Journal of Energy and Power Engineering-2012.01

Volume 6, Number 1, January 2012 (Serial Number 50)
Journal of Energy
and Power Engineering
David Publishing Company
www.davi dpubl i shi ng.org
Publishing David

Publication Information:
Journal of Energy and Power Engineering is published monthly in hard copy (ISSN 1934-8975) and online (ISSN
1934-8983) by David Publishing Company located at 9460 Telstar Ave Suite 5, EL Monte, CA 91731, USA.

Aims and Scope:
Journal of Energy and Power Engineering, a monthly professional academic journal, covers all sorts of researches on
Thermal Science, Fluid Mechanics, Energy and Environment, Power System and Automation, Power Electronic,
High Voltage and Pulse Power, Sustainable Energy as well as other energy issues.

Editorial Board Members:
Prof. Ramesh K. Agarwal (USA), Prof. Hussain H. Al-Kayiem (Malaysia), Prof. Zohrab Melikyan (Armenia), Prof.
Pinakeswar Mahanta (India), Prof. Carlos J. Renedo Estbane (Spain), Prof. Mohamed Ahmed Hassan El-Sayed
(Trinidad and Tobago), Prof. Carlos Redondo Gil (Spain), Prof. Roberto Cesar Betini (Brazil), Prof. Rosrio Calado
(Portugal), Prof. Dr. Ali Hamzeh (Germany).

Manuscripts and correspondence are invited for publication. You can submit your papers via Web Submission, or
E-mail to energy@davidpublishing.org or info@davidpublishing.org. Submission guidelines and Web Submission
system are available at http://www.davidpublishing.org.

Editorial Office:
9460 Telstar Ave Suite 5, EL Monte, CA 91731, USA
Tel: 1-323-984-7526, 1-302-597-7046
Fax: 1-323-984-7374
E-mail: energy@davidpublishing.org; info@davidpublishing.org

Copyright2012 by David Publishing Company and individual contributors. All rights reserved. David Publishing
Company holds the exclusive copyright of all the contents of this journal. In accordance with the international
convention, no part of this journal may be reproduced or transmitted by any media or publishing organs (including
various websites) without the written permission of the copyright holder. Otherwise, any conduct would be
considered as the violation of the copyright. The contents of this journal are available for any citation. However, all
the citations should be clearly indicated with the title of this journal, serial number and the name of the author.

Abstracted / Indexed in:
Database of EBSCO, Massachusetts, USA
Database of Cambridge Science Abstracts (CSA), USA
Chinese Database of CEPS, American Federal Computer Library Center (OCLC), USA
Ulrichs Periodicals Directory
Summon Serials Solutions
Chinese Scientific Journals Database, VIP Corporation, Chongqing, China
Chemical Abstracts Service (CAS)

Subscription Information:
Price:
Print $600 (per year)
Online $480 (per year)
Print and Online $800 (per year)

9460 Telstar Ave Suite 5, EL Monte, CA 91731, USA
Tel: 1-323-984-7526, 1-302-597-7046; Fax: 1-323-984-7374
E-mail: order@davidpublishing.org

www.davidpublishing.org
DAVID PUBLISHING
D

Jour nal of Ener gy and
Power Engi neeri ng

Volume 6, Number 1, January 2012 (Serial Number 50)

Contents
Clean and Sustainable Energy
1 Simulation of Fast Gas Injection Expansion Phase Experiments under Different Pressures
Donella Pellini, Werner Maschek, Nicola Forgione, Francesco Poli and Francesco Oriolo
12 Assessment of Solar-Coal Hybrid Electricity Power Generating Systems
Moses Tunde Oladiran, Cheddi Kiravu and Ovid Augustus Plumb
20 A Comparison of Different Communication Tools for Distance Learning in Nuclear Education
Glenn Harvel and Wendy Hardmann
34 Aims and First Assessments of the French Hydrogen Pathways Project HyFrance3
Alain Le Duigou, Marie-Marguerite Qumr, Pierre Marion, Philippe Menanteau, Pascal Houel, Laure
Sinegre, Lionel Nadau, Aline Rastetter, Aude Cuni, Philippe Mulard, Loc Antoine and Thierry Alleau
41 Experimental Study of Airflow-Mixture in HVAC Unit by Using PIV
Yu Kamiji, Atsuhiko Terada and Hitoshi Sugiyama
49 H
2
Production from Wind Power in a Wind Farm in Spain
Milagros Rey Porto, Mnica Aguado, Raquel Garde, Gabriel Garca and Trinidad Carretero
60 Experimental Studies on Critical Heat Flux in a Uniformly Heated Vertical Tube at Low Pressure
and Flowrates
Husham M. Ahmed
70 Environmental Impacts of Oil and Gas Production in Nigeria
Cecily. O. Nwokocha, Donatus. U. Ugwu, Abiodun. B. Fagbenro and Emmanuel. J. Cookey
76 Impact of the Target Film Thickness on the Properties of the Neutron Tube
Shuang Qiao, Shiwei Jing and Mingrui Zhang
Power and Electronic System
81 A Novel Transient Current-Based Differential Algorithm for Earth Fault Detection in Medium
Voltage Distribution Networks
Mohamed F. Abdel-Fattah and Matti Lehtonen
90 Chaos Control and Modified Projective Synchronization of Chaotic Dissipative Gyroscope Systems
Faezeh Farivar, Mahdi Aliyari Shoorehdeli, Mohammad Teshnehlab and Mohammad Ali Nekoui

101 Investigation of Interline Dynamic Voltage Restorer with Virtual Impedance Injection
Ahmed Elserougi, Ahmed Hossam-Eldin, Ahmed Massoud and Shehab Ahmed
110 Innovative Renewable Energy-Load Management Technology via Controlled Weight Motion
Mohammed A. El-Kady, Mamdooh Saud Thinyyan Al-Saud and Majeed A.S. Alkanhal
117 A Parameter Determination Method of Distribution Voltage Regulators Considering Tap Change
and Voltage Profile
Yuji Hanai, Yasuhiro Hayashi, Junya Matsuki, Yoshiaki Fuwa and Kenjiro Mori
126 The Development of the Generation Expansion Planning System Using Multi-Criteria Decision
Making Rule
Seokman Han, Koohyung Chung and Balho H. Kim
132 Study on the Photovoltaic System with MPT and Supercapacitors
Jan Leuchter, Pavel Bauer and Vladimir Rerucha
140 An Experience of On-site PD Testing for Condition Monitoring of an 11 kV PILC Cable Insulation
System
Xiaosheng Peng, Chengke Zhou and Xiaodi Song
146 Research on IPv6 Transition Evolvement and Security Architecture of Smart Distribution Grid
Data Communication System
Xin Miao and Xi Chen
150 Operational Reliability in Transmission Power Grid
Daniel Morar, Basarab Guzun and Ioan Rodean
Journal of Energy and Power Engineering 6 (2012) 1-11

Simulation of Fast Gas Injection Expansion Phase
Experiments under Different Pressures
Donella Pellini
1
, Werner Maschek
1
, Nicola Forgione
2
, Francesco Poli
2
and Francesco Oriolo
2
1. Karlsruhe Institute of Technology, Karlsruhe D-76021, Germany
2. Dipartimento di Ingegneria Meccanica, Nucleare e della Produzione, Universit di Pisa, Pisa 56126, Italy

Received: November 11, 2010 / Accepted: April 28, 2011 / Published: January 31, 2012.

Abstract: The injection of a high pressure gas into a stagnant liquid pool is the characteristic phenomenon during the expansion phase
of a hypothetical core disruptive accident in liquid metal cooled fast reactors. In order to investigate lots of mechanisms involved in this
phase of the accidents evolution, an experimental campaign called SGI was performed in 1994 in Forschungszentrum Karlsruhe, now
KIT. This campaign consists of nine experiments which have been dedicated to assess the effects of different pressure injection, the
nozzles size and the presence of inner confinement in the formation of the rising bubble. Three of these experiments, which were
focused on the pressure effects, have now been simulated with SIMMER III code and with FLUENT 6.3, a commercial CFD code. Both
codes, despite their different features, have showed a good agreement with the experimental results. In particular, time trend evolutions
of pressures and bubble volumes have been well reproduced by simulation. Furthermore, both codes agree on the shape of the bubble,
even though they have evidenced same discrepancies with the experimental shape.

Key words: Expansion phase, SIMMER III code, FLUENT.

1. Introduction

The sodium cooled fast reactors (SFR) represents
one of the proposed Generation IV systems [1] chosen
from the Generation IV International Forum in order to
enhance the future contribution and benefits of nuclear
energy utilization.
Therefore, the simulation of core disruptive
accidents and the assessment of potential, thermal and
mechanical energy for new design becomes of interest
too. Also new simulation tools have been developed
meanwhile and are applied in this domain.
In case of a severe accident, a mixture of fuel and
steel vapor, which forms an expanding bubble, might
be driven out of the damaged core into the sodium pool
(so-called expansion phase). This causes the sodiums
level raise and, consequently, the cover of gas pressure

Corresponding author: Donella Pellini, Ph.D., research
field: thermo-hydraulic of liquid metal cooled nuclear reactors.
E-mail: donella.pellini@kit.edu.
increase.
Experiments dealing with the injection of a high
pressure gas into a stagnant liquid pool, such as the
demonstration fast breeder reactor program (DFBR) in
Japan [2] or the previous campaigns performed by
Tobin and Cagliostro [3], describe characteristic
phenomena taking place during the expansion phase
and are of interest for testing and benchmarking.
In order to investigate further the mechanisms
involved in this phase of the accident evolution, an
experimental campaign called SGI was performed
in1994 in former Forschungszentrum Karlsruhe, now
KIT [4].
In addition to the experimental studies, the
expansion phases knowledge needs, also, a lot
numerical analysis focused on studying the bubbles
growth, the interfacial instabilities effects and the
entrainment which plays a crucial role in the expansion
phase [5, 6].
D
DAVID PUBLISHING
Simulation of Fast Gas Injection Expansion Phase Experiments under Different Pressures

2
Therefore from the computational viewpoint, a
simple and clear geometry together with the use of
simple materials is helpful in the assessment of codes
which need models to handle the complex phenomena
involved in this phase.
Thus SGI campaign, which fulfils these
requirements, has been chosen to compare the results
obtained from the simulations SIMMER III and
FLUENT codes of three of the nine experiments of
which the campaign is made of.
FLUENT 6.3 [7] is one of the most known
computational commercial fluid-dynamics code. Its
flexible features make it suitable to deal with several
different fluid dynamics problems concerning mixing,
turbulence, chemical reactions thus allowing to carry it
out in a lot of conventional industrys fields [8] and in
the nuclear industry, too [9].
SIMMER III is a two-dimensional (2D), three
velocity-fields, multi-component, multiphase, Eulerian
fluid-dynamics code coupled with a space and energy
dependent neutron kinetics model.
It was originally developed to simulate the events
sequence of the Core Disruptive Accidents (CDA) in
Liquid Metal Fast Reactors (LMFR) [10, 11], even
though, over the years, it has been developed into a
flexible tool which can deal with various problems
consistent with its modeling framework such as safety
analysis in advanced fast reactors up to the new
accelerator driven systems [12], steam explosion and
fuel coolant interaction problems [13, 14] and, more
generally, multiphase flow problems [15, 16].
2. Experimental Facility
The SGI (acronym of Schnelle Gas Injection)
campaign has been aimed to investigate on the
mechanism and phenomena involved in the expansion
phase.
In order to achieve this goal, an experimental facility,
which was built by SRI International, California, has
been used in Forschungszentrum Karlsruhe (now KIT)
for a series of experiments in which a certain amount of
gas has been injected into a water pool aiming at
reproducing the typical phenomenology of the
expansion phase in a nuclear reactor. The facility
consists of two vessels connected through a short tube
as it is shown in Fig. 1.
The first vessel is the so-called pressure vessel
which is completely filled with nitrogen at 293.15 K
and with a pressure ranging from 0.3 and 1.1 MPa.
The second vessel is the so-called main vessel and
it consists of an acrylic cylinder with an inner diameter
of 33 cm. Since it simulates the reactor coolant pool, it
has been partially filled with water and partially with
air at 293.15 K at atmospheric pressure.
Furthermore, inner walls have been placed inside
this vessel aimed at reproducing the reactors inner
structures.

Fig. 1 Scheme of the experimental facility for the SGI
campaign [4].

3
As it was previously mentioned, the two vessels
have been connected through a short tube, which was
closed by two sliding doors and a thin metal foil above
the doors.
The doors were opened by exploding
hydrogen-oxygen gas which drove two pistons towards
the doors, pushing them sideways in opposite
directions. Their role was to open the gas flow cross
section within a short time with the minimum
disturbance of the gas flow.
The campaign has investigated the effects of
different pressure, nozzle diameter and the presence of
inner confinement walls on the formation of the rising
bubble. Therefore, nine tests have been performed
depending on these parameters, as reported in Table 1.
Three tests, respectively 91, 93 and 95, have been
considered in this work.
3. Numerical Simulation
In order to perform a comparison between two
different kinds of codes the tests 91, 93, 95 have been
chosen. In fact, these tests have been arranged with the
same geometrical features, that is the presence of an
inner vessel wall with a diameter of 23 cm and the
nozzle diameter equal to 9 cm.
The only parameter changed has been the nitrogen
injection pressure which has been respectively set
equal to 1.1 MPa for test 91, 0.6 MPa for test 93 and 0.3
MPa for test 95 (see Table 1).
3.1 SIMMER III Features and Geometrical Domain
As mentioned previously, SIMMER III is a
two-dimensional (2D), three velocity-fields,
multi-component, multiphase, Eulerian fluid-dynamics
code coupled with a space and energy dependent
neutron kinetics model.
SIMMER III handles by default LMFR core
materials (fuel, coolant, control and fission gas) in solid,
liquid and vapor phases.
The materials mass distribution is modeled through
27 density components and the energy distributions are
modeled by only 16 energy components, because some
Table 1 Experimental conditions of the nine tests performed.
Test
number
Presence
of inner
structures
Inner
structure
diameter (cm)
Nozzle
diameter
(cm)
Initial
pressure
(MPa)
21 yes 23 6 1.1
29 no 33 9 1.1
30 no 33 6 1.1
32 no 33 9 0.6
33 yes 9 9 1.1
37 yes 9 9 0.3
91 yes 23 9 1.1
93 yes 23 9 0.6
95 yes 23 9 0.3

density components can be assigned to the same energy
components (e.g. a mixture of vapor components can
be defined with the same energy component).
The materials density and energy components are
obtained through the calculation of mass, momentum
and energy conservation equations using three
velocity-fields (two liquids and one vapor).
The fluid-dynamics solution algorithm is based on a
time factorization method (Four Step Algorithm)
which determines intra-cell interfacial area source
terms, heat and mass transfer and momentum exchange
functions separately from inter-cell fluid convection.
A higher-order spatial differencing scheme is the
standard scheme to reduce numerical diffusion and, in
particular, it is necessary in treating multiphase
phenomena.
Concerning the simulations, a cylindrical domain has
been set up for representing the test section (see Fig. 2). It
has been subdivided into 31 radial and 68 axial cells
and inside it, the main vessel takes up the radial cells
from 1 to 31 and the axial cells from 32 to 68.
The cover gas region is included in the axial cells
from 56 to 68 and in the cells from 1 to 31 radially,
according to the volume provided.
The pressure vessel, the connection tube and the
inner walls inside the main vessel have been shaped
throughno calculation regions.
3.2 CFD Code and Computational Domain
FLUENT features make this code suitable for
carrying out a CFD simulation of the experiments
chosen and for the comparison with SIMMER III code.

4

Fig. 2 SIMMER III geometrical domain.

The simple geometry of the experimental facility
allows to set up a two dimensional axial symmetric
computational domain (see Fig. 3), which has been
divided into about 10800 cells.
The time step chosen for the simulations is equal
to10
-6
s.
The time step and the mesh have been settled on
after a preliminary activity aimed to verify the
solutions independence from both parameters.
Among the several turbulence models available in
FLUENT, the standard - model has been chosen
aimed at describing the turbulent motion of the two
fluids.
Furthermore, a standard wall functions approach has
been used for the near wall region modeling. A first
order upwind scheme interpolates the convective terms
in the momentum equation while the pressure-velocity
coupling is numerically taken into account through the
SIMPLE algorithm.
In order to simulate the phenomena involved in the
SGI experiments, the Volume of Fluid (VOF) model
has been selected. This model, as largely suggested
[17], is the most suitable to deal with problems
involving the formation and the motion of a large gas
bubble in a liquid. In fact, the VOF model determines

Fig. 3 Computational domain for FLUENT simulation.

the interface motion between two immiscible phases
with different density and viscosity through a phase
indicator f defined as volume fraction function.
Therefore, f = 1 and f = 0 represent the two extremes,
respectively, for the liquid and the gas phase while the
range included between these values is representative
for a two-phase cell with a part of interface.
The continuity equations solution for each phase
accomplishes the interface tracking. Obviously, the
assumptions of absence of phase change, the
incompressibility of the liquid phase, homogeneity and
immiscibility of the two fluids with a well defined
interface has to be made. According to this, for the SGI
simulations gas has been considered the primary phase
and water the secondary phase.
4. Comparison among Calculations and
Experimental Results
4.1 Test 91
4.1.1 Pressure
One of the parameter chosen for the comparison
among codes and experimental results is the pressure
because of the availability of direct experimental

5
measurements. As it has been mentioned previously,
the test 91 has been performed injecting nitrogen at 1.1
MPa and 293.15 K in the main vessel containing water.
As can be seen in Fig. 4, which shows the
experimental and calculated pressure in the cover gas
region, at about 0.0074 s a sharp peak of about 12 MPa
arises due to the injection of the gas. In fact, the gas
bubble begins to grow inside the water, raising the
water level and at the same time compressing the cover
gas.
The maximum pressure is reached when the bubble
pushes the water up to the top of the main vessel
moment and reaches the maximum dimension before it
breaks up. Therefore, the pressure decreases rapidly.
The experimental curve shows a very small second
peak of about 2 MPa at about 0.08 s. This indicates the
experimental pool surface is domed but the pressure
wave following the impact at the top corner of the
facility is reduced on its way to the top center.
Both codes are in good agreement with the
experimental results, even though SIMMER simulation
shows a very small delay in timing with respect to
FLUENT, which results overlaps the experimental ones.
Concerning the second small peak, SIMMER
estimates a value of 2.3 MPa at about 0.0085 s while
FLUENT presents timing in coincidence with
experimental data but a value of 2.8 MPa. In any case,

0.0
2.0
4.0
6.0
8.0
10.0
12.0
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT

Fig. 4 Pressure transient in cover gas region for Test 91.
the second peak is more defined from the simulations
with respect to experiment. This is probably due to a
slight overestimation of the pressure wave from both
codes. Experimentally, the pressure stabilizes at about
0.7 MPa after 0.009 s. SIMMER III results are in
agreement with this trend, even though the calculated
value is slightly lower than the experimental results
(about 0.6 MPa). FLUENTs results show some
oscillations around a value of 0.7 MPa.
It must be note that after 0.015 s FLUENT evaluates
a pressure increase up to 1.6 MPa, perhaps due to an
overestimation of bubble generation.
The pressure trend in the pressure vessel is shown in
Fig. 5. When the sliding doors open the nitrogen begins
to flow into the water pool, causing the expanding
bubble to rise and, at the same time, decreasing the
pressure in this vessel.
The pressure reaches its minimum of 0.52 MPa at
about 0.01 s due to the break-up of the expanding
bubble in the water region which decrease the pressure
in the main vessel, allowing the lower vessel pressure
to rise.
Experimental data are in very good agreement with
both codes, even though SIMMER reproduces almost
exactly experimental values and trends. FLUENT
shows more oscillations during the first 0.005 s and a
small underestimation of the minimum pressure value.

0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT

Fig. 5 Pressure transient in pressure vessel for Test 91.

6
4.1.2 Gas Bubble Volume
Since the bubble growth plays a crucial role in the
experiment, the evaluation of the gas bubble volume
was chosen for the comparison.
The experimental data are available up to 0.00538 s
that is the instant when considerable cavitation at the
upper edge of the inner cylinder walls sets in.
During this time there is a very good agreement
between experimental and calculated results, as
highlighted in Fig. 6.
In particular, the two codes evaluate in the same way
the bubble volume, therefore, an overlapping of the
results obtained from simulations can be noted during
this time range.
Despite this agreement, as can be seen in Fig. 7
where the sequence of the bubble shapes development
is shown, both codes exhibit some discrepancies in the
prediction of the bubbles shape with respect to the
contour plot of the bubble surface obtained in the
experiment through high speed photographs. Since the
earlier moments of the bubbles growth (0.00277 s) the
experimental bubbles shape is domed while, at the
same time frame, the bubbles shape obtained from
both codes presents a concavity which becomes more
marked as the time increases, up to 0.00538 s.
4.1.3 Gas Velocity
The last parameter chosen for this work has been the
gas exit velocity from the nozzle.
The experimental data have been estimated from the
displaced water volume and therefore they are not
obtained from a direct measurement.
Furthermore, they are available just for a very short
time ranged from 0.0029 to 0.0046 s. Therefore, this
comparison must be focused on the FLUENT and
SIMMER results.
The calculated velocity data has been taken from the
bottom cell closed to the axis, both for FLUENT and
SIMMER.
As can be noted in Fig. 8, even though the two codes
exhibit the same trend, FLUENT shows more
oscillation during the first 0.0085 s with respect to
SIMMER and the experimental results.
0.0E+00
5.0E-04
1.0E-03
1.5E-03
2.0E-03
2.5E-03
3.0E-03
3.5E-03
4.0E-03
2.0E-03 3.0E-03 4.0E-03 5.0E-03 6.0E-03
Time [s]
V
o
l
u
m
e

[
m
3
]
Experimental
SIMMER III
FLUENT

Fig. 6 Gas bubble volume comparison for Test 91.

Fig. 7 Experimental and numerical bubbles shape
evaluation for Test 91.

-3.0E+02
-2.0E+02
-1.0E+02
0.0E+00
1.0E+02
2.0E+02
3.0E+02
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
G
a
s

E
x
i
t

V
e
l
o
c
i
t
y

[
m
/
s
]
Experimental
SIMMER III
FLUENT

Fig. 8 Gas exit velocity for Test 91.

7
SIMMER tends to overestimate slightly the
maximum velocity, although the timing of both codes
agrees completely.
From 0.0086 s to 0.011 s the two codes show same
trends and values, but afterwards FLUENT, the
minimum velocity value is smaller than SIMMER III
value. This might explain the overestimation of the
minimum pressure value observed in Fig. 5.
4.2 Test 95
4.2.1 Pressure
Fig. 9 shows the experimental and calculated
pressure in the cover gas region considering an
injection gas pressure of 0.3 MPa. This test has to be
considered as the lower extreme of the three tests
chosen for this comparison because of its lowest initial
gas injection pressure (0.3 MPa).
Due to the lower injection pressure, the bubble
expands slower with respect to test 91, thus
compressing the cover gas region in a weaker manner.
Therefore, the experimental pressure peak reaches the
maximum value of about 0.5 MPa in about 0.0175 s
and the pressure raise exhibits a mild peak. The
pressure decrease is milder than in test 91, too.
Concerning the simulations, Fig. 9 highlights a quite
good agreement between the codes and the
experimental data, although some differences can be
noted.

0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.0E+00 5.0E-03 1.0E-02 1.5E-02 2.0E-02 2.5E-02 3.0E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT

Both codes highlight a little delay (about 0.001 s) in
reaching the maximum. FLUENT results show a slight
underestimation of the pressure curve, with a
maximum peak of about 0.45 MPa.
SIMMER pressure curve for the first 0.0165 s can be
placed between the experimental and the FLUENT
curve.
Afterwards, SIMMER evaluates a first peak which is
sharper in the experiment and then it overestimates the
main peak, reaching a value of 0.546 MPa.
Furthermore, the pressure decrease presents a
steeper slope with respect of the other two curves and a
lower peak at 0.022 s.
From 0.025 s on both codes show some oscillations.
The pressure curve of the pressure vessel, shown in
Fig. 10, is similar to the trend shown in Fig. 5, because
of the same phenomenology.
As can be expected, the minimum of 0.18 MPa is
reached at about 0.0185 s, therefore later than in test 91.
Experimental data are in better agreement with both
codes than in the cover gas region. In particular,
SIMMER curve matches the experimental results up to
the minimum and then shows a little underestimation of
the pressure values. FLUENT curve presents lots of
oscillations during the whole transient and a little
underestimation of the minimum.
As can be seen in Fig. 11, the evaluation of the
bubble volume is in good agreement between FLUENT

0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.0E+00 5.0E-03 1.0E-02 1.5E-02 2.0E-02 2.5E-02 3.0E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT


8
0.0E+00
5.0E-04
1.0E-03
1.5E-03
2.0E-03
2.5E-03
3.0E-03
3.5E-03
0.0E+00 2.0E-03 4.0E-03 6.0E-03 8.0E-03 1.0E-02 1.2E-02
Time [s]
V
o
l
u
m
e

[
m
3
]
Experimental
SIMMER III
FLUENT

Fig. 11 Gas bubble volume comparison for Test 95.

and SIMMER, even though the experimental curves
slope is slightly steeper than those obtained from the
simulations.
Observing Fig. 12, which shows the bubbles shape
growth at 0.0048 s (early stage) and at 0.0117 s
(advanced stage), it is possible to note that from the
early stage on, the bubble growth numerically
evaluated shows a very pronounced concavity similarly
to what has been observed in Fig. 7 for test 91.
Therefore, the bubble volume is underestimated in
both codes. However, it must be pointed out that the
experimental expanding bubble exhibits a dimple in the
dome, instead of homogeneous dome as shown in Fig.7
for test 91. The reasons for these differences in the
bubble shape have not been clarified yet.
4.2.3 Gas Velocity
The evaluation of the gas exit velocity from the
nozzle is shown in Fig. 13.
Likewise to test 91, the availability of experimental
data just for a very short time (from 0.0061 s to 0.01 s)
allows focusing on the comparison between the two
codes.
The experimental data shows a linear trend, which is
better approached from SIMMER simulation, although
the evaluation of the velocity is slightly
underestimated.

Fig. 12 Experimental and numerical bubbles shape
evaluation for Test 95.

-150
-100
-50
0
50
100
0.0E+00 5.0E-03 1.0E-02 1.5E-02 2.0E-02 2.5E-02 3.0E-02
Time [s]
G
a
s

E
x
i
t

V
e
l
o
c
i
t
y

[
m
/
s
]
Experimental
SIMMER III
FLUENT


FLUENTs curve shows oscillation that are not
evaluated from SIMMER even though SIMMER curve
seemed to represent a sort of averaged curve for
FLUENT.
Similarly to what observed in Fig. 8 for test 91
FLUENT underestimates the minimum velocity and
the evaluation of the last part of the curve (from 0.022 s
on) with respect to SIMMER, although both codes
agree on the reaching of a relative maximum of 30 m/s
at about 0.026 s and the following velocity decrease.

9
4.3 Test 93
4.3.1 Pressure
Test 93 has been performed injecting nitrogen at 0.6
MPa, therefore, it might be considered as a middle case
between test 91 and test 95. In Fig. 14 cover gas
pressure curves are shown.
As can be seen the experimental pressure trend looks
like more similar to the test 91 than to the text 95,
exhibiting a sharp peak. The peak reaches its maximum
of about 4.76 MPa in about 0.01 s. This value is very
close to the value found for the test 93.
Even the pressure decrease shows similarities with
test 91, although a series of three lower peaks can be
noted during the phase of pressure decrease which lasts
up to 0.016 s.
The peaks width therefore can be placed between
the two widths of test 91 and test 95. Afterwards, the
pressure stabilizes at about 0.35 MPa.
Concerning the simulations, Fig. 14 highlights a
substantial agreement between the codes and the
experimental data, although some differences can be
noted.
Both codes exhibit a delay which starts for both at
0.0075 s thus the maximum is reached after about
0.005 s from SIMMER and 0.004 s from FLUENT
with respect to the experimental results.
The peaks value is underestimated, too. While
SIMMER evaluates a peak of 4.3 MPa, FLUENT
evaluates the maximum at 3.3 MPa.
Furthermore, during the pressure decrease FLUENT
curve shows three defined peaks that might be connected
to the three close peaks observed experimentally.
Afterwards, both codes are in good agreement with
experimental results.
Fig. 15, which shows the pressure in the lower vessel,
highlights the good agreement between the codes and
the experimental results, especially up to 0.012 s.
In the following time, similarly to the simulations of
the two previous cases, FLUENT and SIMMER
minimum pressure value is smaller than the
experimental one.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT


0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
P
r
e
s
s
u
r
e

[
M
P
a
]
Experimental
SIMMER III
FLUENT


As can be seen in Fig. 16, the curves obtained from
FLUENT and SIMMER match each other in the time
range considered. Nevertheless, the codes evaluated a
milder slope than that obtained from experiment.
Comparing these results with Fig. 6 and Fig. 11, it is
possible to note that the evaluation of the bubble
volume behaves in an intermediate manner. In fact, in
test 91, the numerical curves almost match the
experimental one, while in test 95 it is possible to
observe a certain deviation from the experimental
values.
The deviation of test 93 is smaller than in test 95 and,
obviously larger than in test 91.

