cut-in and rated of
7.2 MW while when the wind speed dropped below approxi-
wind speed does not follow the cubic relationship precisely. At
mately 5 m/s no power was produced.
rated power and wind speed (600 kW and 12 m/s in this ex-
Due to the intermittent nature of the wind resource it is
ample) the power developed by the rotor is limited either by not
viable for the power output from a wind farm to be sched-
changing the pitch angle of the blades (pitch control) or by uled
or dispatched by the electricity utility responsible for op-
allowing the blades to enter aerodynamic stall (stall control).
erating the generators of the power system. Hence the power
The choice of rated power and wind speed is made by the
output from a wind power plant is exported to the local elec-
designer of the turbine. At shut-down wind speed the turbine
tricity system on a ‘‘take all’’ basis. Appropriate energy pur-
1300
1200 750
1100 700
650
1000
H 600
900
,
r W 550
a
e 800 , 500
y )
r 700 W 450
e k
p 600 ( 400
t
s u
r 500 p 350
u
t
o u 300
H 400 O
250
300
200
200
150
100 100
0 0
25
20
s
/ 15
m
10
0
01/01 02/01 03/01
04/01 05/01 06/01 07/01 08/01
Date
Main shaft
Hub
Gearbox
Generator
Yaw section
300 WIND
FARM ELECTRICAL SYSTEMS (6)
Without PFC
In
almost all cases the power collection systems within wind
farms
in Europe and North America use underground cable.
This
is to reduce visual impact of the wind farm and to avoid
the
hazards associated with the large cranes required for
r
erection of the turbines operating in the vicinity of overhead
A
lines. The voltage of the main power collection circuits de-
V
k pends
on the size of the wind farm and the voltage at the
m 1
R 1 jX 1 jX 2 R2
S
–jX c jX m R2(1–s)/s
dc Tx
Network
Carbon
dioxide 800 10,500–16,100
where C is the capacitance of the power factor correction ca-
Sulfur
dioxide 10 150–240
pacitors, Lm is the magnetizing inductance, and # is the angu- Nitrogen
oxides 3.4 50–80
lar frequency of the islanded system, which, in this simple
analysis, is assumed to be equal to the rotational speed of the
generator. In this self-excited condition the induction ma-
chine is able to sustain a terminal voltage without a connec- common
with the capacities of wind generation now in ser-
tion to the network and, in fact, as the rotational speed of vice.
Therefore, wind power has an impact on the power sys-
the turbine rises, very large transient overvoltages may be tem both
because of the nature of the wind resource but also
experienced. Many utilities are extremely concerned about because
of its location within the power system.
main system under fault conditions. Thus, most wind farms The main
impact of wind power plants on the generation sys-
will have sophisticated protection to detect this loss-of-mains tem is
the saving in fossil fuel. This translates directly into a
or islanding condition. This protection is located at the point reduction
in gaseous emissions and hence in environmental
of connection of the wind farm to the public electricity supply damage
due to acid rain and climate change or global warm-
network and often uses relays either based on detection of the ing. The
extent of this reduction in environmental impact
rate of change of system frequency or on a rapid change in clearly
depends on the plant mix of the generation system
the position of the voltage vectors. Both conditions will occur into
which the wind farms are connected. However, typical
when a wind farm is islanded. savings
for the United Kingdom generating system, which is
Obtaining a low impedance connection to the mass of earth still
heavily reliant on coal-fired plants, are shown in Table 1.
at all points in a large wind farm is difficult. This is particu- The
proportion of generation due to wind power in all global
larly the case where the wind turbines are located on rocky, major
power systems remains low (generally much less than
high-resistivity ground. A low impedance earth or ground con- 5%) and
is unlikely to rise above 10% for many years. At these
nection is required for personnel and animal safety as well as levels of
penetration, the effect of the intermittent nature on
to protect the wind farm plant from the effects of both electri- the wind
is small as any changes in wind generation are
cal faults and of lightning (8). Usual practice is to install some masked
by the much larger changes in conventional central
local earth rods and conductors at each wind turbines site, generator
loading due to variations in customer demand and
but also to lay bare copper earthing conductor with the main plant
breakdowns.
power collection cables and to bond the earths of the entire In
addition to the saving in fuel it is also possible to allo-
wind farm together. This can result in extremely long hori- cate some
capacity credit to wind power plants (9). The capac-
zontal earth electrodes of up to, say, 5 km in length. These ity
credit indicates the capacity of other forms of generation,
conductors then have a significant self-inductance and series which is
no longer required because of the connection of the
resistance which need to be considered in the design and test- wind
power plants. It is usual to calculate this using a substi-
ing of the earthing system. It is also important to be aware of tution
technique based on loss-of-load-probability (LOLP). Al-
the possibility of external earth connections being transferred though
the results of such calculations depend on the type of
into the site via telephone circuits or other services because
generating system considered, a number of studies in differ-
these can lead to large potential differences in the earthing ent
countries indicate that at low penetrations of wind power
system under fault conditions. (less
than 1%) the capacity credit of wind power plants is at
least
equal to the mean annual output while it drops to half
this
value at penetrations of 10%.
IMPACT OF WIND POWER PLANTS ON THE POWER SYSTEM
scenario would occur, and hence the impact of the wind farm
P is the real power exported by the generator and Q is the
throughout the year is determined.
reactive power absorbed by it; R and X are the real and imagi- In
a simple radial distribution circuit, wind power plants
nary components of the network source impedance. Thus it do not
provide capacity credit as the circuit is required to pro-
may be seen that the real power export from the wind power vide
reactive power to allow the wind farm to operate. In more
plant acts on the circuit resistance to raise the voltage while
complex, interconnected distribution systems, wind power
the reactive power drawn by the induction generators acts on plants
may make some contribution to the capacity of the
the circuit inductive reactance to lower the network voltage. system.
NICHOLAS JENKINS
UMIST
ANDREW VAUDIN
National Wind Power
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELECTRICA...RONICS
%20ENGINEERING/22.%20Energy%20Conversion/W3051.htm
} { { } }
HOME
ABOUT US
CONTACT US
HELP
Home / Engineering / Electrical and Electronics Engineering
Your Librarian
Electronics Engineering
● Save title to My ● Search this title
Profile
Wind Turbines
Abstract
file:///N|/000000/0WILEY%20ENCYCLOPEDIA%20OF%20ELE...%20ENGINEERING/22.%20Energy
%20Conversion/W3051.htm17.06.2008 16:51:03