10
0.0E+00
5.0E-04
1.0E-03
1.5E-03
2.0E-03
2.5E-03
3.0E-03
0.0E+00 2.0E-03 4.0E-03 6.0E-03 8.0E-03
Time [s]
V
o
l
u
m
e

[
m
3
]
Experimental
SIMMER III
FLUENT

Fig. 16 Gas bubble volume comparison for Test 93 .

4.3.3 Gas Velocity
The evaluation of the gas exit velocity from the
nozzle is shown in Fig. 17.
Likewise to the other two tests, the experimental
data are available just for a time ranging from 0.004 s
to 0.006 s. In this time range SIMMER shows a good
agreement with the experimental data.
In analogy to what has been observed for test 91 and
test 95, FLUENT exhibits oscillations during the first
0.011 s, underestimating the minimum gas exit velocity
value.
5. Conclusions
The injection of a high pressure gas into a stagnant
liquid pool, which characterizes phenomena of the
expansion phase of a hypothetical core disruptive
accident in liquid metal cooled fast reactors, has been
investigated through the experimental campaign SGI
performed in 1994 in Forschungszentrum Karlsruhe,
now KIT.
Three tests of the campaign have been simulated
with two different codes, SIMMER III and FLUENT,
and the results have been compared with the
experimental data.
This activity has turned out to be helpful in assessing
the possible differences in the results owing to the use
of a multi-field, multi-phase, multi-component
accident code like SIMMER III and a CFD code like
-250
-200
-150
-100
-50
0
50
100
150
200
0.0E+00 3.0E-03 6.0E-03 9.0E-03 1.2E-02 1.5E-02 1.8E-02
Time [s]
G
a
s

E
x
i
t

V
e
l
o
c
i
t
y

[
m
/
s
]
Experimental
SIMMER III
FLUENT


FLUENT, which of course have different features and
different levels of details.
Moreover, this activity has provided a further
qualification of the SIMMER III code, in simulating
Core Disruptive Accidents (CDA) phenomena
involved in sodium fast reactors.
The main results concerning the activity performed
can be summarized as in the following:
the comparison among the experimental pressure
curves and the numerical results obtained from the two
codes has shown a very good agreement, both for the
main and the pressure vessel;
the evaluation of the bubble volume is matching
between the codes and it is in good agreement with
experimental data. An underestimation of the bubble
volume can be observed at lower initial gas injection
pressure;
the analysis of the bubbles shape has revealed
some discrepancies between the experimental results
and the numerical simulation performed with both
codes. In fact, SIMMER and FLUENT have exhibited
a concavity in the center of the bubble which are not
observed in the experiments;
the comparison of the gas exit velocity performed
with the two codes has shown a general agreement
between them, although an oscillatory trend has been
noticed in the simulation of the three test with
FLUENT.

11
The simulation activity has revealed the comparable
capability of SIMMER III with respect to FLUENT
code in reproducing the experimental findings, getting
similar results with reduced computational resources
and time.
References
[1] World Nuclear Association, Generation IV Reactors [online],
http://www.world-nuclear.org/info/default.aspx?id=530&
terms=Generation%20IV (updated Dec. 2010).
[2] T. Nakamura, H. Kaguchi, I. Ikarimoto, Y. Kamishima, K.
Koyama, S. Kubo, et al., Evaluation method for structural
integrity assessment in core disruptive accident of fast
reactor, Nucl. Eng. Des. 227 (2004) 97-123.
[3] R.J. Tobin, D.J. Cagliostro, Energetics of simulated
HCDA bubble expansions: Some potential attenuation
mechanisms, Nucl. Eng Des. 58 (1980) 85-95.
[4] L. Meyer, D. Wilhelm, Investigation of the Fluid Dynamics
of a Gas Jet Expansion in a Liquid Pool, Technical report
KfK5307, Forschungszentrum Karlsruhe, 1994.
[5] M. Epstein, H.F. Fauske, S. Kubo, T. Nakamura, K.
Koyama, Liquid entrainment by an expanding core
disruptive accident bubbleA Kevin/Helmotz
phenomenon, Nucl. Eng. Des. 210 (2001) 53-77.
[6] M.J. Tan, J.M. Delhaye, An experimental study of liquid
entrainment by expanding gas, J. Fluids Engineering 109
(1987) 436-441.
[7] FLUENT 6.3 Users Guide, Fluent Inc., Cavendish Court,
Lebanon, Sept., 2006.
[8] W.J. Rider, D.B. Kothe, Reconstructing volume tracking,
Journal of Computational Physics 141 (2) (1998) 112-152.
[9] P. Di Marco, N. Forgione, M. Tarantino, Analysis of the
transient rise of an initially spherical gas bubble in a
stagnant liquid, in: Proceedings of XXIV National
Congress UIT on Heat Transfer, Napoli, Italy, 2006.
[10] M. Sharabi, W. Ambrosini, N. Forgione, H. Shuisheng,
SCWR rod bundle thermal analysis by a CFD code, in:
Proceedings of 16th International Conference on Nuclear
Engineering, Orlando, FL, USA, 2008, pp. 1-7.
[11] S. Kondo, Y. Tobita, K. Morita, N. Shirakawa, SIMMER
III: A Computer Program for LMFR Core Disruptive
Accident Analysis, Technical report JNC TN 9400
2003-071, Research Document, O-arai Engineering
Center, Japan Nuclear Cycle Development Institute, 2003.
[12] Y. Tobita, Sa. Kondo, H. Yamano, S. Fujita, K. Morita, W.
Maschek, et al., The development of SIMMER-III, an
advanced computer program for LMFR safety analysis, in:
Proceeding of the IAEA/NEA Technical Meeting on Use
of Computational Fluid Dynamics (CFD) Codes for Safety
Analysis Reactors Systems Including Containment, Pisa,
Italy, Nov. 11-14, 2002.
[13] K. Morita, A. Rineiski, E. Rineiski, E. Kiefhaber, W.
Maschek, M. Flad, et al., Mechanistic SIMMER III
analyses of severe transient in accelerator driver systems
(ADS), in: Proceedings of the 9th International
Conference on Nuclear Engineering (ICONE 9), Nice
Acropolis, France, Apr. 8-12, 2001.
[14] K. Morita, S. Kondo, Y. Tobita, D.J. Brear, SIMMER III
applications to fuel-coolant interactions, Nucl. Eng. Des.
189 (1999) 337-357.
[15] S. Wang, M. Flad, W. Maschek, P. Agostini, D. Pellini, G.
Bandini, et al., Evaluation of a steam generator tube
rupture accident in an accelerator driven system with lead
cooling, Prog. Nucl. Energy 50 (2008) 363-369.
[16] D. Wilhelm, P. Coste, Experience with the multiphase
code SIMMER and its models for interfacial area
convection, in: Proceedings of International Meeting on
Trends in Numerical and Physical Modelling for Industrial
Multiphase Flows, Cargese, France, Sept. 27-29, 2000.
[17] K. Morita, T. Matsumoto, K. Fukuda, Y. Tobita, H.
Yamano, I. Sato, Experimental verification of the fast
reactor safety analysis code SIMMER-III for transient
bubble behavior with condensation, Nucl. Eng. Des. 238
(2008) 49-56.


Assessment of Solar-Coal Hybrid Electricity Power
Generating Systems
Moses Tunde Oladiran
1
, Cheddi Kiravu
1
and Ovid Augustus Plumb
2

1. Faculty of Engineering and Technology, University of Botswana, Gaborone, P/Bag 0061, Botswana
2. Department of Mechanical Engineering, University of Wyoming, Laramie, WY 82071, USA

Received: February 04, 2011 / Accepted: June 09, 2011 / Published: January 31, 2012.

Abstract: Botswana currently depends on electricity generated from coal-based power plant or electricity supplied from the border in
South Africa. The country has good reserves of coal and the solar radiation is sufficiently high to make solar thermal attractive for
generating electricity. The paper presents two conceptual coal-fired power station designs in which a solar sub-system augments heat
to the feed heaters or to the boiler. The thermal and economic analyses showed enhanced system performance which indicates that
solar power could be embedded into existing fossil fuel plants or new power stations. Integrating solar energy with existing or new
fossil fuel based power plants could reduce the cost of stand-alone solar thermal power stations, reduce CO
2
emissions and produce
experience necessary to operate a full scale solar thermal electricity generation facility.

Key words: Hybrid systems, solar, coal, economics, performance.

Nomenclature

A Annuitized construction costs, $
CC Total construction costs, $
dr Discount rate
h Specific enthalpy, kJ/kg
m
1
Total system mass flow rate, kg/hr
Q Heat input, kJ/kg
Q
s
Total solar input, kJ/hr
Q
mc
Maximum heat load for open feedwater heater, kJ/hr
Q
sc
Solar input to closed feedwater heater, kJ/hr
W Pump of turbine work, kJ/kg
y
c
Fraction of total mass flow extracted for closed feedwater
heater, dimensionless
Plant thermal efficiency
y
o
Fraction of total mass flow extracted for open feedwater
heater, dimensionless
1. Introduction
Electricity requirements in Botswana are supplied
through local generation at Morupule coal-based power
station and imports from the Southern Africa Power

Corresponding author: Moses Tunde Oladiran, Ph.D.,
associate professor, research fields: applied energy, energy
management, engineering education. E-mail:
oladiran@mopipi.ub.bw.
Pool, mainly from Eskom of the Republic of South
Africa (RSA). Botswana has abundant reserves of coal
estimated at 212 billion tones in various fields though
only one field is currently operational. With these huge
coal resources it is likely that Botswana would depend
on coal fired power stations for future electricity needs.
Indeed several such power stations are at various stages
of development to make the country self sufficient in
electricity and possibly become a power exporter to
neighbouring states. As CO
2
generated per capita will
grow steadily, environmental issues would increasingly
be of concern.
The country is a signatory to the United Nations
Framework Convention on Climate Change (UNFCCC)
and the Kyoto Protocol to control greenhouse gas
emissions. Some of the options under consideration to
limit CO
2
emission from power stations include use of
advanced power systems such as integrated gasification
combined cycle (IGCC) systems and carbon
sequestration technologies to recapture and store carbon
dioxide (i.e. carbon capture and storage). To generate
D
DAVID PUBLISHING
Assessment of Solar-Coal Hybrid Electricity Power Generating Systems

13
electricity on a long term basis there are aggressive
plans to introduce renewable technologies especially
solar thermal and biomass systems. Botswana has one
of the highest solar radiation regimes in the world
because the country receives over 3200 hours of
sunshine per annum and the average daily insolation on
a horizontal surface is approximately 21 MJ/m
2
[1].
From an economic, strategic and environmental point
of view it is appropriate to promote the use of solar
energy in Botswana. However, solar thermal electricity
involves huge capital costs.
Current economic assessments place the cost of
generating electricity using solar thermal power,
including storage, at 0.16 $/kWh compared to hard
coal at 0.05 $/kWh [2]. This has both discouraged
investment in solar thermal power and negatively
impacted innovation that results from operating
experience with new facilities. Solar-coal hybrid
thermal power plants have been proposed more
recently as an economically viable alternative to
stand-alone solar thermal power plants [3].
The advantage of the hybrid facility is that, in many
locations, transmission capacity already exists
whereas for a stand-alone solar facility transmission
capability may not be available in locations where the
solar radiation is sufficient to warrant investment in
solar thermal generating plants. In addition, the
solar-coal hybrid does not require storage which is
estimated to be approximately 15% of the total cost of
construction for a solar thermal facility utilizing
parabolic trough technology [2]. Furthermore, the
hybrid technology appears to offer retrofit potential
for existing coal fired power plants located where
annual solar radiation is sufficiently high. Finally, it
may be possible to generate additional power without
adding turbine capacity. Therefore, it is plausible to
use hybrids of solar-coal as a transition stage to fully
renewable energy production [3-5].
The current paper examined a generic coal fired
facility located where the solar insolation is
approximated by that at Morupule, the location of
Bostwanas single coal-fired power plant. Two possible
hybrid configurations are considered. In the first option
the solar input is utilized to gradually replace feedwater
heating. For this plant configuration (Fig. 1) the coal
consumption is constant but the turbine output increases
as solar energy replaces steam bled from the turbines in
order to provide feedwater heat. The second
configuration (Fig. 2) considered is one for which the
solar input replaces a portion of the boiler heat reducing
the quantity of coal as the solar input increases.
It was assumed that the first configuration would be
easier to implement at an existing facility because the
boiler does not require any modification. The only
necessary change to the existing facility is the addition
of heat exchangers to the feedwater heaters. However,
it is likely to be more difficult to control because of
the complexity of the feedwater heating scenario. The
second configuration appears to be more
straightforward in terms of control and, perhaps, more
economical to implement during new construction. In
both cases the amount of CO
2
produced per unit of
electricity generated is reduced in comparison to the
coal-fired power plant.
2. System Analysis
The generic Rankine cycle that is selected for
analysis is a standard superheat cycle having re-heat
and two feedwater heatersone open and one closed
[6]. The temperatures and pressures at various points
in the cycle are shown in Table 1. This cycle has a
thermal efficiency (reversible) of 43.1% [6] and a
parasitic energy cost for feedwater heating
amounting to about 18%. Just less than 25% of the
total mass flow of steam is extracted for feedwater
heating before the lowest pressure stages of the
turbine. If all of the feedwater heating is supplied by
solar energy the plant efficiency (reversible) could be
increased to 51%, based on the same coal input,
resulting in a significant increase in power output
(assuming that the turbines can accommodate the
additional steam flow rate).


14

Fig. 1 Option 1solar thermal feedwater heating.

Fig. 2 Option 2solar thermal input at boiler.


15

Table 1 Characteristics of basic Rankine cycle.
Operating characteristics Value
Boiler inlet 205
o
C, 8 MPa
Boiler (superheater) outlet 480
o
C, 8 MPa
Closed feedwater heater inlet 2 MPa
High pressure turbine outlet 7 MPa
Low pressure turbine inlet 440
o
C, 7 MPa
Open feedwater heater inlet
Low pressure turbine outlet
Condenser outlet
Open feedwater heater outlet
0.3 MPa
0.008 MPa
Saturated, 0.008 MPa
Saturated, 0.3 MPa
Power output (electric)-option 1
Power output (electric)-option 2
Steam rate
90-100 MW
100 MW
280,000 kg/hr
Boiler efficiency 90%
Generator efficiency 90%

For most of Botswana the average annual direct
normal irradiance (DNI) ranges from 6.5-7.5 kWh/m
2
.
For the Morupule area the DNI is approximately 6.9
kWh/m
2
but a value of 6.75 kWh/m
2
was used in the
current analysis to err on the conservative side. An
east-west tracking on a polar axis was considered for
parabolic trough collectors. The hourly irradiance is
illustrated in Fig. 3 [7]. Fig. 3 includes the assumption
of 75% efficiency for the parabolic trough field [7].
The results presented for both options are based on
hourly calculations for one day per year which is
assumed to be representative of the annual
performance. For option 1 the solar heat is added to
the feedwater heaters. Therefore, when the hourly
solar input is less than the total heat load for the open
feedwater heater will be given by:
mc
m
1
= b
9
- (1 - y
c
)b
8
- y
c
b
11
(1)
The fraction of steam extracted for the open
feedwater heater is given by:
y
o
=
h
9
-h
8
+
c
(h
8
-h
11
)-
s
m
T
,
h
S
-h
8
(2)
When the hourly solar input is greater than the total
heat load for the open feedwater heater then a portion
of the solar heat goes to the closed feedwater heater.
For this case the enthalpy balances for the two
feedwater heaters must be solved simultaneously to
determine the fraction of steam extracted for the
closed feedwater heater. The two enthalpy balances
can be written (for y
o
= u)

Fig. 3 Hourly direct normal irradiance.

(1 - y
c
)b
8
+ y
c
b
11
+
mc
m
1
, = b
9
(3)
y
c
b
7
+ b
10
+
sc
m
1
, = y
c
b
11
+ b
1
(4)
noting that
sc
=
s
-
mc

The work output by the turbines and work input to
the pumps per total system mass flow rate are given
by:
w
t1
= (b
2
- b
7
) + (1 - y
c
)(b
7
- b
3
) (5)
w
t2
= (1 -y
c
)(b
4
-b
5
) +(1 -y
c
-y
o
)(b
5
-b
6
)(6)
w
p1
= (1 - y
c
-y
o
)(b
8
- b
co
) (7)
w
p2
= b
10
-b
9
(8)
The total heat input is:
q
= (b
2
- b
1
) + (1 - y
c
)(b
4
-b
3
) (9)
The turbine total output, pump inputs, and heat
input were obtained for a 24 hour period and then the
plant thermal efficiency was calculated, by dividing
the net work by the heat input, i.e.
p = (w
t1
+ w
t2
- w
p1
- w
p2
) q
(10)
For option 2 the solar field was assumed to provide
heat up to 350
o
C. This allows for a much larger solar
contribution to the 100 MW (thermal) plant than that
for feedwater heating. The analysis is
straightforwardas the hourly thermal input from the
solar field increases the amount of thermal input from

16
coal is decreased to maintain a constant plant output.
The lower heating value for the coal is taken to be 20
MJ/kg.
For this scenario the turbine and pump work terms
remain constant but the heat input from combustion
process varies as the solar field contributes to the
thermal input from the boiler. For this case,
q
= q
mux
- q
soIu
(11)
where, again, the solar input must be summed over the
period when solar radiation is available and subtracted
from the total daily heat load, q
mux
to obtain the
thermal efficiency using Eq. (10).
3. Performance Analysis
In order to determine the economic performance, a
variable size of collector field was employed for both
configurations. Variable collector field costs were also
investigated. The base cost is $500/m
2
which is
slightly more than that in the analysis by Kaltschmitt,
et al. [2]. The annual operating and maintenance cost
is assumed to be $10/m
2
in current dollars and the cost
of coal is assumed to be 0.030$/kg. For option 1 the
base cost of construction for the solar facility is
annuitized over the assumed plant life using:
A = CC
d(1+d)
n
(1+d)
n
-1
(12)
The annuitized construction costs along with the
annual operating and maintenance costs are then used
to compute the current cost per kilowatt-hour for the
additional electricity generated. The plant life (n) is
assumed to be 25 years. This leads to a determination
of the optimum size for the solar contribution to the
existing facility. In addition, it is important to
establish the sensitivity of the solar field cost and
performance assumptions. For this case the coal
consumption remains constant and the coal-fired plant
is assumed to be operational in situ.
The costs for construction and operation and
maintenance for both the coal and solar facilities are
included in the analysis for the second option. These
costs for the solar part of the facility are assumed to be
the same as those utilized in option 1. For the coal
plant the construction costs are assumed to be 1.36
10
6
$/MW and the annual operating and maintenance
costs are assumed to be 50,000 $/MW [2]. The coal
plant capacity factor is assumed to be 0.8 leading to
7008 operating hours per year. The various costs and
operating parameters employed in the economic
analysis are summarized in Table 2.
4. Results and Discussion
The system thermal efficiency for option 1 without
solar input is 43.1%. With solar input, as shown in
Fig. 4, the efficiency increases to slightly above 46%
as the size of the collector field is increased. The
efficiency reaches a limit when the solar field provides
all of the feedwater heating for the period when solar
radiation is available. The results shown in Fig. 4 do
not include the pumping costs for the working fluid in
the solar field or other parasitic loads like the power
used for tracking.
For option 1 the cost of the excess electricity generated
in dollars per kilowatt hour is shown in Fig. 5 as a
function of the size of the solar field in square meters.
As illustrated in the figure there is an optimum solar
field size of approximately 90,000 m
2
that results in
the lowest cost for the electric power, just less than
0.108 $/kWh for the $500/m
2
field cost. For this case
the solar field provides 5.8 MW in addition to the 90
MW (electric) delivered by combustion of coal. For a
6% discount rate this leads to an overall cost of
electricity of just over 0.040 $/kWh assuming the base
cost of coal generated power to be 0.036 $/kWh as
computed for option 2. It was noted that, as a result of
non-availability of solar input, particularly in the early
morning and late afternoon hours, the results contain
discontinuities. Fig. 5 shows that the cost of electricity
per unit kilowatt hour increases as the assumed cost of
the solar field increases. Therefore, augmentation of
electricity production from fossil fuel plants with solar
system will be attractive where insolation is very high.
The effect of discount rate on the cost of electricity for
this option is illustrated in Fig. 6.

17
Table 2 Parameters used in economic analysis.
Name of variable Value
Cost of solar field, $/m
2
300-700
Annual O &M for solar field, $/m
2
10
Cost of coal, $/kg 0.030
Plant life, yrs. 25
Construction costs for coal plant, $/MW 1.36 10
6

Annual O &M for coal plant, $/MW 50,000
Plant capacity factor 0.8 (7008 hr/yr)
Discount rate (dr), % 6-10

1 2 3 4 5 6 7 8 9 10 11
x 10
4
0.43
0.435
0.44
0.445
0.45
0.455
0.46
0.465
collector field size, m
2
p
l
a
n
t

t
h
e
r
m
a
l

e
f
f
i
c
i
e
n
c
y

Fig. 4 Plant thermal efficiency versus collector field size
for option 1.

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
5
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0.26
2
c
o
s
t

o
f

e
l
e
c
t
r
i
c
i
t
y
,

$
/
k
W
h
$700/m
2
$500/m
2
$300/m
2

Fig. 5 Cost of excess electricity generated versus the size
of the solar field for option 1.

For small solar fields, less than 40,000 m
2
, the solar
contribution affects only the open feedwater heater.
As the fraction of steam extracted to supply heat to the
feedwater heater decreases the output from the low
pressure turbine increases in proportion to the
enthalpy difference between the extraction point and
the turbine outlet. The result is a constant estimated
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
5
0.1
0.15
0.2
0.25
2
c
o
s
t

o
f

e
l
e
c
t
r
i
c
i
t
y
,

$
/
k
W
h
dr = 10%
dr = 6%

Fig. 6 Cost of excess electricity generated for option 1 for
2 discount rates (dr).

cost for the extra power produced of just over 0.16
$/kWh. For solar fields above 40,000 m
2
the solar
input begins to contribute to the closed feedwater
heater increasing the output from both the high and
low pressure turbines. This causes the cost of extra
power produced via solar to decrease significantly to
just under 0.11 $/kWh as mentioned above.
With 90,000 m
2
of collector field all of the
feedwater heat is supplied by solar energy for 5 hours
of the day. Fig. 7 illustrates the quantity of excess
electricity generated as a function of the size of the
collector field. The optimum design (minimum cost of
excess power generation) occurs slightly before the
solar energy input reaches the point of supplying the
entire feedwater load for the sunshine period (11-hour
day) analyzed. For example, if the collector field is
increased to greater than 160,000 m
2
all of the
feedwater load is covered (11 hours per day) resulting
in 7.5 MW of extra power at a cost of 0.1439 $/kWh
(6 MW when multiplied by the plant capacity factor of
0.8). The average cost for power produced by the
hybrid facility is then 0.044 $/kWh.
For option 2 the plant thermal efficiency increases
as indicated in Fig. 8 beginning at 43.1% for no solar
contribution and peaking at 63.4% when the solar field
provides the maximum possible (350
o
C at 8 MPa) for
the 11 hours in which solar radiation is available. As

18
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
x 10
5
0
1
2
3
4
5
6
2
a
d
d
i
t
i
o
n
a
l

p
o
w
e
r
,

M
W

Fig. 7 Excess electricity generated by option 1 as a
function of solar field size.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
x 10
5
0.4
0.45
0.5
0.55
0.6
0.65
solar field area, m
2
p
l
a
n
t

t
h
e
r
m
a
l

e
f
f
i
c
i
e
n
c
y

Fig. 8 Plant thermal efficiency as a function of solar field
size for option 2.

for the option 1 analysis the parasitic losses associated
with the solar field, pumping and tracking power, are
not considered when calculating the plant thermal
efficiency.
As indicated above, for option 2 the base cost of
electricity generated from coal at a 6% discount rate is
0.036 $/kWh. As illustrated in Fig. 9 the average cost
of electricity increases as the size of the solar
contribution is increased. However, the cost is much
less than it would be for 100% solar, particularly if
storage is included. For example, for a 200,000 m
2
the
estimated electricity cost is 0.050 $/kWh providing 19
MW (electricity) of the total 100 MW. With 400,000 m
2
of solar resource the solar contribution is 30.9 MW at
0 0.5 1 1.5 2 2.5 3 3.5 4
x 10
5
0.04
0.045
0.05
0.055
0.06
0.065
0.07
solar field area, m
2
c
o
s
t

o
f

e
l
e
c
t
r
i
c
i
t
y
,

$
/
k
W
h

Fig. 9 Average cost of electricity for option 2 as a function
of solar field size.

a cost of 0.066 $/kWh. The solar heat was assumed to
be available at a maximum temperature of 350
o
C,
thus, if the size of the collector field is increased
significantly above 400,000 m
2
the average cost of the
electricity produced increases rapidly because all of
the energy collected is not utilized.
5. Conclusions
The results illustrate that solar-coal hybrid plants
can produce electricity at costs only slightly greater
than those for coal-fired power plants. Utilizing solar
thermal for feedwater heating demonstrated that by
adding solar energy system to an existing facility
potentially increased the power output by 5%-8%
while only very modestly increasing the power cost.
One of the advantages of this approach is that it
allows the construction of a small solar thermal unit
without storage or transmission that will produce
power at a moderate increase over coal and increase
the total output from an existing plant. It also allows
for the establishment of local construction and
operating and maintenance costs without the
construction of a larger and potentially more risky
stand alone solar thermal facility.
Option 2 which utilizes solar thermal to replace coal
at the boiler may be more applicable to new power
plant installation. For this case providing 20% to 25%

19
of the total plant capacity with solar can be
accomplished with a modest increase in the cost of the
electricity generated. This may be attractive in regions
where the peak load is a result of summer air
conditioning demand and corresponds with the peak in
solar radiation.
In conclusion, integrating solar thermal power with
existing or new fossil fuel-based power plants is a
useful strategy to reduce the cost of stand-alone solar
thermal power stations, reduce CO
2
emissions and
gain low risk experience necessary to operate a full
scale solar thermal electricity generation facility.
Acknowledgments
This work was completed while OAP was
supported by a Fulbright Award at the University of
Botswana. The authors express their appreciation to
the Fulbright Scholar Program, the University of
Botswana, and the University of Wyoming for making
this collaboration possible.
References
[1] Energy Trend Analysis, Energy Affairs Division
(Planning and Documentation Unit), Energy and Water
Resources, Ministry of Minerals, Gaborone, Botswana,
2003.
[2] M. Kaltschmitt, W. Streicher, A. Wiese, Renewable
Energy-Technology, Economics and Environment,
Springer-Verlag, Berlin, Germany, 2007.
[3] K. Bullis, Mixing solar with coal to cut costs: A new
strategy could reduce coal plant emissions and cut the cost of
solar power [online], 2009, MIT Technology Review,
http://www.technologyreview.com/energy/23349/
(accessed Apr. 10, 2010).
[4] J. Guerrerio, Hybrid Power Plants Pair Solar with Coal to
Reduce Emissions [online], 2010,
http://www.examiner.com/x-2903-Energy-Examiner~y20
09m2d2-Hybrid-power-plants-pair-solar-with-coal-to-red
uce-emissions (accessed Apr. 10, 2010).
[5] P. Fairley, Cutting Coal Use with Sunshine [online], 2009,
http://www.technologyreview.com/energy/22080/?a=f.
(accessed Apr. 10, 2010).
[6] M.J. Moran, H.N. Shapiro, Fundamentals of Engineering
Thermodynamics, 5th ed., John Wiley & Sons, NJ, 2004.
[7] S.A. Kalogirou, Solar Energy Engineering, Processes and
Systems Academic Press, MA, USA, 2009.


A Comparison of Different Communication Tools for
Distance Learning in Nuclear Education
Glenn Harvel and Wendy Hardmann

Faculty of Energy Systems and Nuclear Science, University of Ontario Institute of Technology, Oshawa, Ontario L1H 7K4, Canada

Received: December 07, 2010 / Accepted: May 05, 2011 / Published: January 31, 2012.

Abstract: Recent advancement in nuclear education learning has been through the use of computers and simulation related tasks
such as the use of industry codes. Further enhancements in nuclear education are being considered through the use of distance
learning technologies. The purpose of this work is to explore distance learning related tools to determine if they can provide an
enhanced learning environment for nuclear education. In this work, a set of tools are examined that can be used to augment or replace
the traditional lecture method. These tools are Mediasite, Adobe Connect, Elluminate, and Camtasia. All four tools have recording
capabilities that allow the students to experience the exchange of information in different ways. This paper compares recent
experiences with each of these tools in providing nuclear engineering education and assesses the various constraints and impacts on
delivery through direct feedback from students and instructors. In general, the tools were found to be useful for mature students on
the condition that the lecturer was comfortable with the tools and in some cases, adequate support from IT groups was provided.

Key words: Distance education, web based tools, nuclear engineering education.

1. Introduction

Nuclear generated electricity was first produced in
Canada in 1962, from the Nuclear Power
Demonstration (NPD) unit, the first CANDU (Canada
Deuterium Uranium) power plant. From that small
beginning grew an industry that conducts world-class
research, development, design, construction,
commissioning, operation and maintenance of nuclear
power plants, as well as research reactors and several
related technologies. The Province of Ontario, with a
population of 13 million, has been the centre of
Canadas nuclear industry. In particular approximately
50% of the electricity used in Ontario is generated
from CANDU nuclear units. Two other provinces
each operate 700 MW CANDU units and other
provinces are considering to build new nuclear power
plants. The current fleet of Canadas nuclear plants
ranges in age from 15 to 37 years, with a substantial

Corresponding author: Glenn Harvel, associate professor,
research fields: nuclear education, nuclear design, and thermal
hydraulics. E-mail: Glenn.Harvel@uoit.ca.
amount of life-extension work having already been
carried out on the older units, and similar projects
being planned for the remaining ones. The continuing
increase in demand for electricity as the population
grows and its standard of living increases, and despite
strong conservation efforts and the transfer of
significant manufacturing capacity abroad, the
Government of Ontario has requested proposals for
new nuclear generation. In addition to meeting
increased demand, the new units are required to
enable the closing of the remaining coal fired
generation, so as to meet environmental targets to
reduce greenhouse gas as well as particulate
emissions.
Corresponding to the aging of the nuclear power
plants is the demographics of the workforce. About
half of the engineers and scientists in the nuclear
industry are reaching retirement age in the next five
years. A similar situation exists in several other high
technology industries in Canada, and world-wide in
the nuclear technology sector. During the design and
D
DAVID PUBLISHING
A Comparison of Different Communication Tools for Distance Learning in Nuclear Education

21
construction of the early units, particularly in the
1970s and early 1980s, a combination of university
science and engineering programs and
industry-specific training produced the required
skilled personnel. As the design activities decreased in
the late 1980s and the construction of new units in
Ontario essentially stopped by mid 1990, universities
no longer had the number of students required to keep
offering nuclear-specific courses and degree programs.
At the start of the new millennium the industry
recognized the lack of specialized graduates being
produced by universities, and decided to establish the
University Network of Excellence in Nuclear
Engineering (UNENE) to address this problem [1].
The specific mandates of UNENE are to fund research
chairs at the seven member universities located in
Ontario, to fund additional nuclear-specific research at
any Canadian university, and to offer a course-based
Master of Nuclear Engineering program. Unique
features of the latter are the offering of courses on
weekends, and the option to take any course at any of
the participating universities. The appointment of
professors with specialist knowledge in the nuclear
field has enhanced the offering of undergraduate
courses, and has provided new consulting services to
industry. The increased profile of nuclear research at
the universities, and the publics awareness of
employment opportunities in the nuclear sector have
also raised student interest in these courses. However,
the majority of university graduates hired into the
nuclear industry gain their nuclear-specific knowledge
after becoming employed in the industry, typically via
company or other specialized training courses. Apart
from universities, community colleges are graduating
technicians and technologists, and are supporting
apprentice programs that produce trades people who
are in even greater demand than scientists and
engineers. With a few exceptions these graduates have
received little nuclear-specific education if any, and
will require specialist training after joining a particular
company. Industry training remains a critically
important part of ensuring that the required specialist
knowledge and skills are given to the men and women
working in the nuclear sector. Training that requires
such specialized equipment as full-scope replica
simulators, systems and components unique to a given
unit, tasks to be performed within the plant, are best
conducted by the particular company that is
responsible for the operation and maintenance of the
equipment.
2. UOITs Nuclear Degree Programs
A somewhat unique combination of events led to
the establishment of nuclear degree programs at the
University of Ontario Institute of Technology (UOIT)
[2]. These events included the aspects of plant aging
and demographics described above, the location of 12
CANDU units within a 25 km radius of UOIT, the
decision by the operator of these units (Ontario Power
Generation) to move their nuclear head office staff
from downtown Toronto to within the above
mentioned 25 km radius of UOIT, and the recognition
that the combination of demographics, plant life
extension and new build projects will require an
unprecedented number of nuclear science and
engineering graduates.
The first intake of students into the undergraduate
nuclear programs took place in September 2003, with
110 students selected from over 400 applicants. The
first graduates of these programs joined the work force
in 2007. The education program is centered around a
nuclear plant engineer. Courses in reactor physics,
thermal hydraulics, health and radiation physics,
materials and chemistry, and heavy emphasis on
design are included in the program [2]. To date, over
100 engineers have graduated and either entered the
work force or graduate school. Initial feedback
indicates that the nuclear specific education they
received has resulted in enabling them to become
more productive sooner than graduates of the more
typical science and engineering programs.
The Master of Nuclear Engineering program

22
consisting of a both research and a course based
option was initiated in the 2008-2009 academic year.
The Ph.D. in Nuclear Engineering program was
approved in 2010. The graduate program has
developed a set of advanced course topics that extend
the knowledge obtained in the undergraduate program
to include more detailed information useful for the
working environment. The course based program is
designed specifically to enhance the student skill set
for industry while the thesis based program is
designed for those interested in research and further
advanced studies.
In addition to the graduate program, a diploma
program has been initiated where students can take
only four courses and establish a specialty in such
areas as reactor systems, radiological applications, and
safety, licensing, and regulatory affairs.
While classroom lectures remain an essential part of
the education provided at UOIT, the use of computers
to enhance learning has been a key aspect of the
success of the programs and our graduates. The
CANDU reactors, by their unique characteristics,
required the use of computer monitoring as early as
the 1960s, and the use of computers was extended to
control and safety systems with each subsequent
power plant. As well, Canada was an early developer
of full-scope nuclear plant replica training simulators,
with the first such training tool for the Pickering A
units becoming operational in 1976. The unique
synergy of the use of computers in the CANDU
reactors and operator training programs made the
extensive use of computers in UOITs nuclear degree
programs a logical outcome.
3. Distance Learning for Nuclear Education
UOITs goal from the outset was to be both a
research-intensive and a student-centric institution. An
important additional mission of the university was to
investigate how strategies for college-university
transition could be facilitated through co-location and
the application of information and communication
technologies (ICT) to facilitate movement of students
from one level of education to the other. Consequently,
the university was designed as a fully laptop
university, the second in Canada, and the first in
Ontario. Planning included building both an online
infrastructure and a purpose-built physical
infrastructure, to facilitate learning both on and off
campus, from any location and at any time in lecture
halls, research laboratories, the library, and in public
areas such as study halls, cafeterias and on-campus
restaurants. The combination of ubiquitous Internet
access and the provision of standards-based laptop
hardware has allowed UOIT to establish a teaching
and learning environment that addresses total student
access to learning technologies, creating common
access to a unique web-centric learning environment.
The web-centric learning environment at UOIT can
be defined as the strategic integration of information
and communication technologies into all aspects of
the teaching and learning processes. The adoption of
digital technologies within higher education
institutions requires investment in both hardware and
software, and in the necessary support systems to
provide a learning environment where students
experience both a high-touch and a high-tech
educational experience, and where faculty are
supported in their use of both new network
technologies and new pedagogical approaches to
instruction [3, 4] .
Four support components and/or services systems
form the core of the web-centric environment at UOIT:
Learning Infrastructure, Online Educational Services,
the Mobile Learning Program, and Research Support.
Each component provides a foundational element for
the strategic use of information and communication
technology at UOIT. Rather than merely applying ICT
to enable distance learning or distributed education,
UOIT has developed a unique learning model,
whereby digital technologies enhance face-to-face
(F2F) education through the application of digital
technologies to enhance student engagement, both

23
online and in F2F classroom settings, in a coordinated
strategy. All learning spaces are equipped with smart
podiums to facilitate F2F instruction using laptop
computers, data projectors, DVD players, and Internet
access. As well, students in the Faculty of Energy
Systems and Nuclear Science have access to nuclear
plant simulation programs and other nuclear specific
software on their laptop computers. Research indicates
that students and faculty experience frustration with
the introduction of new technology without also
having the appropriate support structures in place [5].
To successfully merge traditional F2F lectures with
the development of an online learning community,
UOIT recognized that it is imperative to develop
strategic support systems that can provide a seamless
environment whereby students and faculty could
receive assistance when required. By empowering
both faculty and students with right-fit technologies,
the institutional focus was to enhance students
education/experiences. The UOIT institutional support
centers are depicted in Fig. 1 [6].
4. Education Requirements for Distance
Students
Distance education is changing with the times. It has
shifted from a somewhat solitary self study experience,
which Wedemeyer [7] described as the independence
of the student, to a rich dynamic teaching and learning
environment supported by web-based resources and
multiple interactions between individual students
and/or an instructor. Increasingly the format of
distance education has become online or web-based
learning enabling personal relationships and a stronger
sense of engagement as in fact the actual distance
between students and teachers has in some cases nearly
disappeared [8].

Fig. 1 Schematic of UOIT webcentric learning environment [6].

24

Simonson [9] described distance education as being
institution-based, formal education where a student is
separate from teacher, and interactions occur between
learner, resources and instructors. This sharing of
learning and experiences can occur via many different
asynchronous (at different times) and synchronous (at
the same time) means. With the advancement of
technological tools to be lower cost, easier to install
and use, and very little if any special equipment
required, communication between the student,
instructor and other students is commonplace in most
distance education courses. Asynchronous
communication is considered the most flexible form
of communicating [10], enabling contributions and
responses at any time by participants. This form of
communication has its drawbacks, and some students
and instructors view it as cumbersome and struggle to
cope with the slow pace of the dialogue. Synchronous
communication offers opportunities for immediate
dialogue between the learner and instructor or the
learner and other learners. Synchronous format
imposes a fixed time on the participants which,
depending on the number of time zones involved, can
be onerous. Synchronous communication can include
video, audio and text interactions which gives
participants familiar cues that can build rapport and
community more quickly and asynchronous dialogue.
Distance education has evolved, influenced by
technological tools and capabilities from being a
one-dimensional type of learning with the student
interacting solely with learning materials in a situation
of absent instruction and independent student or as
Holmberg [11] described it guided didactic
interaction to being a multidimensional learning
experience with interaction between student, materials,
instructor and other students.
Traditionally, the distance student has been a person
who could not attend school because they were
located in a different city or region. They
communicated with the school/lecturer via mail and
essentially self-taught the material. In todays world,
the nuclear student falls into several categories. There
are those that can and do attend school because they
live close enough and can afford transportation. There
are those students who live close enough, but due to
work commitments are not available for study at the
same time the classes are being delivered, and hence a
fixed structured program does not work. The third
group of students are those that are at a distance and
also have significant work commitments. As such,
distance education is no longer just for students that
live outside the region of the University. Universities
report that the majority (85%) of the students
enrolling in distance education courses are within
commuting distance of campus [12]. With the
technological tools available and used in courses, the
flexibility of place and time is very attractive to
on-campus students.
The majority of the students able to attend class are
those in the undergraduate programs. They need the
information being provided by the nuclear education
program to obtain sufficient skills to enter the work
force. Conversely, the majority of the students that
cannot attend class already have employment.
Those students employed have two main reasons
for wanting further education. Some are in the nuclear
industry and believe that additional education will
increase their skills to allow them to perform better at
their work and improve their chances for a successful
career. However, several are employed in industry
sectors other than nuclear such as the automotive or
manufacturing sectors and wish to retrain to find
employment in the nuclear sector. In either case, there
is a need to obtain advanced skills without interfering
with current employment duties.
From the students perspective, there are two main
requirements. One is obtaining useful knowledge
and the other is to be able to do so at reasonable times.
The first requirement means that textbooks alone
are insufficient. Students require that their learning
experience includes consultation with the professor
and preferably with those that have a significant

25
experience base related to the topic. The ability to
relate the learning topic directly to the work that the
students are doing is extremely important, especially
for those students that are working in the industry
already.
The second requirement means that the students
need to have access to a professor after typical work
hours and also the flexibility to obtain the information
outside of class time. Many of the students will work
at odd hours or perform shift work which means that
for one week they can attend lectures, and for the next
week they can only review the material at a different
time from the delivery. This second requirement
means that face to face communication at a fixed
time during the day, as per traditional education
methods, would fail to meet the students needs.
Essentially, a delivery method was needed that
allowed the students to attend the classes that they
were available for the live delivery, or synchronously,
and allowed students to review the class delivery and
material for those classes that occurred while they
were working yet still participating in an environment
where a Professor provided course information and
real life practical experience.
5. Distance Learning Tools
In choosing the appropriate learning tools,
professors can begin by looking at the course design
and outcomes. Identifying the teaching and learning
activities that will enable the students to achieve the
course outcomes and that can be completed by the
students independently versus those that are better
suited to an instructor facilitated session is the next
step. Consideration of the skills of the students and
their familiarity with various technology is important.
The selection of media using Holden & Westfalls
model takes into account the learning context, the
content, the intended outcome and the students [13].
There are many different ways to communicate
today and young people appear to thrive on using as
many as possible. Some techniques utilize devices
such as ipods, cell phones, or small portable readers
that can carry electronic books, videos, or even
education games. Familiarity with these tools and
ways of interacting does prepare students for a media
rich learning setting as well as a cultivating some
patience with technology tools when they need
adjusting. Other experiences they bring to distance
courses include using a CD contains the lecture notes,
software program, and a self-study guide. Thus the
student can work at any place and time as long as they
have a computer. While these techniques are useful,
they do not directly meet the requirements for the
distance student in nuclear education. Orientation time
along with several practice times is essential for
students to become comfortable with an online
learning environment so that the tools fade from focus
and the course learning takes centre stage.
The tools that are useful for meeting the
requirements of the nuclear distance student fall into
two categories: Software that records the lecture for
post lecture viewing and software that allows for
recording and interaction in a synchronous fashion
during the lecture. Software from both of these
categories are used at UOIT.
Camtasia and Mediasite are software programs that
can record lectures for later review. Both can produce
a web based video output/file of the lecture material,
the Professor speaking and the dialogue amongst the
class. Yet each program differs in its capabilities and
ease of use.
Camtasia is a very powerful program that can be
used to record almost any media on a computer. The
most common use at UOIT is to record Microsoft
PowerPoint presentations, but certainly Camtasia can
be used to record any window environment. Hence,
you can record an existing video and overlay mouse
movements and your own audio recording. The user
can also record simulation runs, calculations in
spreadsheets, use a whiteboard, or even record a video
camera. Hence the input into Camtasia is quite
universal.

26
A wonderful feature of Camtasia is the ability to
edit. Essentially, whatever input you produce can be
combined, spliced, edited in a production studio style
environment. This allows for mixed media to be
combined into one format to produce a cohesive
learning presentation to the student. The output can be
done in several formats as well such that it can be a
portable video usable on an ipod or as a web-based
video where the students cannot download the
material but must use the registered course
environment to view it. A typical example is shown in
Fig. 2 where the video and audio streams can be
spliced together and edited into a new output. An
example of the final output is shown in Fig. 3.
With all of the capability of Camtasia, there are also
some drawbacks. The process of producing a high
quality output is very time consuming. To give an
example, a 1.5 hour lecture recording of a PowerPoint
presentation with mouse movements and the professor
audio (no web cam) would require that the CPU be in
use for approximately 1 hour and generate a large file
of approximately 45 MB. It is possible to shorten the
time and the file size, but to do so, requires more
effort from the Professor.
As such, Camtasia is an extremely powerful tool
that requires a significant learning curve for a
Professor to use well and use with ease. Competing
software is available that will work faster and be
easier to use than Camtasia because they are
essentially point and click software. While easy to use,
the recorded output cannot be easily edited.
Mediasite is an alternative technology significantly
different from Camtasia. Note that Mediasite has
several features which will not be discussed in this
paper as the advanced features have not yet been
implemented at UOIT in the teaching environment.
Hence, this paper will discuss the basic features of
Mediasite in use. The main reason for the use of
limited features of Mediasite is that the majority of the
Professors and Sessional instructors that will perform
distance teaching are not comfortable yet with
advanced tools and require a simple point and click
type system so they can concentrate on teaching and
not the technology.

Fig. 2 Typical Cmtasia editing screen showing the recording window information, video stream, and audio stream for editing.

27

Fig. 3 Typical output of a Camtasia recording without video.

Mediasite, as currently used, records the screen
from the computer, the audio of the Professor, and the
video of the Professor. The auditorium is equipped
with a sound system so that the Professor is free to
walk around and talk to the students in class and to
have that conversation recorded. The auditorium is
also equipped with a video camera with zoom
capability to record the actions of the Professor. The
zoom capability is not fully used at this time but does
allow the Professor some movement around the room
and to prepare information on a whiteboard for
recording and presentation to the students.
The output of Mediasite is a web based format that
is stored on a server with controlled access by the
students. A typical catalog list is shown in Fig. 4
where the students need only click on the correct link
to review the lecture at a later time. Fig. 5 shows a
typical output of the Mediasite recording.
Unfortunately, while the video can be seen on the web,
capture to file appears as a black screen.
The tool used for interactive communication at
UOIT is Adobe Connect. A competitor technology
that provides similar capability is Elluminate that is
also in use for the UNENE Nuclear Education
programs.
Adobe Connect is a web based tool that creates a
shared environment for both the Professor and the
students. A typical Adobe Connect view is shown in
Fig. 6.
The Adobe Connect environment uses a common
server for all attendees through the internet. Aside
from a small add-on file, no software is required to
use Adobe Connect other than internet access. The
sessions are reserved on the server by the coordinators
at the University and the Professors can begin a virtual
class at any time as long as the number of users who
do not exceed our license limit.
Adobe Connect sessions can record several items as
shown in Fig. 6. There is an area where any file can be
uploaded and displayed. The audio output and webcam

28

Fig. 4 Section of catalog page for Mediasite.

Fig. 5 Typical output of Mediasite with the screen capture in the main view area and the video capture on the right.

output can also be displayed. From this perspective,
Adobe Connect is similar to Camtasia or MediaSite.
The differences are associated with other more
interactive features. The software will identify the
attendees in the virtual meeting and their status, allow
for chat type discussions amongst the students,
areas to record notes, polls, and virtual whiteboards.
The software is also flexible in that it can switch
between the various features such as a whiteboard and
a presentation relatively effortlessly.


29

Fig. 6 Typical output of Adobe Connect showing a whiteboard, video camera, chat session, poll section, and attendance status.

A key feature of the Adobe Connect software is the
ability to give the students various levels of control.
For example, you can assign a student presenter
status which will allow them to upload their own
presentation or to actual draw on the whiteboard
themselves.
Elluminate is a very similar software to Adobe
Connect. The layout of Elluminate is a bit more
friendly for certain features and allows better
recording control. A typical layout for Elluminate is
shown in Fig. 7.
6. Student/Professor Experiences
The student/professor experiences discussed here
have occurred over a period of 3 years. The majority of
the discussion is based upon course delivery by the
principal author and modified based upon feedback
from several other Professors. The Adobe Connect
software was used for 3rd year, 4th year, and graduate
students. The Elluminate software was used for
industrial students. Camtasia software was used for 3rd
year, 4th year, and graduate students. The Mediasite
software has been used for 4th year and graduate
students.
The third year students belong to the category of
students that can almost always attend class and thus
do not truly require distance learning. As well, they
are students very familiar with technology but not
with virtual learning environments. Camtasia files
were used to record base level material for each
lecture and to place it in the WebCT Vista program so
that students can view the base level material at their
leisure. The lectures were then conducted in a hybrid
format. Half of the lectures occurred face to face and
allowed the students to interact with the Professor
without technology. As such, stories of industrial
experience and additional information not found in the
base level presentations were often discussed.


30

Fig. 7 Typical layout of Elluminate type interactive environment.

The remainder of the lectures were taught using
Adobe Connect. This allowed the students to decide to
work from home or from school. Doing so increased
flexibility of the student work week as the on-line
session was usually chosen on at a time convenient for
the student schedule.
In general, the student experience was positive.
They strongly supported the use of the Camtasia files
although they wanted the files portable so that they
could take with them after the course was over. The
Adobe Connect sessions became positive although it
was very important to have a training session at the
beginning of the course so that each student was
aware and comfortable with the software. One of the
key problems observed was ensuring that the students
stayed focused on the class. The chat area becomes
very active and scrolls too fast with information that it
becomes distracting so it is imperative to train the
students to use the Chat area only for course
information.
An improvement in the use of distance technology
becomes apparent when you use a teaching assistant in
a tutorial setting. The students can then explore the
features of Adobe Connect and come to the virtual
class more prepared to participate.
The 4th year students experienced Camtasia files
and Mediasite recordings. In general, these students
come to class for open discussion with a slightly
higher absentee rate as they are looking for jobs or
performing more part-time work to pay for their
education. The Camtasia files are again desired to take
home for additional information after the course is
over. The use of Mediasite is accepted by the 4th year
students, especially for those courses taught in the

31
evening by Sessional lecturers.
Overall, the undergraduates accept distance learning
tools but are more interested in having the recorded
file to review at a later date and prefer face to face
contact.
Graduate students fall into two categories. Those
that are studying a thesis tend to come to class as their
research is also at the University. However, there is a
significant portion of the classes where the thesis
student attends from home. The decision is usually
based upon the other activities performed in that day.
Course based students are usually those that already
have a job. Approximately half of the course based
students are in the local area and could attend while
the other half work for companies where
transportation to the University would take more than
1 hour to come to class.
The graduate students prefer Mediasite. Essentially,
they are satisfied with the ability to review a recording
at a later time. The main reason for this is that most of
them have very demanding jobs and have difficulty in
attending class even in the evenings. Several of them
work in different cities and a recording available over
the internet is more than sufficient.
However, there is an exception when the student
needs to interact with the Professor. For example
when they need to give a presentation of their research
or they need to discuss the topic interactively. Under
these circumstances, Adobe Connect or Elluminate is
preferred.
Some additional benefits were observed during the
implementation of the distance learning program. In
the fall of 2009, Canada prepared for an H1N1
pandemic outbreak where expectations were that the
class attendance would significantly drop. Alternative
means of education and preparation for extra classes
were being considered. While the pandemic did not
occur in Canada, the availability of previously
recorded Camtasia files allowed for sufficient
flexibility that extra preparations were not necessary.
If a student did miss a class due to flu like symptoms,
the minimum required information was already
available and could be reviewed afterwards without
impacting course delivery. As well, if the Professor
was concerned, then class delivery could be done via
Adobe Connect from outside the classroom reducing
health risks associated with pandemic situations.
Additional benefits have become apparent over time
due to unforeseen weather conditions or closures of
the University due to facilities related issues or
emergencies. As such, the education was not impacted
in such scenarios due to the availability of the
software.
The drawbacks of the distance learning fall into
three categories. The first category is the inexperience
of the Professors with the software. The second relates
to delivery support from IT. The third relates to
training of the students.
With respect to the Professors, not all Professors are
sufficiently knowledgeable in modern information
technology nor do they have sufficient available time
to learn the new technology. The result is that
advanced software such as Camtasia and Adobe
Connect, while powerful, become significantly
underused and the full on-line experience is not
transferred to the students. The short term solution to
this problem is to use distance learning tools that do
not require significant Professor involvement until
additional confidence is gained.
Support from IT or a similar education support
group is essential. The initial problem is ensuring that
the University network and servers are sufficiently
robust that they can handle the extra bandwidth placed
on the system by the use of distance tools. Especially
when video streaming is employed, the network can
be significantly overloaded. While the University is
fully covered by either wired or wireless Lan, it is
apparent that certain locations do not work robustly
and transmission is often clipped resulting in a
frustrating experience to the students. Of interest to
note is that off-campus users do not appear to
experience this problem while on-campus students

32
almost always do and of course the problem is worse
with a wireless connection.
Once the connection issues are resolved, the main
concern is ensuring that there is sufficient support for
the storage and access of output files. Access is a
significant issue as the workload on the Professor
increases as they are required to create links between
the storage locations and the information stored in
WEbCT where the students can access the
information.
One remaining IT related issue is associated with
Mediasite. The interface system is not friendly to the
non-media user. Hence, dedicated staff for Mediasite
audio-visual control are currently required. This
increases the cost of the delivery of the education and
also makes use of advanced features more difficult as
the Professor essentially needs to instruct support staff.
Most of the IT related issues are expected to disappear
as the program matures and solutions are developed.
The remaining issue is perhaps the most difficult to
deal with. Student expectations are high with respect
to technology. Based upon their personal experience
with cell phones, ipods, and computers, they expect
that distance delivery should not be significantly
difficult. As such, any problems that do occur tend to
be very frustrating and only a small number of
problems can be tolerated before a lack of support
amongst the students occurs.
As well, while some students are quite computer
friendly, there are a few that have great difficulty with
computers and thus there is the added complication of
dealing with students at different levels of capability.
Hence, a training session on the technology is required
for successful implementation which usually
consumes one class.
7. Conclusions
Each of the software tested did perform within
their requirements and worked reasonably well. The
majority of the issues are more with training the
professor in the use of the software.
The following conclusions were obtained:
Each software can perform certain tasks in
distance learning.
The professor for the class must be comfortable
with the technology or the experience will suffer.
IT related support needs to be well established
before delivery to minimize problems. Acceptable
problems such as network outages are understood and
accepted by the students. Other than that, the students
expect the software to work immediately upon start of
the course.
Mediasite is excellent for guest lecturers as there
is minimal effort required to implement the session.
Adobe Connect or Elluminate is more useful for
an interactive environment.
Camtasia, while powerful, should be reserved for
the development of a library of learning objects due to
the time intensiveness associated with the software.
At this moment, UOIT is able to deliver a distance
learning program for nuclear education. The
effectiveness of the program is linked to the use of
multiple software to suit different requirements.
Future development of the distance learning
program will concentrate on increasing the number of
students interested in on-line education and using their
feedback to improve the performance of the delivery.
Eventually, it would be best to reduce the number of
software in use to a minimum cost effective set, but
realistically, it is expected that software will continue
to evolve over the next several years in both ease of
use and capability.
Acknowledgments
The authors wish to acknowledge Debbie Jeffrey,
Stuart Reid, and Brendan Calver, Imtiaz Malek, Mike
Dymarski, Dan Meneley, Ben Rouben George
Bereznai, Bill Muirhead, and Victor Snell for valuable
discussions and comments.
References
[1] M. Mathur, People challenges: The UNENE initiative and
World Nuclear University, in: CNS Annual Conference,

33
Toronto, Canada, 2004.
[2] G. Bereznai, Start-up of a nuclear engineering
undergraduate honours degree program in Canada, in:
14th Pacific Basis Nuclear Conference, Hawaii, USA,
2004, pp. 1-7.
[3] A.W. Bates, Managing Technological Change: Strategies
for College and University Leaders, San Jossey-Bass, San
Francisco, USA, 2000.
[4] D.R. Garrison, T. Anderson, E-learning in the 21st
Century: A Framework for Research and Practice.
Routledge Falmer, London, England, 2003.
[5] E. Vogel, W. Muirhead, Teaching and Learning in a
Laptop Nursing Program: Institutional and Pedagogical
Issues, in: M. Bullen, D.P. Janes, (Eds), Making the
Transition to E-learning: Strategies and Issues.
Information Science Publishing, Hershey, PA, 2006, pp.
84-103.
[6] G. Bereznai, W. Muirhead, The use of mobile learning
technology in the nuclear science and engineering
programs at the University of Ontario Institute of
Technology, in: European Nuclear Society NEST
Conference, Budapest, Hungary, May 4-8, 2008.
[7] C. Wedemeyer, Learning at the Backdoor, University of
Wisconsin Press, Madison Wisconsin, USA, 1981.
[8] B. Holmberg, The Sphere of distanced Education Theory
Revisited, ERIC Document Reproduction Service No.
ED386578, 1995.
[9] M. Simonson, Definition of the field, Quarterly Review
of Distance Education 4 (1) (2003) 7-8.
[10] G. Salmon, E-tivities: The Key to Active Online Learning,
Kogan Page Limited, London, UK, 2002.
[11] B. Holmberg, The Feasibility of a Theory of Teaching for
Distance Education and a Proposed Theory (ZIFF Papiere
60), ERIC Document Reproduction Service No.
ED290013, Institute fur Fernstudienforschung
Arbeitsbereich, Fern Universitt, Zentrales, Hagen, West
Germany, 1985.
[12] E. Allen, J. Seaman, Staying the Course, Online
Education in the United Stateds, 2008, Survey report,
Sloan Consortium, USA.
[13] J.T. Holden, G. Westfall, Instructional Media Selection
for Distance Learning: A Learning Environment
Approach, ERIC Document Reproduction Service No.
ED501248, 2008.


Aims and First Assessments of the French Hydrogen
Pathways Project HyFrance3
Alain Le Duigou
1
, Marie-Marguerite Qumr
2
, Pierre Marion
3
, Philippe Menanteau
4
, Pascal Houel
3
, Laure
Sinegre
5
, Lionel Nadau
5
, Aline Rastetter
6
, Aude Cuni
7
, Philippe Mulard
8
, Loc Antoine
9
and Thierry Alleau
10

1. I-Ts, CEA, Gif-Sur-Yvette Cedex, 91191, France
2. Dpartement EPI ECUELLES, EDF R&D, Moret Sur Loing Cedex, 77818, France
3. IFP Energies nouvelles, 1 et 4 avenue de Bois Prau, Rueil Malmaison Cedex, 92852, France
4. LEPII, Universit de Grenoble CNRS, Grenoble Cedex 9, BP 47, 38040, France
5. DRI, GDF SUEZ, 361, av du P
dt
WILSON, Saint Denis La Plaine Cedex, BP 33, 93211, France

6. ALPHEA, EURODEV CENTER, 4 rue Jules Verne, Eurozone, Forbach 57600, France
7. AIR LIQUIDE, 1 chemin de la porte des loges, Jouy-en-Josas, BP 126, 78354, France
8. TOTAL RAFFINAGE MARKETING, La Dfense 10, Paris la Dfense Cedex, 92069, France
9. ADEME, 20 avenue du Grsill, Angers Cedex 01, BP 90406, 49004, France
10. AFH2, 28 rue Saint Dominique, Paris 75007, France


Abstract: The HyFrance Group was originally formed in France to support the European project HyWays, by providing (former
projects HyFrance1 and HyFrance2) the French data and possible hydrogen pathways according to national specificities. HyFrance3
is a new project that focuses on the economic competitiveness of different steps of the hydrogen chain, from the production to end
usage, at the time horizon of 2030 in France. The project is coordinated by CEA with the other partners being: ADEME (co-funding),
AFH2, CNRS, IFP, Air Liquide, EdF, GdF Suez, TOTAL, ALPHEA. The project is divided into 4 sub-projects, that address present
and future French hydrogen industrial markets for chemical & refinery uses, the analysis of the interplay between wind energy
production and storage of hydrogen for different automotive requirements (refuelling stations, BtL plants, H
2
/NG mix), massive
hydrogen storage to balance various offer and demand characteristics, and the supply network (pipeline option competitiveness vs.
trucked in supply) to distribute hydrogen in a French region for automotive applications. Technical and economical issues, as well as
GHG emissions, are addressed.

Key words: HyFrance, hydrogen, industrial, techno-economy, storage, wind energy, infrastructure.

1. Introduction

1.1 Context
Hydrogen is currently attracting a lot of attention in
publications, since its use as an energy vector in fuel
cells for both mobile and stationary applications may
become a very lucrative market in the future

Corresponding author: Alain Le Duigou, Ph.D., research
fields: energetics, hydrogen and techno-economics. E-mail:
alain.le-duigou@cea.fr.
(20302050), and in projects, too [1-4]. Estimates of
the potential development for hydrogen as an energy
source from the past ten years show that automotive
market penetration rates by 2050 could reach as high
as 40% [1] and even 70% in some cases [2, 3]. This
roughly means having to double the current hydrogen
production and multiplying it by at least 5 in the most
industrialised regions such as Europe or the US which
counts about 1 private vehicle for every two
inhabitants. As a major R&D player in the field,
D
DAVID PUBLISHING
Aims and First Assessments of the French Hydrogen Pathways Project HyFrance3

35
France is greatly concerned by such radical change.
Between 2004 and 2007, two projects named
HyFrance1 and HyFrance2 [3] contributed to work out
a national hydrogen energy roadmap, as a French
application of the techno- and socio-economic
methods and tools used in the European project
HyWays [2], which aimed at developing a European
hydrogen energy roadmap. In this project, coordinated
by CEA, the partnership included 15 major national
experts from industry, public research, associations
and ministries (Research, Industry), while searching a
national consensus among these experts. HyFrance1
and HyFrance2 identified the key issues, challenges,
barriers and opportunities related to the development
of hydrogen energy and its applications in France, in
the medium and long term (2020-2030 and beyond, up
to 2050). French hydrogen chains were carried out,
using the techno-economic tool E3database [5], which
calculated the energy need, emissions and costs of the
chains. The French strategy for hydrogen production
was consistent with the orientations of the national
energy policy and resources, and the future modes of
massive production of free-CO
2
hydrogen using
nuclear energy, wind and biomass. The two
significant impacts of this work wer e the
establishment of a survey of hydrogen energy in
France, within the European harmonized framework
of HyWays, and the emergence of a national
collective techno-economic expertise on this topic.
1.2 Aims and Topics of HyFrance3
The HyFrance partnership decided to maintain this
common fruitful experience under a new collaborative
project, HyFrance3 [6], that mainly consists of the
players who already achieved the high level of
international visibility, recognition and acceptance
within the framework of the former projects
HyFrance1 and HyFrance2: ADEME (partner and
co-funding), AFH2, CNRS-LEPII, IFP Energies
nouvelles, Air Liquide, EdF, GdF Suez, TOTAL
Raffinage Marketing, ALPHEA, and the CEA as
coordinator. HyFrance3 has been launched to focus on
the economic competitiveness of different steps of the
hydrogen value chain that had not been, or
insufficiently, achieved in HyFrance1 and HyFrance2,
from the production to end usage use, at the time
horizon of 2030 in France. Then, in addition to the
technical and economic data that will be given,
HyFrance3 offers, as a continuation of HyFrance1 and
HyFrance2, the opportunity of a national debate on
hydrogen by gathering the major French public and
industrial research partners.
This paper describes the topics addressed by this 20
months duration project that has been launched at the
end of May 2009. The work is divided into 4
sub-projects, that cover both large scale systems (SP1,
SP3, SP4) and a smaller scale one (SP2):
SP1: present and future French hydrogen
industrial markets (merchant, captive and co-product),
for chemical and refinery uses depending on different
scenarios including the use of BtL;
SP2: technologies and costs of H
2
production from
wind power for different transport requirements
(refuelling stations, BtL plants, H
2
/NG mix) with
different grid configurations (isolated or connected
wind power plants) and H
2
storage volumes;
SP3: relevance, dimensions and costs of
centralized or decentralized hydrogen mass storage, to
balance various offer and demand characteristics;
SP4: large scale infrastructure to satisfy massive
demand vs. massive production or storage, by
dimensioning the supply network (pipeline option
competitiveness vs. trucked in supply) to distribute
hydrogen in a French region for automotive
applications.
Technical and economic issues, as well as GHG
emissions, are addressed.
2. Present and Future French Hydrogen
Industrial Markets
Similarly to the european and world markets,
hydrogen in France is mainly produced for ammonia

36
and refinery uses, both from fossil resources (captive
and merchant supply) and as chemical processes
co-product. In the short and medium term, given the
prospects of an increasing hydrogen demand
essentially for refinery use (desulfurisation, heavy
crude oil processing, environmental requests), a
progressive awareness of the limited fossil
hydrocarbon reserves as well as increasingly stringent
limitations of the emissions of greenhouse gases, low
carbon content hydrogen production processes such as
water electrolysis will appear relevant and long term
candidate for hydrogen production. HyFrance3 aims
to analyse the short and medium term hydrogen needs
for a chemical and refinery use, up to 2030.
A study has been carried out, based upon literature
data and a complete survey of the French producers and
consumers. The annual production of hydrogen in
France is about 920,000 tons per year. It appears that
by-product hydrogen share is ca. 60% (including steel
industry coke oven gas) of the total hydrogen
production, with a significant ratio being burnt, while
those quantities could be considered as potentially
available to supply car fleets for instance, but for a
transition phase only. Almost all hydrogen is produced
and consumed at the same location, and cross-border
trade is negligible. In addition, emerging markets such
as iron and steel industry could take benefit of a green
hydrogen use, as foreseen for instance by the ULCOS
(Ultra-Low CO
2
Steelmaking) project [7]; this could
represent very high supply requests, but rather in the
long term. Generally speaking, the French hydrogen
industrial market should reasonably increase in the
medium term. Considering the life cycle assessments of
hydrogen production via natural gas steam reforming [8]
and electrolysis [9], the CO
2
emissions associated to H
2

production are about 1% to 2% of the overall French
emissions, but they could only be lowered either
through expensive CO
2
capture and storage
technologies (still to be demonstrated at a large scale)
or by the use of nuclear or French electricity mix driven
electrolysis.
As regards hydrogen needs for the refining industry
(Fig. 1), two petroleum scenarios are analysed. On the
one hand, a trend-setting development scenario that
perpetuates the current energy demand and
technological progress foreseens together with an
increase in bio-fuel use (10%, including a second
generation allothermal bio-fuels ratio) hybridised cars
and electrical vehicles (7.5% of the total number of
cars in France). On the other hand, an
environmentally friendly development scenario
considers more ambitious hypothesis such as lower
fuel consumptions (-20%), a strong increase in
combined rail and road transport (20%), and 21% of
cars without internal combustion engine: 15% electric
vehicles (battery electric and hybrid), and 6%
hydrogen fuel cell vehicles which corresponds to the
HyWay slow scenario for France in 2030.
Calculations will use linear programming, to match
the different fuels market demand with refinery
scheme and crude oil supply, and therefore hydrogen
requests, as a function of the scenarios and hypothesis
chosen. Well-To-Weel BtL costs will be evaluated,
taking into account the interplay with SP2 (above)
when hydrogen is produced by electrolysis from wind
energy. Overall CO
2
emissions will be estimated, too.
3. Wind Energy Hydrogen Economics
Hydrogen production from wind energy often
appears as a promising way to increase the renewable
energy ratio in the energy mix, hydrogen being then
developed as an electrical energy storage medium, as
well as a very low CO
2
energy supply for a direct use

Fig. 1 Current block flow diagram of refinery.
Brut
R Atm
R.S.V.
D Atm Fuels
Gasoil
Kero
Essence
Naphtha
GPL
Slurry
C3-C4
H
2
H
2
H
2
H
2
H
2
H
2
H
2
H
2
H
2
S
H
2
S
H
2
HDS
HDS
Isom
Reformer
Alkyl
ETBE
PG
HDS
V.B.
Prtrait FCC
HDS
SMR
Claus
HDC
D
A
T
M
D
S
V
Brut
R Atm
R.S.V.
D Atm Fuels
Gasoil
Kero
Essence
Naphtha
GPL
Slurry
C3-C4
H
2
H
2
H
2
H
2
H
2
H
2
H
2
H
2
H
2
S
H
2
S
H
2
HDS
HDS
Isom
Reformer
Alkyl
ETBE
PG
HDS
V.B.
Prtrait FCC
HDS
SMR
Claus
HDC
D
A
T
M
D
S
V

37
in transports, together with an important energy
security of energy supply. Different layouts may be
considered: remote wind farms dedicated to hydrogen
production on the one hand, and wind power systems
connected to the grid, on the other hand, capable of
providing a continuous electrolyser operation and
possibly retailing electricity to the grid. Given the
shape of hydrogen demand, different storage
configurations may be necessary to guarantee
continuous supply: irregular or regular requirements
from refuelling stations, or continuous demand from a
chemical process, for instance.
This topic aims at evaluating hydrogen production
costs, given different types of demand, using wind
power plants, electricity network, electrolysers,
compressed hydrogen storage, and SOFC fuel cells to
sell electricity to the grid. The simulation tool
HOMER (NREL) is used to give the optimal
techno-economic layout for each case, according to
the components specifications and under the following
economical assumptions: 3 MW wind systems, IHT
S-556 electrolyser (760 Nm
3
H
2
/h), 23 m
3
Air Liquide
tube trailer storage (200 bar), 20 years lifetimes except
for intermittently used electrolysers (10 years), and a
6% discount rate. The different cases considered are
the supply of a hydrogen motorway service area in the
south of France (important H
2
demand peaks), of a
Hythane bus fleet (continuous supply of a binary
mixture of 20% Hydrogen and 80% natural gas), and
of a 2000 barrels per day BtL station. Unlike the
refuelling station and Hythane bus fleet cases, the last
one does not require any hydrogen storage capacity,
which is provided there by the biofuel plant, and then
considers a supply with the current.
The first simulations show a large hydrogen
production cost range, depending on the very different
characteristics of the demand (Fig. 2). The refuelling
station requires a 28 MW isolated wind farm and the
most important storage capacity. Hydrogen production
is then the most expensive, in some cases over 20 /kg,
but can be lowered by an appropriate overall system

Fig. 2 Decentralized H
2
production costs from wind power
for different end-uses.

of management. Hydrogen production cost is barely
cheaper in the case of a 3 MW isolated wind farm
capable of supplying a 100 buses fleet (12.5 kg/h)
because of H
2
storage. It is much different with the
BtL station that can be supplied without H
2
storage.
An isolated 48 MW wind farm could supply roughly
7% of a 2000 barrels per day BtL station, at an
average cost of 9 /kg.
All these costs can of course be noticeably lowered
when the system is connected to the grid, the lowest
range of 4-5 /kg being reached for the BtL station
with grid connection.
Anyway, the hydrogen production costs we have
obtained are higher than the costs mentioned in the
literature, especially for hydrogen obtained with usual
electrolysis and SMR production: between 3.4 and 5.3
/kg (given an electricity price of 66
/MWh-Roads2HyCom [10]), and between ca. 1.6
and 3 /kg for central and on-site SMR, respectively
(IEA [11], HyWays [2], Roads2HyCom [10]). Table 1
takes stock of the H
2
productions prices obtained
during the Roads2HyCom project. The lower
electrolysis costs correspond to the use of the
electricity grid, while the higher ones correspond to
the use of wind energy referring to Ref. [12] that gives
on the other hand a high cost value (ca. 18 /kg) when
hydrogen is produced by on-site electrolysis.
4. Hydrogen Mass Storage
The former project HyFrance2 promoted a French

Electricity
production
(wind, grid)
Electrolysis
hydrogen
production
Hydrogen
storage
Hydrogen end-uses:
H2 filling station,
HyThane filling station,
BtL production,
electricity storage
Different system configurations

Hydrogen end-uses:
H2 filling station,
BtL production,
electricity storage

Hydrogen end-uses:
H2 filling station,
BtL production,
electricity storage
Up to 20/kg
~ 9/kg*
(*4-5 /kg with grid connection)

Hydrogen end-uses:
H2 filling station,
HyThane
filling
station, BtL production,
electricity storage
Hydrogen end-uses:
H2 filling station,
HyThane
filling
station, BtL production,
electricity storage

38
Table 1 Roads2HyCom: summary of hydrogen production
costs.

Centralized
SMR
Centralized
electrolysis
On-site
electrolysis
Current
production cost
1.6 /kg 3.4-5.33 /kg 3.73-18.33 /kg
Assumptions: natural gas: 7.25 /GJ and electricity: 66 /MWh
(18.3 /GJ).

strategy for hydrogen production with CO
2
-free or
CO
2
-reduced emissions, i.e. mainly steam methane
reforming SMR (assuming the availability of
cost-effective techniques for CO
2
capture and storage),
conventional water electrolysis using the French
electricity mix (90% from non-fossil resources) and
regional biomass gasification (large resources
available). Whatever may be the future hydrogen
production methods, large scale use of hydrogen will
require massive storage comparable to natural gas to
balance supply and demand. This is specifically
relevant for the production of renewable hydrogen
from intermittent sources such as wind energy, as
described above. But storage remains essential to
connect large scale hydrogen continuous production,
as cited above, with seasonal demands. The mass
storage will interact with the hydrogen grid similar to
the buffering capabilities in the current natural gas
grid, and the locations could be very similar. But
technical requirements and strategy may strongly
differ: natural gas storage in France responds to both
winter/summer stationary use fluctuations and
strategic choices, whereas H
2
mass storage would be
linked to seasonal demand changes related to the
variations in automotive uses.
We evaluate hydrogen mass storage dimensioning
and costs connected with an infrastructure network to
be developed in a couple of French regions during the
next decades (up to 2050), as described below
(paragraph 5). In the literature, four underground
facilities are considered in which natural gas can be
stored under pressure: depleted oil fields, aquifers,
excavated rock mines, and salt caverns. Given the
French geological layout, we consider the latter as the
cheapest most convenient in principle. Thick salt
deposits are located in the south and east of France, and
salt caverns have the advantage of good performances
in terms of filling and emptying (cycling), working
volumes, reasonable depths and cushion gas needs. As
said below, both south-east French administrative
regions named Rhne-Alpes and PACA have been
selected for this study. Hydrogen quantities needed to
be stored to balance offer and demand in 2050 for
personal cars and light duty vehicles (hydrogen fuel cell
or internal combustion engine) have been evaluated.
Roughly at a few tenth of thousands tons of hydrogen
need to be stored, depending on the two scenarios of the
French project PROTEC H
2
[1] considered (paragraph
5). Filling and emptying rates are calculated too, which
may have a strong impact on the dimensioning. Fig. 3
considers a one year repetitive scenario that includes a
one month complete supply break (maintenance) in
November.
5. Hydrogen Infrastructure and Transport
in a French Region
Currently, existing hydrogen production, storage
and infrastructure is mostly used for the chemical and
refining industries, and as said above, even if nearly
1600km of hydrogen pipelines in Europe have been
identified, almost all hydrogen is produced and
consumed on-site. In the future, hydrogen use will be
spread all over the country, therefore, a distribution
infrastructure will be necessary. Its design and costs
will be of a primary importance for the global use and

Fig. 3 1 year hydrogen storage evolution (tons), filling and
emptying rates (tons per day)maintenance in November.
-1000
-800
-600
-400
-200
0
200
400
-5000
0
5000
10000
15000
20000
25000
30000
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
F
l
o
w

(

t

/

d
a
y

)
T
o
n
s
Months
16% penetration rate in 2050

39
acceptance of this energy carrier, case specific and
strongly dependent on the scenarios of demand,
essentially for transportation, delivery options and
geographic locations (production, storage, refuelling
stations).
Various projects such as HyWays [2] and
Roads2HyCom [4] evaluated the technological
developments and related costs of hydrogen
infrastructure for transport and distribution. HyFrance3
aims to analyse the hydrogen infrastructure
development in two French regions, selected on the
basis of criteria such as industrial and research
capacities, massive geological gas storage opportunities,
and initiatives in hydrogen projects and demonstrations.
South-East French administrative regions named
Rhne-Alpes and PACA have been selected, which
gather almost 20% of the French population.
Deployment scenarios given by the recent French
project PROTEC H
2
(calculated by the POLES code)
have been selected to characterize hydrogen demands
up to 2050: penetrations rates of 16% and 40% at that
time horizon for passenger cars and light duty vehicles,
together with somewhat lower penetration rate values,
less than 2%, in 2030. Fig. 4 gives the hydrogen filling
stations distribution in 2050 for a penetration rate of
16% in Rhne-Alpes and PACA regions.
The ECOTRANSHY code [13] will be used to
work out an economic model that can provide the
main optimal features (trucks or pipelines? where?
when?) of a hydrogen network originating from
large-scale facilities (production and mass storage)
and supplying a set of GPS-located delivery points
with given demands (refueling stations), at various
time steps from 2010 to 2050. The approach has the
advantage to mix in a unified framework economical
issues (estimation of capital and operational
expenditures), physical features pertaining to
hydrogen (trucks characteristics, hydraulics of
pipelines and compression stations), and geographical
constraints (location of sources and consumptions).
With regard to the supply, salt caverns mass storage

Fig. 4 Hydrogen filling stations distribution in 2050 for a
penetration rate of 16% in Rhne-Alpes and PACA regions.

locations inside the Rhne-Alpes and PACA regions
will be considered at first, as well as long distances
mass storage locations in the case of French regions
lacking of geological mass storage capacities.
Hydrogen transport between large scale production
and geological mass storage facilities will be
evaluated, too.
6. Conclusions
This project aims to assess different steps of the
hydrogen chain at the time horizon 2030 in France,
from the production to the final use. Industrial markets
(chemistry, refinery) should reasonably increase in the
medium term, mainly due to refinery use; steelmaking
could represent very high supply requests, but rather
in the long term. As regards hydrogen needs for the
refining industry, two petroleum scenarios are
analysed: a trend-setting one, and an
environmental one that considers a 6% penetration
rate of hydrogen fuel cell uses in vehicles in 2030.
The wind energy hydrogen production first
simulations show a large cost range: from 4-5 /kg
when the system is connected to the grid for a supply
with the current, up to over 20 /kg for an isolated
refuelling station. As regards hydrogen massive
underground storage, salt reservoirs located in the
south and east of France will be considered; hydrogen
quantities to be stored have been evaluated, roughly at

40
a few tenth of thousands tons to balance supply and
demand in the Rhne-Alpes and PACA regions.
Those two French administrative regions have been
selected to evaluate the 2 deployment scenarios given
by the recent French project PROTEC H
2
:
penetrations rates of 16% and 40% at the time horizon
2050, together with somewhat lower penetration rate
values, less than 2%, in 2030. ECOTRANSHY code
will be used to work out an economic model that can
provide the main optimal features of a hydrogen
network originating from large-scale facilities to
supply refuelling stations, at various time steps from
2010 to 2050. Large scale analysis developed in
HyFrance3 address practical topics that could lead to
detailed developments in future projects, to evaluate
for instance hydrogen scenarios that consider
intermittent energy sources, massive production and
hydrogen storage, and national transport and
distribution infrastructures.
Acknowledgments
The project HyFrance3 is funded by the French
ADEME, Agence de lEnvironnement et de la
Matrise de lnergie (Environment and Energy
Management Agency), contract number 0874C0009.
References
[1] E. Bellevrat, B. Chteau, A. Kitous (ENERDATA), The
role of Hydrogen in Long-Term Energy System: An
Updated Quantitative Analysis with the POLES Model,
in: IEW 2009, Venice, Italy, June 19, 2009.
[2] Ludwig-Blkow-Systemtechnik GmbH, HyWays,
Executive Summary of the European Hydrogen Energy
Roadmap and Action Plan [online], EC Contract N: SES6
502596, http://www.HyWays.de, June 2007.
[3] J.M. Agator, S. Avril, Towards a French hydrogen energy
roadmap: The HyFrance project, in: 16th World
Hydrogen Energy Conference (WHEC 16), Lyon, France,
2006.
[4] Roads2HyCom Hydrogen & Fuel Cell Database [online],
2007, http://www.roads2hy.com/.
[5] E3database-Life-Cycle Analyses and Well-to-Wheel
Analyses [online], http://www.e3database.com/.
[6] HyFrance3 Project, French Environment and Energy
Management Agency (ADEME), contract number
0874C0009, May 26, 2009.
[7] J.P. Birat, Addressing the climate change challenge: The
ULCOS breakthrough program, in: ISIJ 157th Meeting,
Tokyo, 2009.
[8] M. Mann, P. Spath, Life Cycle Assessment of Hydrogen
Production via Natural Gas Steam Reforming,
NREL/TP-570-27637, 2001.
[9] J. Ivy, Summary of Electrolytic Hydrogen Production,
Milestone completion report, NREL/MP-560-36734, 2004.
[10] Analysis of the current hydrogen cost structure,
Roads2HyCom Hydrogen & Fuel Cell Database:
Deliverable report D 2.4, 2009.
[11] Prospects for Hydrogen and Fuel Cells: Energy
Technology Analysis Report, IEA, 2005.
[12] J. Linnemann, R.S. Wilckens, Realistic costs of
wind-hydrogen vehicle fuel production, International
Journal of Hydrogen Energy 32 (2007) 1492-1499.
[13] J. Andre, A. Simonnet, J. Brac, G. Maisonnier, M.O. Sidi,
D. De Wolf, Least-cost development of hydrogen
transmission-distribution pipeline networks,
ECOTRANSHY, in: 17th World Hydrogen Energy
Congress (WHEC 17), Brisbane, Australia, 2008.


Experimental Study of Airflow-Mixture in HVAC Unit by
Using PIV
Yu Kamiji
1
, Atsuhiko Terada
1
and Hitoshi Sugiyama
2
1. Japan Atomic Energy Agency, Ibaraki-ken 311-1393, Japan
2. Graduate School of Engineering, Utsunomiya University, Utsunomiya 321-8585, Japan

Received: December 27, 2010 / Accepted: March 09, 2011 / Published: January 31, 2012.

Abstract: In this paper, experimental results are reported about, especially, effect of turbulence in airflow-mixture in HVAC (heating,
ventilating, air-conditioning) unit. A flow channel in this study has same characters as actual HVAC unit, such as bent rectangular
duct, a heater unit and a flow control door. The experiment was carried out to obtain verification data for analysis by using PIV
(particle image velocimetry) system. Moreover, temperature map in the channel was also obtained by using thermocouples. Reynolds
number is set to 12800. As a result of this study, representative velocity distributions and Reynolds stress distributions were obtained.
It was found that stress value is quite high (maximum 1.4) by preliminary turbulence and mixture in the duct. In addition, it was also
found that temperature diffusion is promoted by mixing around door.

Key words: Airflow-mixture, PIV, flow visualization, separation flow, Reynolds stress.

Nomenclature

D A side length of square in cross-section (= 75 mm)
L Averaged velocity
Re Reynolds number (U
b
D/)
N Number of instantaneous data (= 5000)
T Averaged temperature
T
b
Bulk temperature (inlet temperature = 21
o
C)
T
h
Temperature of heater (= 35
o
C)
T Difference of temperature heater and inlet
(T
h
- T
b
= 14
o
C)
U, V Velocity component for x,y-axis
U
b
Bulk velocity (averaged velocity at inlet = 2.58 m/s)
u,v Fluctuating velocity component for x,y-axis
u',v Fluctuating velocity intensity for x,y-axis
Dynamic viscosity
x,y Coordinates (shown in Fig. 2)
1. Introduction
Gas temperature control mechanisms by mixing of
different temperature fluid are used in various
engineering fields such as architectures, power plants
and automobiles. In particular focusing on a physical

Corresponding author: Yu Kamiji, M.Sc., research field:
flow visualization. E-mail: kamiji.yu@jaea.go.jp.
phenomenon of airflow-mixture, there are quite a few
cases which use it like cooling of gas turbine blade,
thermal striping in nuclear reactors and so on.
Japan Atomic Energy Agency (JAEA) has been
conducting R&D on heat utilizing system of HTGR
(high temperature gas-cooled reactor). Authors are
especially interested in HVAC (heating, ventilating,
air-conditioning) unit as one of the system. HVAC is a
kind of air conditioner and used in many places, so
that some researches on HVAC have been reported
[1, 2]. The unit generically has evaporators, heaters,
blower and air distributing mechanism. Now, we
specifically aim to develop smaller sized and higher
performance unit. To achieve these purposes, we have
been carrying out research on promotion of airflow
mixture for temperature homogenization in the unit.
To design and product something, lower cost and
shorter development time are very important as you
know. Accordingly, it is necessary to use numerical
analysis instead of experiment in designing. There are
some reports about analysis using commercial
D
DAVID PUBLISHING
Experimental Study of Airflow-Mixture in HVAC Unit by Using PIV

42
analysis code for HVAC [3, 4].
In this paper, numerical analysis was performed as
preliminary results. Heater was treated as porous
media and modeled by using the Angarisas method
[5]. This code is developed by Sugiyama et al [6-8].
The code is performed by using algebraic Reynolds
stress model and turbulent heat flux model. So far,
The code was applied to spiral flow [6], rectangular
duct with a sharp turn [7] and square duct with ribs [8],
and verified their results. However, it was not verified
about airflow-mixture. In addition, there are some
experimental results for simple mixing in planar shear
layer [9] and in T-junction [10]. But there are not
results for more complex mixing such as the flow in
HVAC unit. From the above background, we aim to
develop the code to design process gas reactors. This
paper reports the experimental results for
airflow-mixture with preliminary turbulence. The
experiment was carried out as the first step of code
validation and development. In addition to that,
preliminary analysis results are showed below, too.
Velocity fields are measured by using PIV, and
temperature fields are measured by thermocouples.
2. Experiment
2.1 Experimental Apparatus
Fig. 1 shows schematics of experimental apparatus.
A cross-section shape is 75 mm-square at inlet. There
are metal meshes for rectification. Also, this test
channel has heating unit and flow control door. This
heating unit consists of a folded tape heater
(Sakaguchi E.H VOC Corp.), two metallic meshes and
many glass beads (4 mm diameter). The tape heater is
set between the meshes. And the beads fill clearance
between heater and meshes up. The flow control door
separates inlet flow into heated flow and non-heated
flow, and can change their distributions of separated
flows to four degrees. In this paper, those degrees of
door opening are called from 1/4 (opening) to
4/4(opening) as shown in Fig. 2. This 1/4 means
separation rate of heater side flow. Besides, there are

Fig. 1 Schematic diagram of experimental system.

Fig. 2 PIV measuring area with door opening degree
(1/4-4/4).

no mixture in 4/4 case, so that results of 1/4, 2/4 and
3/4 case are showed in this paper. After the heater unit
and door, there is a 90-degree sharp turn to make
impinging jet. Internal flow of the channel is made by
a blower (Hitachi Industrial Equipments Systems Co.,
Ltd.) at the most downstream place. Flow rate is
measured by vortex flowmeter at upstream of the
blower. Ball valve is placed between flowmeter and
blower so as to control flow rate. As mentioned above,
PIV (Dantec Dynamics A/S, TR-PIV system [11])
was used in velocity measurement. A camera and
LASER are connected to workstation (DELL
Precision470, CPU: Intel Xeon 3.00 GHz, RAM: 1
Compressor Seeding generator
Camera
Blower
Flow
meter
Computer
LASER light sheet
Flow direction
Enlarged view of A
(characteristic area)
A
Heater
Flow control door
(variable opening degree)
(a) Near door area
(b) Mixture area
1/4
3/4
Heater
x
y
x
y
2/4
4/4
(Not measured)
Measured area
Inlet
Inlet
Outlet Outlet
Mixture
Distribution

43
GB, HD:1 TB) and synchronizer. LASER light sheet
was set at 37.5 mm height from bottom face of the
channel, which is central plane. Tracer particles are
made from water solution of glycerol, and supplied
from particle generators. Particle size is about 10
micrometer. The four generators are set at the most
upstream.
2.2 Experimental Methodology
2.2.1 Flow Visualization
Photos were used to visualize the flow field, which
was obtained as base picture for PIV.
2.2.2 Velocity
Table 1 shows main experimental conditions on
measurement of velocity. To obtain statistics of
velocity, 5000 instantaneous velocity vector maps (10
sec) are used. In this paper, average velocity
distributions and Reynolds stress distributions were
calculated. In the measurement of velocity, we tried
improving time resolution by shortening each interval
between instantaneous vectors. To be specific, we
used the camera which has 1280 by 1024 pixels
resolution as a maximum, and set it to 1280 by 640
pixels in the measurement for getting higher frame
rate, so that frame rate was increased from 300 Hz to
500 Hz as a max. As a result of this, each measured
areas, however, became smaller, time resolution was
increased. This reduced area was compensated to
increase measurement parts where interesting
phenomena were seemed to be found. In the
experiment, two parts were measured as shown in
Fig. 2. One part might include a separation at door
(called near door part), another part might include
mixture of two flows (called mixture part). The
LASER was double cavity and used as double pulse.
Moreover, frame straddling method was used, because
the method allowed catching particles on fast flow.
2.2.3 Temperature
Before measurement, a top panel of channel showed
by dotted square in Fig. 1 was exchanged for another
panel with holes for thermocouples. Then, temperature
Table 1 Experimental conditions of PIV.
Item Experimental condition
(1) Re number
12800 (Representative velocity (Ub):
2.58 m/s at inlet)
(2) Fluid Air, 21
o
C
(3) Camera
CMOS Camera, 1280 640 pix 8bit,
Frame rate: 500 Hz
(4) Light source
Nd: YAG LASER ( = 532 nm),
Double pulse (t = 100 s), 5 mJ/pulse
(5) Velocity vector
Adaptive correlation, Instantaneous
vectors: 1178 points 5000 images

of heater was set to 35
o
C. Measurement sequence is
below. First, insert thermocouples to holes in top
panel and wait one minute as time of heat transfer.
After that, measure temperature for two minutes. Then,
calculate average from measured temperature. There
are 749 holes for measurement on top panel. Also, the
measurement was carried out with attention for
interaction between each thermocouple. To be more
precise, each thermocouple was inserted 10mm apart.
The bulk temperature (T
b
) used for normalization was
calculated from the measured temperature at inlet.
3. Results and Discussion
3.1 Flow Visualization
The photos in Fig. 3 show flow visualization of
specific flow that is separation and mixture in the
channel.
Fig. 3a shows visualization at near door part in 3/4
case. There is separation of boundary layer from tip of
door and vortex street in downstream of the tip. 2/4
case has same inclination, but smaller size. That
difference is due to door opening.
Fig. 3b shows visualization at mixture part in 1/4
case. There is also vortex street from edge of heater. It
is caused by mixture of heat air and non-heated air.
3.2 Velocity
The velocity value is normalized by dividing average
velocity L [m/s] by bulk velocity U
b
[m/s].
Fig. 4 shows velocity vectors and velocity
distributions in near door part. It is found that there
are recirculation regions after the separation points in

44

Fig. 3 Flow visualization by LASER sheet and high-speed
camera.
case of 2/4 and 3/4 as you see Fig. 4c and enlarged
view of B. The recirculation region in 3/4 has about
10 mm thick. Through increasing area of the region,
main flow becomes narrower. Therefore, maximum
velocity of main flow reaches 3.6 times U
b
.
Additionally, the velocity of reverse flow in
recircuration region is same to U
b
. At the boundary of
main and reverse flow, there is large velocity
differencelarge velocity gradient. This causes the
instability of Kelvin Helmholtz. So, the vortex street
seen in Fig. 3a results from the instability. On the
other hand in case of 2/4, maximum velocity is 2.0
times U
b
. This is because separation in 2/4 is smaller
than in 3/4. This difference comes from inclination of
door toward main flow. This is evidenced in 1/4 case
that there is no separation and lower maximum
velocity 1.8 U
b
.
Fig. 5 shows velocity vectors and magnitude
distributions in mixture part. In case of 1/4 (Fig. 5a),
there is a region of large velocity gradient from heater
edge (displayed as area C). There is instable vortex
street by Kelvin Helmholtz instability as same as 2/4

Fig. 4 Velocity vectors (upper) and velocity distributions (lower) in near door area.
[mm]
50 100
0
0
50
[mm]
Vortex street

(a) 3/4 opening in near door area
(b) 1/4 opening in mixture area
[mm]
50
50 0
0
[mm]
Vortex street
Enlarged view of B Door
x x
x
y
y
[mm]
[mm] [mm]
[mm]
[mm]
B
Door
Heater
Wall
(a) 1/4 opening (b) 2/4 opening (c) 3/4 opening

45

Fig. 5 Velocity vectors (upper) and velocity distributions (lower) in mixture area.

and 3/4 of Near door area. So, it is found that the
region is boundary of heat air and non-heated air
clearly. Non-heated air flows out as a jet at the corner,
and then changes its direction while mixing with heat
air. Simultaneously, velocity arises to 2.4U
b
inside of
the corner. It was thought that mixing of heated flow
and non-heated flow contributed to increasing of
velocity This is also found in other 2/4 and 3/4 cases,
each maximum velocity is 2.4U
b
and 4.0U
b
. 2/4 case b
shows similar feature to 1/4 case, such as mixture area
and increasing inside velocity. Meanwhile, it looks
also different between 1/4 and 2/4, that velocity
gradient of 2/4 is smaller than of 1/4, which comes
from difference of each door opening. So flow rate for
heater side in 2/4 is larger than in 1/4, which leads
heater side flow to increase velocity. As a result of
this, velocity difference of heated and non-heated
reduces. 3/4 case c shows that velocity of heated flow
is increased, and moreover velocity of non-heated
flow is increased at higher rate because of acceleration
in near door part. Besides, it is found that some of
heated flow go to the recirculation region in near door
part (displayed as area D in Fig. 5c).
3.3 Reynolds Stress
Reynolds stress is calculated from fluctuating
velocity and standard deviation defined by formulae
below.
2
1
2 2
' '
) (
1
1
'
u u
u
N
u
N i
i

Figs. 6-8 show each Reynolds stress distributions.
Value in figures is normalized by dividing u'u', u'v'
and v'v' [(m/s)
2
] by U
b
2
[(m/s)
2
]. In near door part of
1/4 case, stress value is so low that they are showed
multiplied by 10
3
. Also, arrows showed in each figure
mean flow direction.
Fig. 6a shows normal stress of x-direction in 1/4. In
this figure, it is found that value in near door is so
small, about 0.005 because there is no separation.
Meanwhile in mixture part, they rise up to 0.1-0.2.
This mixture region corresponds to the position of the
vortex street or mixture interface mentioned above.
The velocity value momentarily changes by rotation
and ongoing of vortices, which results in rising of
fluctuating velocity or stress value. Fig. 6b shows
result of 2/4. Stress value is maximum 0.7 at tip of
door and it decreases to 0.2-0.3 as flow goes
downstream because of separation. Additionally in
mixture part, the value spreads to larger area than 1/4.
Fig. 6c shows result of 3/4. Maximum value is 1.4 at
tip of door. After the peak, once stress value goes
down to 0.5 rapidly, and then rises up to 0.8 again.
Moreover, separation area becomes wider. In mixture
part, stress value indicates high as flow goes downstream
(a) 1/4 opening
(b) 2/4 opening
(c) 3/4 opening
y
y
[mm]
[mm]
[mm]

[mm]
[mm]
x
x
x
C
D
Wall
Heater
Wall

46

Fig. 6 Reynolds stress (u'u') distributions in near door area (upper) and mixture area (lower).

Fig. 7 Analytical results (3/4 opening).

Fig. 8 Reynolds normal stress (v'v') distributions in near door area (upper) and mixture area (lower).
Heater Door
Wall
y
v'v'/U
b
2
10
3

v'v'/U
b
2
v'v'/U
b
2
v'v'/U
b
2
v'v'/U
b
2
v'v'/U
b
2
(a) 1/4 opening
(b) 2/4 opening
(c) 3/4 opening
x x
x

[mm]
[mm]
[mm]
[mm]
[mm]
y
Flow
direction
[mm]
y
y
(a) 1/4 opening (b) 2/4 opening
(c) 3/4 opening
[mm]
[mm]
[mm]
[mm]
x
x x
Heater
Door
Heater
Wall
Wall
4.5 5.5 6.5
0.5
0.5
1.5
2.5
8.2
7.3
8.2
7.3
6.5
5.7
4.9
4.1
3.2
2.4
1.6
0.8
4.9
5.7
X
2
/D=0.0
( u
3
2
/U
b
2
)10
2
X
1
/D
Heater
Door
4.5 5.5 6.5
0.5
0.5
1.5
2.5
X
1
/D
1.0
2.0
3.0
4.0
5.0
6.0
7.0
0.08
6.0
5.0
10.0
8.0 9.0
5.0
X
2
/D=0.0
(u
1
2
/U
b
2
)10
2
E
F
(a) Velocity vector

x/D

y/D
x/D x/D
y/D
y/D
(b) Reynolds stress (u'u')
distributions
(c) Reynolds stress(v'v')
distributions
(v'v'/U
b
2
) (u'u'/U
b
2
) (U
2
+V
2
)
1/2


47

which includes high velocity gradient area. The value
of 3/4 is quite higher than of 1/4 and 2/4. Considering
this difference, the increasing of stress value in
mixture part of 1/4 and 2/4 is caused by only mixture
of two flows, in contrast, the increasing of 3/4 is
caused by remaining turbulence in near door part as
well as the mixture. Fig. 7b shows preliminary
analytical result of normal stress of x-direction in 3/4
case. Comparing this with experimental result,
analysis predicts each maximum value at tip of door
as well as experiment. However, result of analysis is
different from experiment in three points. First, the
value decreases gently and after that, in mixture part,
increases little bit(displayed as area E). Second,
isoline is distorted around the heater (displayed as area
F). Third, the stress value is about ten times smaller
than experimental value. These may come from
inaccuracy of heater modeling. Because flow direction
is aligned vertical to heater (shown in Fig. 7a), in spite
that it is not aligned in experimental results. Therefore,
flow goes to outer wall, as a result of this, nothing go
to recirculation area in near door directly. This leads
decreasing of momentum in recirculation area and
influence on production of turbulence.
Fig. 8 shows normal stress for y-axis (v'v'). Each
figure is similar to the normal stress for x-axis (u'u'),
but the value is quite smaller. Although u'u' is
increasing as flow goes to downstream, v'v' is not. In
fact, v'v' has maximum value inside jet and stress
value is decreasing to downstream. Analytical result
of v'v' shown in Fig. 7c qualitatively conforms to
experimental result.
Fig. 9 shows shear stress (u'v'). It also indicates
high value at tip of the door and interface of flows as
other stress do.
3.4 Temperature
Fig. 10 shows measured temperature maps of
1/4-3/4. Temperature in downstream of 90-degree turn
naturally gets higher as the door opening degree gets
larger. Also, temperature of non-heated flow in 3/4 is
higher than 1/4 and 2/4. It was considered that a part
of heated flow goes to separation area and mixes with
main flow in advance. This phenomenon results in
getting activation recirculation area around door, and
mixes with main flow, which results in activation of
temperature diffusion.
4. Conclusions
This paper reported experimental results in HVAC
unit. Velocity, Reynolds stress and temperature ,which
were useful database, were obtained. Three patterns of

Fig. 9 Reynolds shear stress (u'v') distributions in near door area (upper) and mixture area (lower).
Heater Door
Wall
y
y
u'v'/U
b
2
10
3

u'v'/U
b
2
u'v'/U
b
2
u'v'/U
b
2
u'v'/U
b
2
u'v'/U
b
2
(a) 1/4 opening
(b) 2/4 opening (c) 3/4 opening
x
x x
[mm] [mm] [mm]
[mm]
[mm]

48

Fig. 10 Temperature distributions measured by thermocouples.

door opening were examined focusing airflow-mixture.
In case of 3/4, there is quite high normal stress for
x-axis up to 1.4 because of synergy between
turbulence around door and mixture of two flows.
Furthermore, temperature obviously becomes higher
corresponding to door opening. Particularly in 3/4,
heated flow partially goes to near door part, so that
mixture starts before mixture part and temperature
diffusion becomes active. On the other hands,
preliminary analysis was performed, too. Analytical
results were different from experimental. This is
caused by inaccurate heater setting in analysis. We
need to develop modeling of heater as a next step.
References
[1] J. Wen, T.F. Smith, Development and validation of online
parameter estimation for HVAC systems, Trans. ASME J.
Solar Energy Eng. 125 (2003) 324-330
[2] S.D. Wolin, N.L. Ryder, F. Leprince, J.A. Milke, F.W.
Mower, J.L. Torero, Measurements of smoke
characteristics in HVAC ducts, Fire Technology 37
(2001) 363-395.
[3] L. Bennett, C.W.S. Dixon, S. Watkins, Modelling and
testing of air flow in a HVAC module, in: SAE 2002
World Congress & Exhibition, Detroit, MI, USA, Mar.,
2002.
[4] N.H. Cho, M.R. Kim, Numerical investigation of fluid
flow in an automotive HVAC module, SAE Paper
971778 (1997) 61-67.
[5] D. Angirasa, Forced convection heat transfer in metallic
fibrous materials, Trans. ASME J. Heat Transfer 124
(2002) 13-21.
[6] H. Sugiyama,Y. Sasaki, Numerical analysis of turbulent
flow in a helically curved pipe, Transaction of the Japan
Society of Mechanical Engineers, Series B 73 (2007)
2099-2106. (in Japanese)
[7] H. Sugiyama, T. Tanaka, D. Hitomi, Numerical analysis
of turbulent separation flow in a rectangular duct with
sharp 180-Degree turn, Transaction of Society of
Automotive Engineers of Japan 38 (2007) 181-187. (in
Japanese)
[8] Y. Okagaki, H. Sugiyama, N. Kato, Numerical analysis
of turbulent flow with heat transfer in a square duct with
45 degree ribs, Transaction of Society of Automotive
Engineers of Japan 41 (2010) 709-715. (in Japanese)
[9] Y. Mizuno, H. Asano, K. Tada, M. Hirota, H. Nakayama, S.
Hirayama, Turbulent mixing of hot and cold air in HVAC
unit for automobile (thermal mixing in planar shear layer),
Transaction of the Japan Society of Mechanical Engineers,
Series B 71 (2005) 215-222. (in Japanese)
[10] H. Asano, K. Tada, M. Hirota, H. Nakayama, T. Asano, S.
Hirayama, Turbulent mixing of hot and cold airflows in
automobile HVAC unit characteristics of velocity and
temperature fields in mixing T-junction, Transaction of
the Japan Society of Mechanical Engineers, Series B 71
(2005) 2970-2978. (in Japanese)
[11] DANTEC DYNAMICS Website,
http://www.dantecdynamics.com/.

(b) 2/4 opening (c) 3/4 opening (a) 1/4 opening
x
x x
y
y y
[mm]
[mm] [mm] [mm]
[mm]
[mm]
Heater
Door
H
2
Milagros Rey Porto
1
, Mnica Aguado
2, 3
, Raquel Garde
2
, Gabriel Garca
2
and Trinidad Carretero
1

1. Gas Natural SDG S.A., Plaa del Gas 1, Barcelona 08003, Spain
2. Renewable Energy Grid Integration Department, National Renewable Energy Centre, Sarriguren 31621, Spain
3. Electrical and Electronic Engineering Department, Public University of Navarra, Pamplona 31006, Spain

Received: October 26, 2010 / Accepted: April 29, 2011 / Published: January 31, 2012.

Abstract: A hydrogen production and conversion plant from wind power was installed in the Sotavento wind farm by Gas Natural and
the Galician Government. This facility is the highest electrolysis power installed at the European level. It consists of an electrolyser of
300 kW, a piston compressor, a 1,725 Nm
3
H
2
storage system and an engine of 55 kW. This pilot plant is being operated by Natural Gas
in order to extrapolate its behavior to that of an industrial facility capable of managing all the production of Sotavento wind farm
following different strategies: balancing, peak-shaving and repowering. In this paper, preliminary results at the facility are presented.
The aim of these first tests has been to describe the operation of equipment under operating conditions required in the management of
wind power production, in order to understand the behavior of the different equipment and try to make them suitable for this type of
applications. This paper shows the difficulty of operating these systems under the wind power requirements.

Key words: H
2
, wind power, storage system, electrolysis, pilot plant.

1. Introduction

For decades, a series of progressive decrease in the
reserves of fossil fuels and environmental problems
associated with burning them have promoted the search
for alternative energy, including renewable energies.
The use of renewable energies offers numerous
advantages. The main ones include:
They are inexhaustible;
Both production and consumption of energy are
major emitters of greenhouse gases. The use of
renewable energy allows producing a low
environmental impact;
At country level, they foster the decline of foreign
energy dependence by providing a diversification of
energy sources;
Contribute to improve and modernize the
industrial base, the generation of employment and,
therefore, regional development;
Distributed Generation Systems allow the point of

Corresponding author: Mnica Aguado, Ph.D., research
fields: renewable energies and grid integration. E-mail:
maguado@cener.com.
generation being close to the consumption, and therefore
to reduce electricity losses produced in transport and the
necessity of transportation infrastructure.
Faced with all these advantages, the renewable
energies have three major drawbacks:
The great variability of renewable resources;
The difficulty of the prediction;
The difficulty of adjusting production to demand.
The renewable energy with greatest potential is wind
power. Regarding the wind energy production, the
countries that have increased more their power during
the year 2009 were US and China. Worldwide, wind
power capacity grew 31%, up 37.5 GW to 157.9 GW in
2009. China doubled its capacity from 12.1 to 25.1 GW,
but the U.S. still installed nearly 10 GW in 2009. That
increases its total ability to produce wind power by
39%, to 35 GW.
In Europe, the growth of wind power is strong and
the new installation of 8.9 GW has raised European
wind capacity to 66 GW in 2009. Germany is the
second country with highest installed capacity with
25.8 GW and the first at European level. Spain is the
D
DAVID PUBLISHING
H
2

50
fourth and the second at European level, with 19.1 GW.
As far as Spain is concerned, wind power represents
79% of the total installed renewable power. Spain had
more than 16,700 MW installed at the beginning of
2009, whose production represented 11.5% coverage
of electricity demand.
In areas with growing wind penetration, like Spain,
the renewable energy landscape has two problems: first,
economic, and the second operational.
Regarding the economic issue, the electricity
generation with renewable sources in Spain has a
special economic regime, as laid down in Royal Decree
6/2009. Nowadays, considering the economic context
of international crisis, the economic impact of the
renewable energy support, at state level, is very
important. Moreover, there is a tariff deficit in Spain.
That is the difference between the revenue from the
regulated tariffs set by the Administration and what
consumers pay for their supplies and regulated access
charges that are set in the liberalized market and the
real costs associated to those rates. This shortage is
causing serious problems, threatening the sustainability
of the system itself and damaging the security and
affordability of the investments required for the supply
of electricity in the quality and safety standards
demanded by the Spanish society.
Therefore, it is indispensable to optimize
economically the management of renewable generation,
in order to achieve this profitability independently of
the bonus given by the government.
On the other hand, in areas with growing wind
penetration, the intermittency of renewable energy
sources imposes additional requirements on system
operation and management. It is assumed that wind
power fluctuations can be absorbed by the grid for the
most part with wind power penetrations of less than 20
percent of maximum load [1]. In areas with more
penetration the variability of the wind resource requires:
Additional requirements in the operation and
management system;
The availability of combined cycles to ensure the
power supply;
Strengthening weak networks, very expensive,
and reserve capacity.
Sometimes when wind power generation
exceeds the limits that the network can safely absorb,
the windfarms must be disconnected. The consequence
is that this clean energy is lost.
Given the magnitude of wind power installed and the
expected growth in the coming years, it is very
important to mitigate and correct the weaknesses of
wind power. It becomes imperative to manage the
energy produced into the grid by wind turbines.
Therefore, it is necessary to research and develop
technologies to correct the variability of the resource
and adjust the production to the demand.
These technological developments are focused on
the prediction and storage of wind power delivered to
the grid.
The resolution of the problem of electricity storage
on a large scale is one of the most important
technological challenges to overcome in the field of
electrical engineering. Currently there are various
storage systems based on the amount of stored energy
and storage time needed [2].
Energy storage can solve most problems associated
with wind power generation, specifically:
The energy generated by wind farms is allowed to
be managed, improving its selling price in the pool [3];
Wind energy production is possible to be adjusted
to commitments as well as get a guarantee of wind
production at peak electricity demand;
The penetration level of wind power is allowed to
be increased, once it becomes manageable. Using these
systems the installed capacity could be increased in
resource-rich sites: during times of high wind, wind
power is stored, so the limit of the network is never
reached and electricity is injected into the grid in times
of low wind production [4-6];
It allows implementing wind farms in remote
locations but with high potential, reducing
transmission costs.
H
2

51
2. Hydrogen as Storage System
This article presents an installation that utilizes H
2

technologies as energy storage system. Hydrogen was
chosen, due to its versatility in both energy and power
capacity [7].
Hydrogen is an energy carrier, in other words it is no
free in nature and must be obtained from a primary
source, although there are numerous sources and
hydrogen production technologies.
Thus, hydrogen is one of the energy carriers with
greater environmental benefits because its combustion
produces no emissions and its production will have a
greater or lesser environmental impact in terms of the
energy that is used to generate it.
Therefore, if obtained through renewable energy, the
environmental impact it causes is minimal. It is for this
reason that today there is great interest in developing
projects that use hydrogen generated from renewable
energy as energy storage.
The concept of generating H
2
from wind born in
1923, when Haldane [8] proposed the production of H
2

and O
2
from water using this type of energy.
Back in the 1930s Hoffman introduced the use of
high-power wind turbines to generate H
2
[9]. From the
1970s the electrolysis using solar power is studied and
the H
2
generation from wind energy is considered too
[10-13].

The development of such facilities in each country is
conditioned by the quality of the transport and
distribution grid, the existing legislation in terms of
renewable energy and management and the percentage of
renewable energies in the electric generation mix [14].
Currently, there are two types of systems that use H
2

as energy storage of wind power. The former are island
energy systems, whose primary objective is to satisfy
the demands of a particular community with an island
configuration of the grid [15-20].
The others are integrated systems in the electrical grid,
which aim to manage and optimize energy production
from a wind farm, improving their capacity factor [13, 21,
22]. Both approaches must be addressed from different
perspectives, because the objectives, and therefore the
operation mode of the facility, are different.
Another objective of such facilities is the generation
of H
2
as fuel using renewable sources instead of
conventional natural gas reforming.
However, the installation presented here is one of the
largest in global capacity, where hydrogen is used in
the energy management of the production of a wind
farm. Therefore, this facility can be used as a platform
for tests development, obtaining data representative of
industrial-scale operation of such facilities.
3. Description of the Facility
The facility is located in the Sotavento wind farm in
the municipality of Xermade (Lugo), at its border with
the municipality of Monfero (A Corua). This has a
rated capacity of 17.56 MW with 24 wind turbines of 5
different technologies.
This project was born thanks to a framework
agreement between Gas Natural and the Department of
Innovation, Industry and Trade of the Xunta de Galicia
for the development of renewable energy. In 2006, Gas
Natural held the main investment buying and installing
all equipment. In 2008 an agreement with the National
Renewable Energy Center (CENER) was signed by Gas
Natural to study, analyze and characterize the plant.
The facility uses the surplus electricity generated by
the wind farm, in other words the excess of energy
produced over expected, to generate hydrogen with an
electrolyser. This breaks down water using electricity
(electrolysis) in four stacks of electrolytic cells,
generating separate H
2
and O
2
.
Fig. 1 shows the scheme of the plant with the main
equipments of this installation: electrolyser,
compressor, H
2
storage and H
2
engine and a picture of
the real systems.
The O
2
that is not going to be used in this process is
vented to the air and the H
2
, produced at a rate of 60
Nm
3
/h and a pressure of 10 bar, goes through a process
of purification and drying to obtain purity higher than
99.99%.
H
2

52

electrolyser
engine
compressor
storage

Fig. 1 Scheme and picture of the H
2
facility.

To increase storage capacity, the H
2
generated is
compressed to 200 bar in two compressor groups
admitting up to 61.8 Nm
3
/h at 4 bar.
The storage system at 200 bar is composed by 7
blocks of 28 bottles each, with a maximum capacity of
1,725 Nm
3
. These blocks are interconnected forming
two groups of H
2
storage, with the possibility of
isolation of each group.
The H
2
stored can be used in a 55 kW engine in case
of energy deficit, that is, if the amount of energy
produced by wind turbines in the wind farm is less than
the expected one.
The engine has a consumption of up to 70 Nm
3
/h of
H
2
at a pressure of 25-60 mbar H
2
. The compressed H
2

at 200 bar is decompressed in a first stage to 14 bar and
a second until the suction pressure of the engine.
The electrolyser in Hydrogenics is mounted in a
container consisting of three separate rooms: the
process plant, control room and electric power and
ancillary items room. In Fig. 2, we show a scheme of
the main components of the electrolyser module.
There are four cell stacks in the process plant. These
consist of a series of interconnected electrolytic cells.
Each contains the cathode, the anode and an inorganic
ion-exchange membrane (patented technology IMET).
The membrane prevents contact and reaction between
O
2
and H
2
produced in the electrodes. As electrolyte a
basic aqueous solution of KOH 30% is used. The
operation pressure is 10 bar.
The control room contains the electrical board, two
closets EPS (Electrolyser Power Supply) and the
control system of the unit.
The electrolyser power system comprises four AC to
DC converters. Those four converters are grouped two
to two and are fed by two lines.
All components of the compression plant are located
inside a soundproof booth for outdoor installation, with
ventilation and heating system. The piston compressor
comprises two groups from Bauer Model HFS
15.4-13-DUO II (Fig. 3).
The motor-generator from Continental Energy
Systems with a rated power of 55 kW is a diesel
engine genset rebuilt for using hydrogen as fuel. It is
mounted on a rigid base structure and has a radiator
for heat dissipation, gas pipeline and control system.
The H
2
engine consumes a maximum rate of 70 Nm
3

H
2
/h. The whole equipment is in a soundproof
container (Fig. 4).
Each system has its own control system. In this
project we have developed an integrated
communications system. We use MODBUS as
communication protocol and integrate signals from the
engine, compressor, electrolyser and network into a
common server.
H
2

53

Fig. 2 Operating diagram of the electrolyser.

Fig. 3 Picture of the compressor.

Fig. 4 Picture of the motor-generator group.
The control system has human interface software
that allows the sending of orders by the operator and
displaying the status of all equipment and variables
involved in the operation of the facility. In addition,
there is an acquisition of historical data (15 seconds) of
each of the equipments.
A major feature of this facility is the absence of any
other energy storage system in addition to H
2
. This
facility does not use redundant elements (batteries, sea
water desalination systems or H
2
filling stations) to
ensure system availability and better energy
management. This makes that it behaves like a pure
energy storage system using hydrogen and therefore
the behaviour of its components as though acting
without any redundant storage system can be analyzed.
In this sense, this installation will serve as a test bench
to analyze and study current technology only working
under conditions of H
2
as energy storage system.
4. Preliminary Results and Discussion
During the first month of operation of the facility, it
has been necessary to conduct a series of modifications
and adaptations, in order to integrate the operation of
each of the equipments to enable the work of the whole
plant.
The electrolyser is a system designed to meet
demands for continuous hydrogen production, as it has
Gas and Electrolyte
Gas to the atmosphere (O
2
)
Gas to driers (H
2
)
Coalescence
filter
Gas separator
Rinser
Cooler
Electrolyte
Gas
Water
supply
H
2

54
been traditionally used to supply hydrogen to users
who used it as a raw material in their manufacturing
processes. Its operating strategy is based, therefore, on
meeting demand at any time you ask, keeping constant
process pressure (about 9.5 bar).
Therefore, if more hydrogen is consumed
downstream than is produced, the electrolyser
increases its production to avoid a pressure drop, and
vice versa, within the limits of production (100% -40%
of each cellstack). Between the electrolyser and
compressor there is a buffer to stabilize the pressure.
The electrolyser can operate each of the stacks
independently, being activated or deactivated by the
user.
However, if the objective is the management of
renewable energy, operational strategy should be based
upon the energy available for consumption and storage
which defines the flow of hydrogen produced.
Today, as the electrolyser control is defined, energy
orders to consume have to be provided in terms of the
compressor suction pressure. The suction flow of the
compressor depends on the compressor suction
pressure and the operation of one or both compressor
groups. The higher suction pressure and number of
compressor groups are in operation, the more
compressed flow, and due to the electrolyser control,
the more production flow.
The electrolyser flow production implies energy
consumption. Therefore, there is a relationship
between the compressor suction pressure and the
energy consumed in the electrolyser, i.e. a relationship
between the compressor suction pressure and the
energy available for storage in the wind farm.
However, this relationship P
suction_compressor
/E
available
storage
is not direct since the compressor suction flow,
besides depending on suction pressure, also depends on
the upstream pressure that means, on the hydrogen
stored in the high-pressure bottles. Therefore, in these
early tests we want to characterize the relationship
between P
suction_compressor
/E
available storage
.
In order to operate from the control over the
compressor suction pressure, in the facility, it was
necessary to install a pressure regulator acted from the
control, in order to avoid the necessity of being on site
to operate the installation.
According to its control, the electrolyser has to
maintain its downstream pressure in the range 7.5-9.5
bar to meet a demand for hydrogen.
In Fig. 5, the operation at minimum production of
the electrolyser is shown. The four cellstacks and two
compressor groups are working at minimum
production, with compressor suction pressure 2 bar. At
this pressure the production of electrolyser is constant
(around 24 Nm
3
/h, each cellstack is producing 6
Nm
3
/h), maintaining the pressure in the electrolyser
around 9.5 bar. When the inlet pressure of the
compressor increases, the production follows the same
tendence in order to maintain the electrolyser pressure
constant.
The operation of the electrolyser-compressor set can
give the following cases:
(1) Case 1: compressor H
2
suction flow >
electrolyser H
2
production flow
This situation will occur in the event that the
maximum production of the electrolyser with the
number of stacks working is lower than the compressor
suction flow. In this case there will come a time when
the pressure in the buffer between electrolyser and
compressor drops below 7.5 bar.
As the pressure decreases downstream from the
electrolyser, to prevent the pressure in the process from
falling below 7.5 bar, the electrolyser is isolated by

0
20
40
60
80
100
120
140
160
180
200
1700 1750 1800 1850 1900 1950
Prod.total (Nm3/h) Pcompr_inlet (bar) Pelect. (bar)
P
r
o
d
u
c
t
i
o
n
(
N
m
3
/
h
)
;
P
c
o
m
p
_
i
n
l
e
t
(
b
a
r
*
1
0
)
P
e
l
e
c
t
r
o
l
.
(
b
a
r
*
1
0
0
)

Fig. 5 Representation of electrolyser operation.
H
2

55
closing the supply user valve, generating H
2
at minimal
production (40% of nominal output) and keeping its
operating pressure at 10 bar.
The compressor, which will continue working, will
compress the hydrogen in the buffer until the inlet
pressure is less than 2 bar, in which case it will stop
because of low pressure suction. To continue working,
if there is hydrogen in the buffer, a manual reset of the
compressor is required. Therefore, the automatic
operation of the facility is very difficult.
(2) Case 2: electrolyser H
2
production flow >
compressor H
2
suction flow
This situation will occur in the event that the
minimum production of the electrolyser with the
number of active stacks is greater than the intake flow
of the compressor.
In this case the downstream pressure of the
electrolyser will rise above 9.5 bar, then the
electrolyser detects no H
2
consumption because the
downstream pressure does not decrease, closes the
supply user valve and produces a minimum flow of
40% of its nominal production, venting the H
2

produced, generating energy losses in the installation.
(3) Case 3: electrolyser H
2
production flow =
compressor H
2
suction flow
Whenever possible, i.e., if active stacks enable the
electrolyser to produce a hydrogen flow equal to the
compressor consumption, the electrolyser works
properly maintaining the process pressure in the range
7.5-9.5 bar. That is, if the downstream pressure is
within the range indicated, the electrolyser will
modulate its output to meet demand requested by the
compressor at all times.
To prevent the electrolyser from venting situations,
involving a loss of hydrogen and therefore electricity,
the pressure downstream of the electrolyser has to be
maintained within this range (case 3). Achieving this,
combinations of stacks and compressor groups sets
have to be defined in order to tight the production flow
with the compression flow to optimize the operation of
the system, achieving maximum efficiency.
Therefore, the existence of two compressor groups
and four cellstacks, that can work independently, have
to be considered.
Manufacturer has provided the following data of
production and compression flows from the
electrolyser and compressor, respectively:
-Electrolyser: Each cell stack produces a maximum
of 15 Nm
3
/h and a minimum of 6 Nm
3
/h. Therefore,
there is a production range 6-60 Nm
3
/h, distributed as
follows:
-Working with a single stack: 6-15 Nm
3
/h;
-Working with two stacks: 12-30 Nm
3
/h;
-Working with three stacks: 18-45 Nm
3
/h;
-Working with four stacks: 24-60 Nm
3
/h.
-Compressor: Each group has a flow consumption in
the range of 18 and 30.9 Nm
3
/h working between 2 and
4 bar of suction pressure, respectively, so that
according to the manufacturer it could work in a range
of 18 to 61.8 Nm
3
/h.
Considering the design flow data, with the
electrolyser-compressor set we can work at a range of
18-60 Nm
3
/h.
Therefore, a single cellstack can not be used, as the
minimum production/compression flow is limited by
the minimum intake flow from a compressor unit,
which is 18 Nm
3
/h, according to the manufacturer. As
indicated above (case 1), working with a single
cellstack the electrolyser maximum production would
be below the minimum compressor intake flow and the
operation of the system would not be stable.
The following are the combinations to be studied in
order to obtain as much information as possible about
the system:
-Two cellstacks (CSs) working with a single
compressor group (CG);
-Three cellstacks working with a single compressor
group;
-Three cellstacks working with two compressor
groups;
-Four cellstacks working with a single compressor
group;
H
2

56
-Four cellstacks working with two compressor
groups.
In Fig. 6 the operation of electrolyser and compressor
with two and three CSs and one CG and with four CSs
and two CGs is shown.
In the initial part of the graphic (1), two CSs and one
CG are working. In grey line the inlet pressure of the
compressor is represented. When the inlet pressure of
the compressor increases the electrolyser production
(Prod.elect.) increases too. We can observe that the
production given by the electrolyser (Prod.elect.) is
always higher than the values given by the flow meter
(Prod. FT) installed after the electrolyser. At minimum
inlet compressor pressure (2 bar), the electrolyser
production is minimum (12 Nm
3
/h), and then it
increases. This minimum production is lower than the
minimum H
2
consumption to the compressor specified
by the manufacturer. In this initial part, the buffer and
electrolyser pressures are constant. This means that the
electrolyser produces the same H
2
quantity that the
compressor consumes.
In the second part of the graphic (2), three CSs and
one CG are operating. Here is shown the same
behaviour, when inlet compressor suction increases as
well as the electrolyser production, until the upper limit
of compressor 35 Nm
3
/h. At the maximum inlet
pressure of 4.5 bar, the compressor consumes 35 Nm
3
/h,
therefore, this is the maximum flow the electrolyser
produces, in order to maintain the pressure in the
process constant, avoiding case 2.

0
1
2
3
4
5
6
7
8
9
10
0:00:00 0:00:00 0:00:00 0:00:00 0:00:00 0:00:00 0:00:00 0:00:00 0:00:00
P
r
o
d
/
1
0
;
P
c
o
m
p
r
e
s
(
b
a
r
)
;
P
b
u
f
f
e
r
(
b
a
r
)
;
P
e
l
e
c
t
(
b
a
r
)
;
C
S
Prod.elect./10(Nm3/h) Pasp.comp(bar)
Pbuffer(bar) Pelec(bar)
Prod.FT/10(Nm3/h) NCS
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3

Fig. 6 Electrolyser operation with (1) 2CSs and 1 CG, (2)
3CSs and 1CG, (3) 4CSs and 2CGs.
The operation with four CSs and the two CGs is
shown in the third part of the graphic (3). Beginning
with 2 bar of inlet compressor pressure, the production
of electrolyser is the minimum (24 Nm
3
/h), when the
pressure increases the electrolyser production does the
same until the buffer pressure decreases and therefore
the electrolyser pressure too. This occurs due to the
electrolyser is producing less H
2
than the compressor
consumes. In the final part of the graphic the pressure
in the buffer continues decreasing and the electrolyser
closes the user valve, maintaining the process pressure
at 7.5 bar and venting the H
2
production (the H
2
flow
measure, Prod. FT, decreases).
In Fig. 7 we can observe the operation of 4 CSs and 2
CGs. In the first part (1) the inlet compressor pressure
increases and, therefore, the electrolyser production
follows the same tendency. At minimum inlet
compressor pressure (2 bar) the 2CGs consume 24-25
Nm
3
/h, less than the minimum flow specified by the
manufacter. The pressure in the electrolyser remains
constant because it produces the same that the
compressor consumes. In the second part (2), the
compressor consumes more H
2
than the one produced
by the electrolyser produces (case 1), so the pressure in
the buffer decreases and therefore the pressure in the
electrolyser decreases too, until it reaches 7.5 bar. At
that moment the electrolyser closes the user valve
partially, making independent its pressure from the
buffer pressure, and uses the H
2
produced to increase
its pressure. We can observe that behaviour in the
figure, because the electrolyser gives data of a H
2

production (Prod. elect) and nevertheless the real
production is much lower (prod. Flow_meter). When
the electrolyser recovers the pressure the production
becomes normal again.
Finally, in the last part of the graph (3), the inlet
compressor pressure decreases, therefore, the suction
flow decreases and the electrolyser production follows
the same tendence. As both variables are equal, the
pressures in the buffer and in the electrolyser remain
constant.
H
2

57
0
1
2
3
4
5
6
7
8
9
10
11
0:00:00 0:00:00 0:00:00 0:00:00 0:00:00
P
r
o
d
/
1
0
;
P
c
o
m
p
r
e
s
(
b
a
r
)
;
P
b
u
f
f
e
r
(
b
a
r
)
;
P
e
l
e
c
t
(
b
a
r
)
Prod.elec/10(Nm3/h) Pcomp_inlet(bar)
Prod.flow_meter(Nm3/h) Pelect.(bar)
Pbuffer(bar)
1 3
2

Fig. 7 Representation of operation of 4 CSs and 2 CGs.

0
1
2
3
4
5
6
7
8
9
10
11
0:00:00 0:00:00 0:00:00 0:00:00 0:00:00 0:00:00
P
r
o
d
/
1
0
;
P
c
o
m
p
r
e
s
(
b
a
r
)
;
P
b
u
f
f
e
r
(
b
a
r
)
;
P
e
l
e
c
t
(
b
a
r
)
Elect_Prod./10(Nm3/h) Pcomp_inlet(bar)
Pbuffer(bar) Pelec(bar)
Flowmeter_Prod./10(Nm3/h)
1 2 3

Fig. 8 Representation of operation of 3 CSs and 2 CGs.

In Fig. 8 the operation with 3 CSs and 2 CG is shown.
Firstly (in 1) the increase of inlet compressor pressure
is shown with the increase of electrolyser production
until reaching the maximum electrolyser capacity with
3 CSs (45 Nm
3
/h). After that (2) the inlet pressure
increase causes a decrease in the buffer pressure and in
the electrolyser pressure, until 7.5 bar. In that moment
the electrolyser closes the user valve partially and uses
the H
2
produced to increase its pressure. We can
observe it twice in the second part of the graph.
Finally, in the third part of the graphic (3) when the
inlet pressure decreases the pressure is recovered in the
buffer and electrolyser and the production to user
decreases, too.
As it was explained before this electrolyser control
mode is not suitable for the energy management of a
wind farm production. In the following test the relation
between P
inlet_compressor
/E
available storage
is the result to be
obtained.
This relationship is not direct, it can be divided into
two parts: the first is the relationship between the
compressor suction pressure and the electrolyser
production flow, and the second the relationship
between the production flow of the electrolyser and its
energy consumption.
Electrolyser H
2
production Compressor H
2
consumption
+ Compressor H
2
consumption P
inlet_compressor
+ 1 or 2
groups
Electrolyser H
2
production Pinlet_compressor + 1 or 2
groups
Therefore,
Electrolyser H
2
production Pinlet_compressor + 1 or 2 groups
+ E
available storage
Electrolyser H
2
production
E
available storage
P
inlet_compressor
+ 1 or 2 groups
Finally, we show some graphs in order to
characterize the electrolyser.
In Fig. 9 the curve U-I of the electrolyser is
represented. It was unable in this test to characterize the
first part of the curve, because the electrolyser, due to
its pressure control, produces at maximum capacity
always linked to the consumption of the compressor.
For security reasons, in order to avoid the H
2
crossover
through the IMET membrane, the CSs are not allowed
to work at H
2
productions below 40% of nominal rated.
The data was adjusted to a logarithmic curve, typical
from electrochemistry systems and electrolysers.
In Fig. 10 the stationary state variables are shown.
The efficiency is calculated as the consumption divided
by the H
2
production, therefore, as all of them produce
the same flow of H
2
(15.1 Nm
3
/h), the more efficient
will be the one with lower consumption. All the CSs
are very similar in behaviour, but we can observe that
the more inefficient is the CS2, because it has the
biggest consumption per m
3
of H
2
produced and the
one with lowest consumption is the CS3. Regarding the
temperature, all of them are working with similar
values, around 70-72 C.
5. Conclusions
The wind power management system based on H
2
in
Sotavento wind farm is being demonstrated. However,
more operation time is necessary in order to obtain
H
2

58

Fig. 9 Curve U-I.

Fig. 10 Representation of the main characteristics of
electrolyser.

definitive conclusions and achieve a complete
know-how of the facility.
Some preliminary conclusions can be established
according to the experience obtained:
-It is necessary to advance in the development of the
regulation and licenses procedures to make easy the
installation of this type of facilities in Spain.
-Regarding the tariffs, in Spain the selling of
renewable energy is regulated and has special price in
the pool as well as the energy produced by Fuel Cells.
However, the energy produced by hydrogen engines is
not included in the legislation and makes it difficult to
sell in the electrical pool.
-There is a high potential for the improvement of
control strategies of the electrolyser to operate with
renewable energy resources, since most of these
equipments are already designed to operate connected
to the grid.
-Also, it is necessary to develop different models of
the systems to simulate the whole system or plant, in
particular in that case, the tandem
electrolyser/compressor.
In this paper we present some results, but in order to
obtain more information from the facility we continue
working on the study of equipments performance and
on the modeling of the system.
References
[1] C. Jorgensen, S. Ropenus, Production price of hydrogen
from grid connected electrolysis in a power market with
high wind penetration, Int. J. Hydrogen Energy 33 (2008)
5335-5344.
[2] S.M. Schoenung, W. Hassenzahl, Long vs. Short-Term
Energy Storage: Sensitivity Analysis, A Study for the
DOE Energy Storage Systems Program, Sandia National
Laboratories, 2007.
[3] J.L. Bernal-Agustn, R. Dufo-Lpez, L. Correas-Usn, I.
Aso-Aguarta, Analysis of a wind-hydrogen system
connected to the electrical grid, in: World Renewable
Energy Congress X and Exhibition, 2008.
[4] L. Fingersh, Optimized Hydrogen and Electricity
Generation from Wind, Technical report, National
Renewable Energy Laboratory, 2003.
[5] M. Korpas, C. Greiner, Opportunities for hydrogen
production in connection with wind power in weak grids,
Renewable Energy 33 (2008) 1119-1208.
[6] M. Korpas, Hydrogen energy storage for gird-connected
wind farms, in: Proceedings of the IASTED International
Conference on Power and Energy Systems, Greece,
2001.
[7] M. Aguado, E. Ayerbe, C. Azcarate, R. Blanco, R. Garde,
F. Mallor, et al., Economical assessment of a wind
hydrogen energy system using WindHyGen software, Int.
J. of Hydrogen Energy 34 (2009) 2845-2854.
[8] J. B. S. Haldane, DAEDALUS or Science and the Future,
Cambridge Eds., 1923.
[9] P. Hoffman, Tomorrows EnergyHydrogen, Fuel Cells
and the Prospects for a Cleaner Planet, The MIT Press,
2001.
[10] M. Dubey, Hydrogen from windA clean energy system,
alternative energy sources, International Journal of
Hydrogen Energy (1978) 3551-3561.
[11] H. De Battista, R. Mantz, F. Garelli, Power conditioning
for a wind-hydrogen energy system, Journal of Power
Sources 155 (2005) 478-486.
[12] S. Kelouwani, K. Agbossou, R. Chahine, Model for
energy conversion in renewable energy system with
hydrogen storage, Journal of Power Sources 140 (2004)
392-399.
[13] S. Sherif, F. Barbir, T. Veziroglu, Wind energy and the
hydrogen economyReview of the technology, Solar
Energy 78 (2005) 647-660.
[14] Set of the existing regulations and legislative framework
H
2

59
related to RES implementation, Stories Project
Deliverable 2.2, June, 2008.
[15] S.R. Vosen, J.O. Keller, Hybrid energy storage systems
for stand-alone electric power systems: Optimization of
system performance and cost through control strategies,
International Journal of Hydrogen Energy 24 (1999)
1139-1156.
[16] S. Klouwani, K. Agbossou, R. Chahine, Model for
energy conversion in renewable energy system with
hydrogen storage, Journal of Power Sources 140 (2004)
392-399.
[17] A.G. Dutton, J.A.M. Bleijs, H. Dienhart, M. Falchetta, W.
Hug, D. Prischich, et al., Experience in the design, sizing,
economics, and implementation of autonomous
wind-powered hydrogen production systems, International
Journal of Hydrogen Energy 25 (2000) 705-722.
[18] L.N. Grimsmo, M. Korpaas, R. Gjengedal, S.
Moller-Holst, A study of a stand-alone wind and hydrogen
system, in: Nordic Wind Power Conference, 2004.
[19] R. Glckner, . Ulleberg, R. Hildrum, C.E. Grgoire, P.
Ife, Integrating renewables for remote fuel systems,
Energy & Environment 13 (4-5) (2002) 735-747.
[20] . Ulleberg, TL pryor optimization of integrated
renewable energy hydrogen systems in diesel engine
mini-grids, in: 14th World Hydrogen Energy Conference
(WHEC 2002), 2002.
[21] J. Burger, P. Lewis, Energy storage for utilities via
hydrogen systems, in: 9th Intersociety Energy Conversion
Engineering Conference, 1974.
[22] J.I. Levene, An Economic Analysis of Hydrogen
Production from Wind, NREL, Golden, CO,
NREL/CP-560-38210, 2005.


Experimental Studies on Critical Heat Flux in a
Uniformly Heated Vertical Tube at Low Pressure and
Flowrates
Husham M. Ahmed
Department of Mechanical Engineering, Gulf University, P.O.Box:26489, Manama, Bahrain

Received: December 05, 2010 / Accepted: April 29, 2011 / Published: January 31, 2012.

Abstract: Investigations into critical heat flux at low flow and pressure conditions are of particular interest when predicting the nuclear
reactor core behavior during Loss of Coolant accident (LOCA). Therefore, critical heat flux (CHF) has been investigated in a uniformly
heated vertical round tube at two low system pressures and six low water flowrates. The results have been compared with two
correlations which have different approaches and CHF look-up table. Good agreements have been obtained for the three comparisons at
the lower sets of mass fluxes. The Bowring correlation was found to be the best to correlate the experimental results with Root Mean
Square Error RMSE of 0.54% and 0.56% for the 5 bar and 15 bar system pressure respectively. A comparisons with the Shim and Lee
correlation yielded RMSE of 0.23% and 5.74% for the two system pressure respectively. When the look-up table of Groeneveld et al.
was used, RMES of 0.55% and 25.2% was obtained for the two system pressure respectively.

Key words: Critical heat flux, dryout, burnout, subcooled boiling, LOCA.

1. Introduction

Heat exchangers using Boiling mechanisms are of
interest in many industrial applications due to the large
heat transfer coefficients that can be obtained.
However, the boiling crises (known as the critical heat
flux phenomena), which is characterized by a sharp and
sudden rise of heater wall temperature, potentially
leads to heater damage, forms and constitutes a very
important restriction to this effective and efficient heat
transfer method [1, 2]. The CHF of forced convective
boiling is one of the most important parameters
limiting the thermal performance of water cooled
nuclear reactors and many other important two-phase
processes equipment such as boilers, evaporators, and
heat exchangers [3]. In these application areas, the
CHF significantly affects the performance, integrity,

Corresponding author: Husham M. Ahmed, Ph.D.,
associate professor, research fields: solar energy and critical
heat flux. E-mail: dr.husham@gulfuniversity.net.
safety, and economic competitiveness of systems and
components [4]. Therefore, the critical heat flux (CHF)
information is essential as this limiting condition forms
an important boundary when considering the
performance of such heat exchange equipment [5].
Many investigators [3, 5, 6] identified CHF in
vertically heated tubes, when the heated surface is not
in contact with cooling liquid, with two possible
regimes depending on the dryout quality. Under low
quality or even under subcooled liquid conditions, CHF
occurs when the vapor generation rate is higher than
the rate of vapor evacuation by the flowing liquid. This
situation leads to vapor accumulation on the heated
wall surface, preventing the liquid from contacting the
surface. The CHF occurrence in this case is called
departure from nucleate boiling (DNB). Under high
quality conditions, the crisis is reached as result of
gradual disappearance of the liquid film adjacent to the
wall due to entrainment and evaporation. In this
situation, the occurrence of CHF is said to be due to
D
DAVID PUBLISHING
Experimental Studies on Critical Heat Flux in a Uniformly Heated
Vertical Tube at Low Pressure and Flowrates

61
dryout (DO). Dryout may occur and a significant
amount of liquid may still flow in from of droplet
entrained by vapor. Dryout is characterized by a more
severe increase in the heated surface wall temperature.
Ref. [7] reported that different type of mechanisms can
trigger the heat transfer degradation and many
hypothetical models can be found in the literature with
no general agreement that has been reached so far.
Over the past several decades, there have been a
large number of experimental studies concerning CHF
in water under forced convective boiling conditions.
The efforts to enhance CHF have mainly been related
to the improvement of nuclear fuel performance [8, 9].
Recently, the cooling requirement of high heat flux
components of fusion reactors has further stimulated
the research in this area [4]. A realistic physical model
or general empirical correlation capable of predicting
the critical heat flux is the ultimate objective of CHF
research. An accurate prediction of CHF requires the
understanding of physical mechanisms and parametric
trends as well as sufficient experimental data bases
[4, 10]. Therefore, for any empirical CHF predictions, a
reliable and extensive experimental database is crucial
[11, 12]. Ref. [13] investigated the influence of two
phase flow characteristics on CHF at low conditions
and reported that there is not yet sufficient data to
understand the details of CHF under both low-pressure
and velocity conditions. Most of the CHF
investigations have been concentrated on conditions at
high pressure and high flow rate, corresponding to the
normal operation ranges of light water reactors. But
when predicting the nuclear reactor core behavior
during a loss of coolant accident, the CHF encountered
under low flow conditions becomes of particular
interest [4]. Therefore, in order to achieve an optimal
design and to ensure a high degree of safety in light
water reactors, the CHF characteristics under low flow
conditions must be clearly understood [14-16].
Ref. [17] reported that there are three representative
methods to predict CHF: the empirical correlation, the
mechanistic model, and the look-up tables. They
concluded that the empirical correlation technique is
the best choice to predict the CHF. This is because the
CHF phenomenon is too complex to be explained by
mechanistic models, and the CHF look-up table is, in
essence, an empirical correlation.
The CHF empirical correlations are classified as
either an inlet (upstream) condition correlation or outlet
(local) correlation [17, 18]. The inlet based correlation
depends on upstream conditions such as system pressure,
tube diameter, fluid mass flowrate, heated length and
inlet subcooling and was developed by early
investigators [19-24] using a dimensionless approach
which depended on upstream conditions. This approach
has been further modified by Refs. [5, 6, 10, 25].
The local condition correlation based on outlet
conditions, such as system pressure, fluid mass
flowrates, tube diameter, and exit quality, was
predicted by Refs. [26-28].This approach has been
further modified and developed by Refs. [17, 29-36]
introducing the dependent variable of true mass
fraction of steam.
The results of Refs. [17, 30] showed that the effect of
L/D (tube length/tube diameter) was insignificant. They
attributed that this L/D insignificance was due to the fact
that the CHF always occurs at the end of the tube in
uniform heating tubes. They concluded that the true
mass fraction is an important parameter in correlating
the CHF effectively. Therefore, they reported that their
new correlations predicted the CHF data much better
than the previous empirical correlations (which
depended on upstream conditions) in a vertical round
tube with uniform and non uniform heat flux.
Ref. [25] argued that the CHF is an upstream
dependent phenomenon, because the CHF is strongly
dependent on the heated length. Furthermore, the
uncertainty in the determination of the thermodynamic
quality is higher than that of inlet subcooling.
Therefore, he concluded that the local conditions
hypothesis is invalid.
Other investigators in Refs. [37, 38] developed
mixed condition correlations that include local and non

62
local (inlet) conditions corresponding to a particular
regime or condition.
Recently, the look-up table has been widely used to
predict CHF for various applications including
computer codes used for design and safety analysis of
nuclear reactors [17, 34, 39]. It was first proposed by
Ref. [40] based on the local condition hypothesis. It has
been further developed by Refs. [41-43]. The first
reliably and widely used look-up table has been
proposed jointly by AECL research (Canada) and IPPE
(Obnink, Russia), to combine their data bases and
proposed a unified CHF look-up table based on 22,946
data points of CHF values obtained in tubes with
vertical upward flow of steam water mixture covering a
wide range of conditions [43]. The look-up table has
been designed to give CHF values for an 8 mm tube at
discrete values of pressure, mass flux, and dry out
quality and assuming a simple correction factor for
tube diameter effect. Linear interpolation can be used
to determine the CHF for conditions between the
tabulated values. CHF values for tube diameters other
than 8 mm were obtained using correction correlations.
The look-up table has many advantages on the
classical empirical correlations or the mechanistic
models and has been considered by many investigators
as one of the most reliable prediction methods for the
CHF and is being used in several best estimate thermal
hydraulic system codes [14, 44-46]. Although the
look-up table was originally developed for steam-water
flow in vertical uniformly heated tubes, it permits the
prediction of CHF in other cases, such as, horizontal
flow, short heated length, rod bundles, and
non-uniformly heat flux condition by using correction
factors to account for these effects on the CHF [45,
47-49]. They can apply for fluids other than water and
various geometries of heat transfer surface [50].
The updated version of the CHF look-up table
contains more than 30,000 data point and provides
values at 24 pressure, 20 mass fluxes, and 23 qualities
which gives it a significant improvement in accuracy
and smoothness [51].
The object of the present study is to experimentally
predict the CHF and also to provide more necessary
data at low flow and pressure conditions in uniformly
heated vertical tubes. The experimental results were
compared with the prediction capabilities of two well
known correlations and the CHF look-up table.
2. Experimental Setup
The CHF experiments reported in this paper have
been conducted in a thermal hydraulic loop designed
and fabricated to perform study state critical heat flux
measurements. The loop was filled with de-ionized
water. The schematic diagram of the experimental
facility is shown in Fig. 1. The test section was a single
vertical tube made of Inconel-600 and has an internal
diameter of 18 mm, an outer diameter of 22 mm, and
was electrically heated uniformly in the axial direction.
The heated length was 3660 mm. The test section
schematic diagram is shown in Fig. 2.

Fig. 1 Schematic diagram of test facility.

63

Fig. 2 Schematic diagram of test section.

An Electrical burn out detector acting on the
Wheatstone bridge principle is used to detect the burn
out situation and to act as a safety device. The bridge
unbalance is due to the onset of the dry out on a part of
the heated tube which causes the power to be switched
off. A D.C electrical supply system can heat the test
section with up to 400 kW maximum power. A 400 kW
preheater was used to adjust the degree of subcooling
of water entering the test section.
A booster pump and two reciprocating pumps were
used to raise the system pressure to the required level
up to 220 bars and to control and adjust the water flow
rate to the prescribed value. A throttling valve located
up stream of the test section inlet was used to avoid
flow fluctuation.
3. Results and Discussions
Critical heat flux experiments have been carried out,
in a uniformly heated vertical tube, over a range of
water mass flux varying from 0.027-0.22 kg/ms and
for two fixed system pressures of 5 and 15 bars. The
water inlet subcoolling was varied up to 80 K, but
controlled and kept constant for each set of experiment.
Fig. 3 shows the effect of inlet subcooling enthalpy
on CHF at different flow conditions for the two system
pressure of 5 and 15 bar. It can be seen that the CHF
increases almost linearly with the inlet subcooling
enthalpy, but the effect decreases as mass flux
decreases. It also can be seen that, for the same mass
flux and inlet subcooling, the CHF slightly decreases as
the system pressure increases from 5 to 15 bar.
The experimental data has been compared with two
correlations of different approaches. Bowring
correlation [19], Shim and Lee correlation [17], and the
Groeneveld et al. CHF look-up table [51]. The three
compressions were found to give good agreements at
the lower sets of mass fluxes (especially for the 5 bar
system pressure), while it over predicts the
experimental data when compared with the data of the
higher mass flux.
The three sets of comparisons are plotted for each
mass flux in Figs. 4 and 5 for the two system pressures
used in the investigation respectively. From these
figures, it can be seen that the experimental data and
that predicted by the Bowring correlation [9] are
comparatively smooth, while the data predicted by the
Shim and Lee correlation [17] and the out look table
[51] contain some fluctuation and scattering as they
strongly depend on the vapor quality.
The Bowring correlation [9] (which adopts the
upstream and inlet subcooling philosophy) was found to
be the best to incorporate the experimental results as it
yields the lowest RMS error of 0.54% and 0.56% for the
5 bar and 15 bar system pressures respectively. In fact
the experimental data agrees or is relatively close to that
predicted by Bowring correlation in many occasions.
Furthermore, there is an excellent similarity in behavior
and trend of the predicted and experimental data using
the Bowring correlation. It slightly over predicts the
experimental data at the higher mass fluxes.

64

Fig. 3 Experimental results of the two system pressures (5 and 15 bar).

RMSE of 0.23% and 5.74% was obtained when the
experimental results, for the 5 bar and 15 bar systems
pressures, were compared using Shim and Lee
correlation [17] (which adopts the down stream
approach, that depends on local variables and the true
mass fraction of steam at the heated tube exit). It was
found that the Shim and Lee correlation [17] is very
sensitive to the value of the true mass fraction of steam.
A small variation in the true mass fraction value would
have a noticeable effect on the predicted value of CHF.
This agrees with the concern reported by Ref. [10]
about the uncertainty in the determination of the
thermodynamic quality at the end of the heated tube.
However, it can be seen that the Shim and Lee
correlation [17] gave mixed trend as it agrees with
some results, under prediction on a few occasions and
over prediction on others. It should be mentioned that
Ref. [17] reported that when their new correlation was
compared with another five well known correlations,
their new correlation gave a better prediction of CHF
than the other four correlations did. Only Bowrings
correlation [9] did better than their correlation. When
the look-up table of Groeneveld et al. 2006 [51] was
used, RMES of 0.55% and 25.2% was obtained for the
5 bar and 15 bar systems pressures respectively. The
look-up table was also regarded as a down stream
approach and therefore it is sensitive to the steam exit
quality. The direct substitution method (DSM) has
been used to predict the CHF data, which is the method
of implementation in most system codes [46].
Unfortunately, it was found that some of the present
experimental data (especially for the 15 bar system
pressure data) was close to the uncertainty, or
discontinuity area of the lookup table. It is the area,
where the boundaries between regions that
experimental data used and regions that correlations
and extrapolations were employed to derive the
Groeneveld et al., 2007 look up table. These regions
referred to in the Groeneveld et al., 2007 look up table
as dark shaded area [51]. For example, as can be seen in
Groeneveld et al. 2007 look-up table [51] shown in
Table 1, at pressure 1000 kPa, the predicted CHF is
832 kW/m at a quality of 0.6 and it rises sharply to
3102 kW/m when the quality is decreased to 0.5 that is
more than two and a half times. This may explain the
discrepancies in some of the results when they were
compared with the look-up table data. It may also
explain why the 5 bar data got a good agreement and
law RMES of 0.55% (as they lie around the
discontinuity areas), while the 15 bar data got RMES of
25.2% (as most of the 15 bar data lies inside or very
close to the discontinuity areas).

65

(a) (b)

(c) (d)

(e) (f)
Fig. 4 Comparison of experimental data with other correlationssystem pressure (5 bar).

66

(a) (b)

(c) (d)

(e) (f)
Fig. 5 Comparison of experimental data with other correlationssystem pressure (15 bar).

67
Table 1 Part of Groenveld et al. 2007 look-up table showing the discontinuity regions.
Pressure
(kPa)
Mass Flux
(kg /ms)
CHF (kW/m)

X
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.6 0.7 0.8 0.9 1
1000 0 678 492 377 318 291 269 254 231 220 193 145 0
1000 50 1351 1179 1068 933 770 723 706 586 522 369 282 0
1000 100 2216 2087 1949 1798 1700 1652 1541 1280 1078 708 501 0
1000 300 3372 3276 3172 2993 2871 2766 2894 1727 1302 815 514 0
1000 500 3953 3938 3790 3678 3579 3537 3445 1564 1067 649 377 0
1000 750 4083 3997 9851 3810 3473 3291 3102 832 623 322 213 0
1000 1000 4099 3998 3896 3708 3148 2620 1199 636 391 189 128 0
1000 1500 4057 3704 3655 3049 1601 1238 843 315 219 138 75 0
1000 2000 3992 3596 3330 1957 1122 873 624 268 155 90 46 0
1000 2500 3811 3566 2927 1197 1012 823 521 300 149 49 26 0
1000 3000 3718 3325 2007 977 895 701 583 313 168 59 24 0
2000 0 970 745 587 491 438 380 361 323 317 232 170 0
2000 50 1834 1613 1447 1331 1209 1089 1069 924 845 658 373 0
2000 100 3072 2933 2801 2599 2487 2393 2313 1996 1720 1123 594 0
2000 300 4362 4026 3725 3563 3475 3375 3276 2998 2266 1252 698 0
2000 500 4584 4373 4197 4076 3929 3828 3695 3362 1896 1109 609 0
2000 750 4870 4700 4590 4478 4396 4033 3811 1948 1194 700 229 0
2000 1000 5083 4992 4961 4792 4196 3311 2751 1260 597 359 174 0
2000 1500 4871 4766 4524 4389 2926 1903 897 512 288 163 97 0
2000 2000 4595 4283 3824 3026 1357 1172 682 418 196 97 60 0
2000 2500 4552 3943 3328 1613 1237 958 556 313 157 53 29 0
2000 3000 4483 3798 2570 1185 1020 731 594 831 172 59 27 0

4. Conclusions
In this study, systematic tests have been conducted to
predict CHF, in a uniformly heated vertical round tube,
for two fixed and relatively low system pressures, a
range of inlet subcooling, and a range of low values of
mass fluxes. Distilled water was used as the working
fluid. It is found that an increase in mass flux resulted an
increases in CHF for a given inlet subcooling and
system pressure. The CHF was also found to be
decreased as the system pressure increases and increased
as the inlet subcooling increases for a given mass flux.
The experimental data has been compared with
Bowring correlation [9], Shim and Lee correlation [17],
and the Groeneveld et al. CHF look-up table [51]. The
three compressions were found to give good
agreements (especially at the 5 bar system pressure) at
the two or three sets of the lower mass fluxes, while it
over predicts the experimental data when it is
compared with the data of the higher mass flux.
Bowring correlation [9] was found to be the best to
incorporate the experimental data. Shim and Lee
correlation [17] and the look-up table [51] were found
to be very sensitive to the calculated value of the steam
quality.
References
[1] J.M.L. Corre, S.C. Yao, C.H. Amon., Two-phase flow
regimes and mechanisms of critical heat flux under
subcooled flow boiling conditions, Nuclear Engineering
and Design 240 (2) (2010) 245-251.
[2] J.G. Collier, J.R. Thome, Convective Boiling and
Condensation, 3rd ed., Oxford University Press, Oxford,
1996.
[3] W.H. Ahmed, A. Meamer, M.A. El-Nakla, A.I. Basel, I.
Ismail, Towards understanding the critical heat flux for
industrial applications, International Journal of Multiphase
Flow 36 (2010) 153-165.
[4] S.H. Chang, W.P. Baek, Understanding, prediction, and
enhancing critical heat flux, in: The 10th International
Topical Meeting on Nuclear Reactor Thermal Hydraulics
(NURETH-10), Seoul, Korea, Oct., 2003, pp. 5-9.
[5] R. Sindhuja, A.R. Balakrishnan, S.S. Murthy, Critical heat

68
flux of R-407C in upflow boiling in a vertical pipe,
Applied Thermal Engineering 28 (2008) 1058-1065.
[6] B.R. Vijayarangan, S. Jayanti, A.R. Balakrishnan, Studies
on critical heat flux in flow boiling at near critical
pressures, International Journal of Heat and Mass Transfer
49 (2006) 259-268.
[7] J.M.L. Corre, S.C. Yao, H.C. Amon, A mechanistic model
of critical heat flux under subcooled flow boiling
conditions for application to one and three-dimensional
computer codes, Nuclear Engineering and Design 240
(2010) 235-244.
[8] J. Yang, D.C. Groeneveld, L.K.H. Leung, S.C. Cheng,
M.A. Nakla, An experimental and analytical study of the
effect of axizl power profile on CHF, Nuclear Engineering
and Design 237 (2006) 1348-1395.
[9] D.C. Groeneveld, On the definition of critical heat flux
margin, Nuclear Engineering and Design 163 (1995)
245-247.
[10] A. Olekhnovitch, A. Teyssedou, A. Tapucu, P.
Champagne, D.C. Groeneveld, Critical heat flux in
vertical tube at low and medium pressures, Part
IExperimental results, Nuclear Engineering and Design
193 (1-2) (1999) 73-89.
[11] D.D. Hall, I. Mudawar, Critical heat flux for water flow in
tubesi. Compilation and assessment of world CHF data,
Int. J. Heat Mass Transfer 43 (2000) 2573-2604.
[12] D.D.D. Hall, I. Mudawar, Critical heat flux for water flow
in tubesII. Subcooled CHF correlations, Int. J. Heat
Mass Transfer 43 (2000) 2606-2640.
[13] A. Inoue, S.R. Lee, Influence of two-phase flow
characteristics on critical heat flux in low pressure,
Experimental Thermal and Fluid Science 19 (1999)
172-181.
[14] S.Y. Chun, H.J. Chung, S.K. Moon, S.K. Yang, M.K.
Chung, I. Schoesse, et al., Effect of pressure on critical
heat flux in uniformly heated vertical annulus under low
flow conditions, Nuclear Engineering and Design 203
(2001) 159-174.
[15] S.W. Ruan, S.M. Yang, Characteristics of the critical heat
flux for downward flow in a vertical round tube at low
flow rate and low pressure conditions, Experimental
Thermal and Fluid Science 7 (4) (1993) 296-306.
[16] S.H. Chang, W.P. Baek, T.M. Bae, A study of critical heat
flux for low flow of water in vertical round tubes under
low pressure, Nuclear Engineering Design 132 (1991)
225-237.
[17] W.J. Shim, J.S. Lee, Prediction of CHF in vertical round
tubes with uniform heat flux for water, Journal of
Industrial and Engineering Chemistry 12 (1) (2006)
127-133.
[18] R. Kadi, S. Aissani, M. Ferrouk, A. Bouam, Prediction of
critical heat flux (CHF) in subcooled water flow boiling by
application of the mechanistic model, Nuclear Research
Center of Birine CRNB P.0. Box: 180 An-Oussera 17200,
Algeria, 2006.
[19] R.W. Bowring, A Simple but Accurate Round Tube
Uniform Heat Flux, Dryout Correlation over the Pressure
Range 7-170 bar, UKAEA report, AEEW-R789, 1972.
[20] M. Caira, G. Caruso, A. Naviglio, A correlation to predict
CHF in subcooled flow boiling, International
Communications in Heat and Mass Transfer 22 (1995)
35-45.
[21] A.A. Katto, A generalized correlation of critical heat flux
for forced convection boiling in vertical uniformly heated
round tubes, Int. J. Heat Mass Transfer 21 (1978)
1527-1542.
[22] A.A. Katto, A generalized correlation of critical heat flux
for forced convection boiling in vertical uniformly heated
round tubes-a supplementary report, Int. J. Heat Mass
Transfer 22 (1979) 783-794.
[23] A. Katto, Prediction of critical heat flux for annular flow in
tube taking into account the critical liquid film thickness
concept, Int. J. Heat Mass Transfer 27 (1984) 883-891.
[24] Y. Katto, L. Ohno, An improved version of the generalized
correlation of critical heat flux for convective boiling in
uniformly heated vertical tubes, Int. J. Heat Mass Transfer
27 (1984) 1641-1648.
[25] A. Olekhnovitch, A. Teyssedou, P. Tye, Critical heat flux
under choking flow conditions: Part IIMaximum values
of flow parameters attained under choking flow conditions,
Nuclear Engineering and Design 205 (1-2) (2001) 175-190.
[26] S. Levy, Forced convection subcooled boiling-prediction
of vapor volumetric fraction, Int. J. Heat Mass Transfer 10
(1967) 951-965.
[27] P.C. Kroger, N. Zuber, An analysis of the effects of
various parameters on the average void fraction in
subcooled boiling, Int. J. Heat Mass Transfer 11 (1968)
211-233.
[28] P. Shah, N. Zuber, Point of net vapor generation and void
fraction in subcooled boiling, in: Proceedings of the 5th
Int. Heat Transfer Conference, Tokyo, 1974, pp. 175-179.
[29] W.J. Shim, H.L. Jae, D.J. Gyoo, CHF for uniformly heated
vertical tube under low pressure and low flow conditions,
in: 35th Intersociety Energy Conversion Engineering
Conference and Exhibit (IECEC), 2000, Vol. 1, pp.
411-419.
[30] W.J. Shim, S.W. Lee, J.Y. Park, J.S. Lee, A new CHF
correlation for medium pressure and high mass flux,
Applied Chemistry 6 (1) (2002) 476-479.
[31] H.C. Kim, W.P. Baek, S.H. Chang, Critical heat flux of
water in vertical round tubes at low pressure and low flow
conditions, Nuclear Engineering and Design 199 (2000)
49-73.
[32] H.C. Kim, W.P. Baek, S.H. Chang, Towards unification of

69
critical heat flux prediction models for flow boiling based
on liquid film dryout mechanism, in: Proc. of Second
Japan-Korea Symposium on Nuclear Thermal Hydraulics
and Safety (NTHAS-2), Fukuoka, Japan, 2000, pp.
762-767.
[33] J.M. Kim, J.Y. Park, S.W. Lee, W.J. Shim, Comparison of
models for prediction of true mass fraction of steam, J.
Applied Chemistry 7 (1) (2003) 297-300.
[34] W.J. Shim, J.Y. Park, Analysis for a generalized CHF
model in vertical round tubes with uniform heat flux,
Journal of Industrial and Engineering Chemistry 9 (5)
(2003) 607-613.
[35] W.J. Shim, J.Y. Park, Prediction of critical heat flux (CHF)
for vertical round tubes with uniform heat flux in medium
pressure regime, Korean Journal of Chemical Engineering
21 (1) (2004) 75-80.
[36] S.W. Lee, J.Y. Park, J.S. Lee, W.J. Shim, Generalized
CHF correlation with uniform heat flux for vertical round
tube, Applied Chemistry 8 (1) (2004) 315-318.
[37] T. Jafri, T.J. Dougherty, B.W. Yang, Local and non local
correlations for critical heat flux at low and medium
pressures, ASME Heat Transfer Division (HTD) 3 (1996)
229-238.
[38] M.M. Shah, Improved general correlation for critical heat
flux during upflow in uniformly heated vertical tubes,
Heat and Fluid Flow 8 (4) (1987) 326-335.
[39] A. Zoubir, B. Tidjani, S.C. Youcef, Comparison of
correlation for predicting critical heat flux for uniformly
heated vertical porous coated tubes at low pressure,
Nuclear Engineering Design 240 (2010) 71-73.
[40] V.E. Doroshchuk, L.L. Levitian, P.F. Lantsman,
Recommendation for calculating burnout in round tube,
Teploenergetika 12 (1975) 66-70.
[41] D.C. Groeneveld, S.C. Chen, T. Doan, AECL-OU CHF
look-up table, Heat Transfer Eng. 7 (1986) 46-62.
[42] D.C. Groeneveld, L.K.H. Leung, F.J. Erbacher, P.L.
Kirillov, V.P. Bobkov, W. Zeggel, An improved look-up
table method for predicting critical heat flux, in:
Proceedings of Sixth International Topical Meeting on
Nuclear Reactor Thermal Hydraulics (NURETH-6),
Grenoble, France, 1993, Vol. 1, pp. 223-230
[43] D.C. Groeneveld, L.K. Leung, P.L. Kirillov, V.P. Bobkov,
I.P. Smogalev, V.N. Vinogradov, et al., The 1995 look-up
table for critical heat flux in tubes, Nuclear Engineering
and Design 163 (1996) 1-23.
[44] A. Tanase, S.C. Cheng, D.C. Groeneveld, J.Q. Shan,
Diameter effect on critical heat flux, Nuclear Engineering
Design 239 (2009) 289-294.
[45] D.K. Chandraker, P.K. Vijayan, D. Saha, R.K. Sinha,
Investigation on the characteristic of CHF in various flow
pattern regimes based on look-up table data, Nuclear
Engineering Design 238 (2008) 170-177.
[46] W.P. Baek, H.C. Kim, S.C. Chang, An independent
assessment of Groeneveld et al.s 1995 CHF look-up table,
Nuclear Engineering Design 188 (1997) 331-337.
[47] N.I. Kolev, Check of the 2005 look up table for prediction
of CHF in bundles, Nuclear Engineering and Design 237
(2007) 978-981.
[48] I.L. Pioro, D.C. Groeneveld, L.K.H. Leung, S.S. Doerffer,
S.C. Cheng, Y.V. Antoshko, et al., Comparison of CHF
measurements in horizontal and vertical tubes cooled with
R-134a, International Journal of Heat and Mass Transfer
45 (2002) 4435-4450.
[49] D.C. Groeneveld, L.K.H. Leung, L.A.Z. Vasic, Y.J. Guo,
C.A. Cheng, A look-up table for fully developed
film-boiling heat transfer, Nuclear Engineering and
Design 225 (2003) 83-97.
[50] W.J. Shim, J.Y. Park, Analysis for a generalized CHF
model in vertical round tubes with uniform heat flux,
Journal of Industrial and Engineering Chemistry 9 (5)
(2003) 607-613.
[51] D.C. Groeneveld, J.Q. Shan, A. Vasic, L.K.H. Leung, A.
Durmayaz, J. Yang, et al., The 2006 CHF look-up table,
Nuclear Engineering and Design 237 (2007) 1909-1922.


Environmental Impacts of Oil and Gas Production in
Nigeria
Cecily. O. Nwokocha, Donatus. U. Ugwu, Abiodun. B. Fagbenro and Emmanuel. J. Cookey
Physics Department, Alvan Ikoku Federal College of Education, P.M.B 1033 Owerri Imo State, Nigeria

Received: October 27, 2010 / Accepted: April 01, 2011 / Published: January 31, 2012.

Abstract: Energy and environment are essential for sustainable development of any nation hence the need for effective and efficient
energy production. UNFCCC reports indicate that those who are least responsible for climate change are the most vulnerable to its
projected impact and the masses are the ones affected by environmental degradation and lack of access to clean and affordable
energy services. This paper therefore addresses the effects of these issues, some forms of energy production in Nigeria and the effects
on our environment and how it can be effectively managed through innovative policies and partnership with organizations that can
help pilot projects that will build a sustainable energy.

Key words: Energy, environment, production, oil and gas, Nigeria.

1. Introduction

The objective of the energy system is to provide
energy services. Energy services are the desired and
useful products, processes or indeed services that result
from the use of energy, such as for lighting, provision
of air-conditioned indoor climate, refrigerated storage,
transportation, appropriate temperatures for cooking
etc. The energy chain to deliver these cited services
begins with the collection or extraction of primary
energy, which is then converted into energy carriers
suitable for various end-uses [1]. These energy carriers
are used in energy end-use technologies to provide the
desired energy services. From the foregoing, [2, 3] it is
clear that energy is an essential input to all aspects of
modern life. It is indeed the life wire of industrial
production, the fuel for transportation as well as for the
generation of electricity in conventional thermal power
plants. The situation was such that nations were
complacent with the oil dominated scenarios of the
1950s and 1960s during which time regular and

Corresponding author: Cecily. O. Nwokocha, research
fields: atmospheric physics, renewable energy. E-mail:
doncecily@yahoo.com.
reasonably cheap supplies were available. However, oil
producing countries caused a world-wide reaction by
deciding to increase the prices of crude oil in the early
seventies. Of course, the oil-rich countries like Nigeria
recorded tremendous economic gains. On the other
hand, those developing nations that did not have oil
were subjected to serious economic problems as they
suddenly found themselves utilizing, in some cases, up
to 50% of their foreign exchange earnings for
importing petroleum products or crude oil in order to
sustain their oil-based industries and public utilities.
Nigeria, the most populous nation in Africa with a
population of over 140 million, about 1/5th of Africas
population has been a major concern when it comes to
energy usage and its sustainability. Though classified
as a developing country, the oil rich country has
witnessed developments over the years with the influx
of foreign companies to make use of the energy forms
(oil, gas coal etc.) which has been of social and
economic importance to the country, but also with
disadvantages that has affected the society over the
years the discovery of oil can be traced back to the
early 1950s and it has been the major source of energy,
D
DAVID PUBLISHING
Environmental Impacts of Oil and Gas Production in Nigeria

71
before the discovery of he other energy forms like
natural gas and other renewable. By 2004, the energy
consumption mix was dominated by oil (58%)
followed by natural gas (38%) and hydroelectric (8%)
with coal and other energy forms not part of the mix
[3, 4]. The effect of these non renewable has been the
one noted over the years.
2. Nigerias Energy Forms
Nigeria uses about three forms of energy, the natural
gas, the hydro-electricity and the solar electricity but
the major concern of this paper is on the natural gas
which is the major source of revenue for the country
(Fig. 1). According to the Oil and Gas Journal [5],
Nigeria has 36.2 billion barrels of proven oil reserves
as at January 2007 emerging as the highest crude oil
producer in Africa, meaning more revenue for the
country. But this dependence on oil has negatively
affected the Nigerian environment especially areas
where it is being exploited.
Various environmental issues such as soil and water
contamination due to oil spillage affecting productivity,
aquatic life and human health are not far fetched. This
leads to migration to urban areas causing congestion
and increased crime rates in the cities.
In addition to this is pipeline vandalism, kidnapping
and militant takeover of oil facilities in the Niger Delta
areas of exploration (Fig. 2). This has effect on the
income rates because since 2005 (December) Nigeria
has lost about $16 billion in export due to shut in oil
production, with an oil income generation of about
$300 billion in the past 3 decades without appreciable
comparative development.
The Niger Delta militant groups and other militia
organizations in search of monetary compensation and
or political leverage are behind these attacks. Chief
among the conditions is equitable sharing of the oil
wealth, controlled oil spilling and transparency of
government budget, from the crude oil revenue which
generated about 25 trillion naira in the past nine years
for Nigeria, which is about 90% of income with no
commensurate development.

Fig. 1 Showing oil exploration in Nigeria.

Fig. 2 Map of Nigeria showing the study area.
2.1 Natural Gas
Nigeria has the largest natural gas reserves in Africa
and the 7th largest in the world (Fig. 3). Because many
of Nigerias fields lack infrastructures to prospect
natural gas, it is flared. Currently, effort to harness is on
with a deadline of 31st December 2008 against flaring.
According to NNPC, the petroleum corporation in
Nigeria, the country flares 40% of its annual natural gas
while the World Bank estimates that Nigeria accounts
for 12.5% of the worlds totally flared gas, with gas
possibly providing approximately 3 times of current
fund generated from oil if properly harnessed.
Nigeria supplies its neighboring countries (Ghana,
Togo and Benin) natural gas to generate revenue but
with its rate of flaring it has caused more harm than
good.
Environmental issues such as acid rains, air pollution,
and ozone layer depletion [3, 6, 7] are on the increase.

72

Fig. 3 Natural gas exploration as a source of energy in Nigeria.

The current ongoing West African gas pipeline project
spearheaded by the Nigerian government is one major
project to harness the gas flared with support to other
sister nations in West Africa.
Current NIPP (National Integrated Power Projects)
which are being constructed in the major industrial
regions of the nation are planned to make use of gas.
Power stations are constructed with no gas to power
them due to vandalisations in the Niger Delta region.
Similar projects such as gas powered electric
generating stations have been sited all over zones of
high demands, such as those about to be commissioned
Shell- Elf NNPC Afam 650 MW project, Geregu,
Papalanto, Omotosho gas stations amongst others in
the Independent Power Stations Projects (IPPS).
2.2 Hydro-Electric Energy Source
This sector has been the least developed growing
only from 5% to 8%. Hydropower (Fig. 4), major
sources of which are still underdeveloped, has an
estimated potential of the range 2-3 TW with an annual
output of 10,000-20,000 TWh [6-9] but is only
available in certain areas of the world, also available in
limited areas of the world are wind energy and biomass.
The rate at which power outage occurs has resulted in
increased use of generators which affects the natural
environment with the various gases released.
Despite these blackouts, customers are billed for
services not rendered which explains the widespread
vandalisation, power theft and other problems the
Power Holding Company of Nigeria experiences.

Fig. 4 Hydroelectric energy as a source of energy in Nigeria.

In addition is the environmental impact of dams
when it overflows causing erosion.
Other forms of hazard the energy sector has caused
the Nigerian environment are global warming, adverse
effects of automobiles which is the main form of
transport, sea level rise and climate change just to name
a few.
In conclusion, the development of any economy is
pivoted on the capacity of energy and its utilization. It
is expedient that one major index of determining
development as nations are rated as 1st, 2nd, or 3rd
world in energy capacity and utilization.
A nation like South Africa with a population of less
than 1/5th of Nigerias has energy capacity of over 35
thousand Megawatts generation of electricity.
Industrial developments are achieved based on
energy capacity. Industries are built with
infrastructural support, based on energy provisions.
Hence it is no gainsaying to conclude that there can be
no meaningful development in the real sense with the
occurrence of affective energy development. Also it
can be observed that the energy sector of Nigeria is
dependent on non renewable which though beneficial
has more adverse effect on the natural environment.
Therefore, the need to start making use of the
renewable which are less harmful and even in abundance.
3. Environmental Concerns of Oil Gas in
Nigeria
Oil has been an important part of the Nigerian
economy since vast reserves of petroleum were

73
discovered in Nigeria in the 1950s. For example,
revenues from oil have increased from 219 million
Naira in 1970 to 10.6 billion Naira in 1979. Shell Oil
operates many of its oil facilities in the oil-rich Delta
region of Nigeria. The Ogonis, an ethnic group that
predominate in the Delta region, [5, 10] have protested
that Shells oil production has not only devastated the
local environment, but has destroyed the economic
viability of the region for local farmers and producers.
The Nigerian Federal Government, on the other hand,
has been charged with failing to enact and enforce
environmental protections against oil damage by Shell
and other oil companies. Furthermore, many Ogonis
have been harassed and even killed by the Federal
government for organizing protests and threatening
sabotage of oil facilities.
Oil production in Nigeria has had severe
environmental and human consequences for the
indigenous peoples who inhabit the areas surrounding
oil extraction. Nigerias export of 12 million barrels of
oil a day comes from 12% of the country's land, and
indigenous minority communities in these areas receive
no economic benefits. Indigenous groups are actually
further impoverished due to environmental degradation
from oil production and the lack of adequate regulations
on multinational companies, as they become more
vulnerable to food shortages, health hazards, loss of land,
pollution, forced migration and unemployment.
The social and environmental costs of oil production
have been extensive. They include destruction of
wildlife and biodiversity, loss of fertile soil, pollution
of air and drinking water, degradation of farmland and
damage to aquatic ecosystems, all of which have
caused serious health problems for the inhabitants of
areas surrounding oil production. Pollution is caused
by gas flaring, above ground pipeline leakage, oil waste
dumping and oil spills. Approximately 75% of gas
produced is flared annually causing considerable
ecological and physical damage to other resources such
as land/soil, water and vegetation. Gas flares, which are
often times situated close to villages, produce soot
which is deposited on building roofs of neighboring
villages. Whenever it rains, the soot is washed off and
the black ink-like water running from the roofs is
believed to contain chemicals which adversely effect
the fertility of the soil. Gas pipelines have also caused
irreparable damage to lands once used for agricultural
purposes. These pipes should be buried to reduce risk
of fracture and spillage. However, they are often laid
above ground and run directly through villages, where
oil leaks have rendered the land economically useless.
Oil spills and the dumping oil into waterways has
been extensive, often poisoning drinking water and
destroying vegetation. According to an independent
record of Shells spills from 1982 to 1992, 1,626,000
gallons were spilt from the companys Nigerian
operations in 27 separate incidences. Of the number of
spills recorded from Shell-a company which operates
in more than 100 countries-40% were in Nigeria.
Shell is also being accused of engaging in
widespread ecological disturbances, including
explosions from seismic surveys, pollution from
pipe-line leaks, blowouts, drilling fluids and refinery
effluents, and land alienation and disruption of the
natural terrain from construction of industry
infrastructure and installations. For example, oil spill
contamination of the top soil has rendered the soil in
the surrounding areas unsuitable for plant growth by
reducing the availability of nutrients or by increasing
toxic contents in the soil. Gas flaring, on the other
hand, has been associated with reduced crop yield and
plant growth on nearby farms, and disruption of
wildlife in the immediate vicinity. Shell and other oil
companies have developed an easy and inexpensive
way to deal with by-products from oil drilling:
indiscriminate dumping.
The crisis over environmental pollution and
economic marginalization from the oil industry
reached a peak in January 1993 when 300,000 Ogoni
protested against Shell Oil. This organized protest was
followed by repeated harassment, arrests, and killing of
Ogonis by Federal government troops.

74
4. Policy Implications and Way Forward
Through these cases and several large oil spills that
have happened in recent years, the world is beginning
to realize the environmental consequences that arise
from the economic dependence on oil resources. Even
if companies and governments would make a total
effort towards environmental safety, significant
accidents will still occur. Industries can make strides
toward reducing the damage on resources that people
use and on the wildlife and habitats. Industries cannot,
however, completely prevent accidents, such as an oil
spill into rivers or oceans, from happening.
Furthermore, [5, 11] environmental damage that occurs
as a result of conflict cannot be completely prevented
either. If oil resources are present during fighting, then
they may be accidentally or deliberately damaged. In
fact, deliberate attacks on oil resources and production,
with resulting environmental devastation, could be
common in future international conflicts. The recent
Iraqi burning of oil fields during the Kuwait War could
be as prelude to future international aggression.
One proposed method of reducing the amount of oil
production is the imposition of an oil tax. A tax would
increase the price of oil, and it has been predicted that
this would reduce the world demand for oil by about
2.6 million barrels per day (b/d) and 5.3 million b/d in
the respective years 2000 and 2010. Although this plan
is designed to reduce the amount of gas emissions, it
could also be useful in reducing the amount of
environmental damage that stems from oil production.
However, it could be argued that a tax that
discriminates against oil is not only unfair to oil
producing countries, but also is unlikely to be efficient
for reducing gas emissions and oil production. One
reason for the inefficiency is that a probable energy
substitute is coal, which is even more environmentally
damaging. The ultimate solution may be the discovery
and production of a new energy resource to replace oil.
A new energy resource that is environmentally-safe
and cheaply accessible could solve many of the
problems associated with oil production. Surveys of the
American public dating back to 1979 show that the
populace favors research and development for
renewable energy technologies and energy
conservation over research and development for
nuclear energy, petroleum, and coal. This new resource
would be especially useful for countries such as the US,
which are heavily dependent on oil resources that are
found outside its borders. In addition, universal access
to this resource could help to alleviate the international
conflicts that have thus far arisen over natural resources.
This new path is a possibility within the US if the
government puts full effort into acquiring a new energy
resource; although, without action from the
government, no current company would be interested
in pursuing another resource other than low-cost oil.
5. Conclusion
The development of any economy is pivoted on the
capacity of energy and its utilization. It is expedient
that one major index of determining development as
nations are rated as 1st, 2nd, or 3rd world [12] in
energy capacity and utilization. Thus the energy sector
of Nigeria being dependent on non renewable which
though beneficial has more adverse effect on the
natural environment. Therefore, the need to start
making use of the renewable which are less harmful
and even in abundance is paramount this includes:
The need for increased awareness in the country
because theres less information circulated about the
impacts of the energy sector.
The major advantages of the renewable energy
technologies include the simplicity of the technologies,
ease of maintenance as well as their enhanced
environmental friendliness over fossil fuel systems.
There is clear evidence of the use of renewable
energy technologies at the moment. However, there is
the necessity to increase the use of the system
especially for rural development. In this regard there is
the urgent need for more support of research,
development, demonstration and diffusion activities in
the existing research centers as well as identified

75
groups in other institutions further harnessing of gas
being flared to generate more energy diversification of
the economy into more of non oil and P.P.P (Public
Private Partnership) financing with provisions of basic
and essential infrastructures.
References
[1] W.W.S. Charters, Solar and Wind Power Technologies for
Remote Applications, CSC Technical Publication Services
No. 187, Commonwealth Science Council, 1985.
[2] G. Boyle, Renewable Energy Power for a Sustainable
Future, Oxford University Press, 1996.
[3] D.M. Considine, Energy Technology Handbook,
McGrawhill Book Company, 1977.
[4] C.O. Folayan, Estimate of global solar radiation bound for
some Nigerian cities, Nigerian Journal of Solar Energy 7
(1988) 36-48.
[5] A.H.I. Ibrahim, Capital limitations, environmental
movements may interfere with expansion plans, Oil & Gas
Journal (1994) 60-68.
[6] A.S. Sambo, Empirical models for the correlation of
global solar radiation with meteorological data for
Northern Nigeria, Solar and Wind Technology 3 (2) (1986)
89-93.
[7] U.O. Aliyu, S.B. Elegba, Prospect for small hydropower
development for rural applications in Nigeria, Nig. Journal
of Renewable Energy 1 (1990) 74-86.
[8] M.D.C. Doyle, A.S. Sambo, Correlation of diffuse solar
radiation with air mass, Solar and Wind Technology 5 (1)
(1988) 99-102.
[9] A.S. Sambo, The Measurement and prediction of global and
diffuse components of solar radiation for Kano in northern
Nigeria, Solar and Wind Technology 5 (1) (1988) 1-5.
[10] A.S. Sambo, Wind Energy Assisted Solar Electricity
Generating Schemes for the Rural Areas of Nigeria, Large
Scale Systems in Developing Countries, Joja Educ.
Research and Pub. Ltd., 1987, pp. 45-160.
[11] R.J. Goodland, Tropical origin of ecology: Eugens
Warmings jubilee, Oikos 26 (1975) 240-245.
[12] A.G. Armstrong, M. Vallejo, Ecuador Oil, Trade,
Environment, and Human Rights, ECUADOR HTM
Bonne.


Impact of the Target Film Thickness on the Properties of
the Neutron Tube
Shuang Qiao, Shiwei Jing and Mingrui Zhang
Department of Physics, Northeast Normal University, Changchun 130024, China

Received: October 21, 2010 / Accepted: March 30, 2011 / Published: January 31, 2012.

Abstract: Neutron tube is a kind of accelerator neutron source, which has been applied extensively. It is mainly composed of an ion
source system, an accelerator system, and a target system. The target system is one of the most important parameters of the neutron tube,
and it impacts directly the yield, the lifetime, and the stability. By far the relation between the property of neutron tube and target
thickness has not been studied drastically in the Institute of Radiation Technology of Northeast Normal University. High-quality
titanium target should be produced in order to manufacture the neutron tube with intense yield, long lifetime and high stability. In our
experiment, the pure titanium film was evaporated on the ceramic target by using the evaporation technique. The impact of the
thickness of the titanium film on the yield of the neutron tube was studied, and the yield of the neutron tube was on optimistic state if the
thickness of titanium film was 2.2 m. The properties of training and stability of neutron tubes with different thick titanium film target
were also present in this paper.

Key words: Neutron tube, target, titanium film, yield, training, stability.

1. Introduction

Small neutron generators using the deuterium
(
2
H)-tritium (
3
H) reaction are the most common
accelerator based (as opposed to isotopic) neutron
sources. Neutrons are produced by creating deuterium
ions and accelerating these ions into a tritium or
deuterium target. Compared to reactors and large
accelerators which have higher neutron output, neutron
generators can provide sufficient neutron output for
many applications, which are portable or mobile, they
operate in pulsed mode, and they are orders of
magnitude less expensive to purchase and to operate.
Compared to isotopic sources, neutron generators
produce higher energy neutrons that are more
penetrating, can be turned off, operate in pulsed or
continuous mode, are easier to transport, and they dont
waste neutrons between analyses or during
maintenance periods. So diagnostic techniques using
neutron beams have a broad spectrum of applications in

Corresponding author: Shiwei Jing, Ph.D., research fields:
neutron tube and its application. E-mail: jingsw504@nenu.edu.cn.
advanced manufacturing, explosives and contraband
detection, medicine, and industry [1-14].
Target property is one of the most important
parameters of the neutron tube [15-19]. The primary
study on the target was carried out in the Radiation
Technology Institute of Northeast Normal University.
By comparison, the yield with about 2 m thickness
target of the tube was relatively high, and the
characteristics of the stability and the training of the
neutron tube was also studied in the experiment.
2. Experimental Setup
The pure titanium (99.999%) film was evaporated on
the ceramic substrate. The experimental parameters
were shown in Table 1. Dfferent thick titanium targets
were got by changing the mass of pure titanium
filament and by changing the heating electric current.
The thickness of the titanium films were characterized
by SEM.
The output of the neutron tube was monitored by
3
He
neutron monitor. The
3
He tube (5 cm) is 25 cm long
D
DAVID PUBLISHING
Impact of the Target Film Thickness on the Properties of the Neutron Tube

77
Table 1 The experimental condition and parameters of
titanium film deposition.
Target
code
Heating
electric
current (A)
Substrate
Temperature
(
o
C)
Pressure
(Pa)
Film
thickness
(m)
1# 100-120 160 5 10
-3
0.8
11# 100-120 160 5 10
-3
2.2
29# 100-120 160 5 10
-3
6.1

with 5 cm polythene enclosement. The connection of
neutron monitor and MCA was shown in Fig. 1, the
neutron tube was placed in the training box full of
silicon oil, and the low voltage and high voltage of
3
He
neutron monitor were supplied by NIM box. The
signals were collected through preamplifier and
amplifier then into MCA and stored in computer.
3. Experimental Results
The experiment was carried out in the neutron hall
10 meters underground. Three neutron tubes with
different thick titanium film were tested under the same
measurement condition. Before test, the high voltage
of neutron monitor was kept at 1450 V, and then the
neutron generator was operated. By adjusting the
parameter of the reservoir, the ion source, and the
acceleration voltage, the yield, the training time, and
the stability of the neutron tube can be obtained in the
experiment.
The MCA data of 29# neutron tube is shown in
Fig. 2; it showed that neutron output increased with the
increase of acceleration voltage. The correlation
between the acceleration voltage and the output of
three neutron tubes with different target was shown in
Fig. 3. From Fig. 3, the target film thickness affected
greatly the output of the neutron tube not only the
neutron output at the same acceleration voltage, but the
functional relation between two parameters. The yield
of 11# neutron tube was relatively high, and its target
titanium film thickness was 2.2 m. Based on the
calibration test in the Station of Radiation Dose in
Chinese Institute of Atomic Energy (CIAE), the output
of 29# neutron tube was 1.16 10
6
n/s under the
condition of 80 kV acceleration voltage, 2500 V ion
source voltage, anode current 530 A, target beam

Fig. 1 Neutron monitor and its connection.

0 500 1000
0
50
100
150
200
1
2
3
4
40kV 40kV 40kV 40kV 40kV 40kV 40kV 40kV 40kV 40kV
5
c
o
u
n
t
s
channel
1: 40kV
2: 50kV
3: 60kV
4: 70kV
5: 80kV

Fig. 2 The neutron spectra in MCA to reflect the
correlation between neutron yield and acceleration voltage
for 29# neutron tube. Line 1 to 5 is the MCA spectrum when
the acceleration voltage is 40, 50, 60, 70, 80 kV, respectively.

40 50 60 70 80
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
C
o
u
n
t
s
High voltage (kV)
29# y=0.00504x
3.32332
1# y=0.16714x
2.22368
11# y=(1.23457E-4)x
4.25162

Fig. 3 The correlation between neutron yield and
acceleration voltage for 29#, 11#, 1# neutron tube.

current 130 A. With the same electric parameter, the
yield of 11# and 1# neutron tube was 1.71 10
6
n/s and
0.35 10
6
n/s, respectively. The yield of 11# neutron
tube was three times of that of 1# tube, and 1.5 times of
29# tube. It showed that there was an optimized target

78
film thickness, the film thickness would not be too
thick nor too thin.
By simulating the relation between the yield and the
acceleration voltage of the neutron tube, the yield of the
neutron tube with different thickness titanium target
varied greatly. The functional relation between the
yield and the acceleration voltage of 11#, 29# and 1#
neutron tube was Y
1
= 0.00012 U
4.25
,Y
2
= 0.00504U
3.32
,
and Y
3
= 0.16714 U
2.22
, respectively. The result
showed that the target film thickness affected the
exponential relation between the output and the
acceleration voltage. By computation based on the
simulation result, the output of three neutron tubes with
2.2 m, 0.8 m, and 6.1 m, respectively, was 1.30
10
7
n/s, 5.68 10
6
n/s, 1.03 10
6
n/s, respectively, at
130 kV acceleration voltage. If the acceleration voltage
reached 185 kV (just like the neutron tube of Russia),
the yield of 11#, 29# and 1# neutron tube was 5.82
10
7
n/s, 1.83 10
7
n/s, 2.25 10
6
n/s, respectively. It is
very significative when the yield of a D-D neutron tube
reaches 10
7
n/s, because at this time the D-D neutron
tube can displace Am-Be or
252
Cf neutron source and
can be applied in many areas. At the same time D-D
neutron tube has no tritium element, so there is no
radioactive pollution in the manufacture and operation
period and it is a green neutron tube.
The training characteristics of three neutron tubes
were shown in Fig. 4. The electric parameter was ion
source voltage 2500 V, reservoir heating current 2.7 A,
anode current 300 A, acceleration voltage 80 kV,
respectively. The count time was 600 s and the counts
was just from the peak of full energy of thermal neutrons.

It can be seen from Fig. 4 that tube 29# and 1#
showed normal trend of neuron yield increase with the
increase of gas in target, and finally to saturation state.
The saturation time for tube 29# and 1# was about 110
minutes. The saturation time of tube 11# showed two
time process. Former 100 minutes was normal and the
last three data may root of the change of the electric
parameter since the acceleration voltage affected
extensively the output of the neutron tube. The saturation
-20 0 20 40 60 80 100 120 140 160
1000
2000
3000
4000
5000
6000
7000
C
o
u
n
t
s
Time (minutes)
11#
29#
1#

Fig. 4 The training properties of 29#, 11#, 1# neutron tube.

time of tube 11# was 20 minutes more than that of tube
29# and 1#, which may root of the thickness of target
titanium film or other reasons. The exact cause of
saturation time variation will be given in the future
experiments.
The stability of three neutron tubes was tested and
the results were shown in Fig. 5. The electric
parameters of the neutron generator were kept same
during the test period as ion source voltage 2500 V,
heating current of reservior 2.7 A, anode electric
current 300 A, acceleration voltage 80 kV. The count
time was 600 s every time and the counts were just
from the peak of full energy of thermal neutrons. The
total test time for 1#, 29#, and 11# was 720, 510, and
330 minutes, respectively. The stability of tube 1# was
relatively best and the relative undulation of most data
was within 2%. The relative undulation of tube 29# and
11# is within 4% and the stability of tube 29# was
better than tube 11#. During the test of stability, after
the electric parameter was fixed on, any other
adjustment was not carried out, so the undulation of
city electric grid was also included in the stability test.
Some data varied extensively maybe came from this
reason, so the correlation between the stability and the
titanium film thickness can not be perorated by now.
4. Discussion
The characteristics of neutron tube were highly
related with the thickness of target titanium film,

79
0 100 200 300 400 500 600 700 800
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
R
e
l
a
t
i
v
e

u
n
d
u
l
a
t
i
o
n
Time (minutes)
1#
29#
11#

Fig. 5 The stability properties of 29#, 11#, 1# neutron tube.

especially for the output. The exact thickness of the
titanium film would be controlled in the future
experiment of film deposition. The micro-structure of
the film would also be considered. It can be seen from
Fig. 3 that the relation between the acceleration and the
output of the neutron tube was not the same if the target
film thickness had changed. Of course the target
thickness is not the only determinant parameter, other
parameters such as the target material, the substract
material, the substract temperature, the deposition
speed, can also affect the output of neutron tube [19].
These parameters would be studied earnestly for high
yield neutron tube. But for the parameter of target
thickness, 1.5-2.5 m target thickness should be the
best choice for the compact neutron tube, and the result
coincides with the conclusion reported by Carlole
Monnin and Pierre Bach [19].
The experiment results from the neutron monitor
were modified by multi-measurement to decrease the
impact of instability of the neutron monitor. The
measurement time was at least 300 seconds and the
count in full-energy peak was at least 1000 to decrease
the statistical error. Furthermore, the stable-voltage
power-supply would be adopted in the future
experiment to keep the acceleration voltage the same in
every measurement (the ion source voltage and the
reservior current were very stable) to ensure the
credibility of the data.
Acknowledgment
This work was supported in part by the Project of
Technicians Service for Enterprises from the Ministry
of Science and Technolgy of the Peoples Republic of
China (Contract No. 2009GJB10017).
References
[1] N. Cerullo, RF irradiation facility for boron neutron
capture therapy application based on a rf-driven D-T
neutron source and a new beam shaping assembly,
Review of Scientific Instruments 73 (2002) 3614-3618.
[2] E.V. Gromov, Progress in neutron logging in Russia
describing a new portable neutron generator, Appl. Radiat.
Isot. 46 (6/7) (1995) 639-640.
[3] L. Petrizzi, Sensitivity and uncertainty analysis
performed on a 14-MeV neutron streaming experiment,
Fusion Engineering and Design 51-52 (2000) 843-848.
[4] S.W. Jing, Structural and optical properties of TiO
2
-
x
N
x

films deposited by the RF magnetron sputtering, Chinese
Phys. Lett. 23 (3) (2006) 682-685.
[5] S.W. Jing, S. Qiao, Y.R. Liu, Development of fast pulse
neutron coal analyzer, Review of Scientific Instruments
76 (2005) 045110-045110-5.
[6] S.W. Jing, L.M. Liu, The ceramic-cup microwave ion
source for sealed-tube neutron generator, Review of
Scientific Instruments 72 (1) (2001) 92-95.
[7] S.W. Jing, Coal analysis using the pulsed neutron
generator generator, Nuclear Science and Techniques 14
(4) (2003) 265-267.
[8] S.W. Jing, L.M. Liu, Comparison of different plasma
chambers in microwave ion source for the intense neutron
tube, Nuclear Science Techniques 14 (1) (2003) 83-85.
[9] L.M. Liu, Y.R. Liu, S.W. Jing, Neutron Generator and Its
Applications, Atomic Energy Press, 2005, pp. 30-50.
[10] D.L. Chichester, B.W. Blackburn, Radiation fields from
neutron generators shielded with different materials,
Nuclear Instruments and Methods in Physics Research B
261 (2007) 845-849.
[11] B. Perot, C. Carasco, S. Bernard, A. Mariani, J.L. Szabo,
E. Mercier, et al., Development of the EURITRACK
tagged neutron inspection system, Nuclear Instruments
and Methods in Physics Research B 261 (2007) 295-298.
[12] D. Koltick, Y. Kim, S. McConchie, I. Novikov, M.
Belbot, G. Gardner, A neutron based vehicle-borne
improvised explosive device detection system, Nuclear
Instruments and Methods in Physics Research B 261
(2007) 277-280.
[13] S. Pesente, G. Nebbia, G. Viesti, F. Daniele, D. Fabris, M.
Lunardon, et al., Progress in tagged neutron beams for

80
cargo inspections, Nuclear Instruments and Methods in
Physics Research B 261 (2007) 268-271.
[14] J. Reijonen, Neutron generators developed at LBNL for
homeland security and imaging applications, Nuclear
Instruments and Methods in Physics Research B 261
(2007) 272-276.
[15] J.M. Verbeke, Development of a sealed-accelerator-tube
neutron generator, Applied Radiation and Isotopes 53
(2000) 801-809.
[16] I.J. Kim, Development of D-D neutron generator, Nucl.
Instru. Meth. B 241 (2005) 917-920.
[17] B.J. Hughey, A long-lived tritiated titanium target for fast
neutron production, Nucl. Instru. Meth. B 95 (1995)
393-401.
[18] A.Y. Kuznetsov, A method for accurate measuring of
yield parameters for neutron generators having complex
targets, Radiation Physics and Chemistry 64 (2002)
257-260.
[19] C. Monnin, P. Bach, P.A. Tulle, M. van Rompay, A.
Ballange, Optimisation of the manufacturing process of
tritide and deuteride targets used for neutron production,
Nucl. Instru. Meth. A 480 (2002) 214-222.


A Novel Transient Current-Based Differential Algorithm
for Earth Fault Detection in Medium Voltage Distribution
Networks
Mohamed F. Abdel-Fattah
1,2
and Matti Lehtonen
2

1. Department of Electrical Engineering, Aalto University, PO Box 13000, Espoo, FI-00076 Aalto, Finland
2. Department of Electrical Power and Machines Engineering, Faculty of Engineering, Zagazig University, Zagazig, Sharqia, PO
Box 44519, Egypt

Received: August 05, 2010 / Accepted: April 28, 2011 / Published: January 31, 2012.

Abstract: This paper presents a novel transient current differential algorithm for earth fault detection in unearthed (isolated) and
compensated neutral medium voltage (MV) networks. The proposed algorithm uses the transient residual currents, which are very
sensitive for earth faults detection. The transient values of residual currents are calculated for each feeder in the network and used as an
earth fault indicator. The flow of residual currents is investigated. It is found that the residual current for the faulted feeder is equal to
the summation of all residual currents for all other healthy feeders. Based on this investigation, a differential technique is proposed. A
percentage restrain performance is proposed to ensure the selectivity and security of the algorithm. The transient algorithm is very
sensitive for earth fault incidence. To apply the proposed algorithm, the residual currents can be measured easily by one sensor for each
feeder with no need to voltage measurement. The proposed algorithm is less dependent on the fault resistance and the faulted feeder
parameters. The network is simulated by ATP/EMTP program. Different fault conditions are covered in the simulation process:
different fault inception angles, fault locations and fault resistances.

Key words: Earth faults, earth capacitance, transient impedance, transient frequency, unearthed and compensated MV networks.

1. Introduction
Differential protection is a fast, selective method of
protection against short-circuits which is applied in
many power system elements: generators, transformers,
busbars and transmission lines. Most differential-relay
applications are of the current-differential type.
Percentage differential relays create a restraining signal
in addition to the differential signal and apply a percent
(restrained) characteristic. The main advantage of
unearthed neutral in power systems is small earth fault
currents which do not require immediate shut down,

Corresponding author: Mohamed F. Abdel-Fattah,
postdoctoral researcher/lecturer, research fields: earth-faults
detection, diagnosis and location using transient and fast
transient-based algorithms and the intelligent protection
systems for self-healing capability of the smart grids. E-mail:
mohamed.abdel-fattah@aalto.fi.
but the main problem is the over-voltage that resulted
by charging of the system capacitance of the sound
phases, which may lead to flashover or breakdown.
Also, it may establish a double line to earth fault. In
networks with an unearthed (isolated) neutral, the
currents of single phase to ground faults depend mostly
on the phase to ground capacitances of the lines. When
the fault happens, the capacitance of the faulty phase is
bypassed, leading to unsymmetrical system. The fault
current is composed of the currents flowing through the
earth capacitances of the two sound phases [1]. The
faulted phase current will equal to the summation of the
healthy phases currents. If these equal currents
considered as an input and output currents to the
protected zone then the differential protection
technique can be applied. Only the current signals are
D
DAVID PUBLISHING
A Novel Transient Current-Based Differential Algorithm for Earth Fault
Detection in Medium Voltage Distribution Networks

82
required and no need for voltage signals which leading
to a practical strong technique. The fault current in
unearthed (isolated) neutral systems is small and in
compensated neutral power systems (resonance
grounded/earthing systems or Petersen coil systems)
the system earth capacitance is compensated by the
connected inductance leading to decreasing in the earth
fault currents. Therefore, the sensitivity of
conventional relays, that normally are based on the
fundamental components of the voltage and current at
power frequency, to very small values of fault currents
will be reduced and it is found that the transient based
schemes are more sensitive to fault incidence in these
systems. The transient based schemes utilize the
transient components in the fault signals to detect the
fault. The initial transients of earth-faults are of
importance primarily in unearthed and compensated
networks. The transient components include charge
and discharge of the network capacitance in
earth-faults and consist of different frequencies. The
voltage of the faulty phase falls suddenly, giving rise to
the discharge transient while the charge transient is
generated by the voltage rise in sound phases during a
single-phase to ground fault. This means that a charge
transient is always a side effect of the ground fault.
Moreover, the charge component dominates the
amplitude of the composite transient and therefore it is
reasonable to be used for single-phase to ground fault
detection [1-3]. The charge component has to flow
through the transformer winding and consequently its
frequencies are substantially lower than those of
discharge component. The impedance of the
compensation coil is relatively high at transient
frequencies. Consequently, the transients are about
similar in both of unearthed and compensated neutral
networks.
Transient-based protective schemes can detect high
impedance and intermittent arcing faults, leading to
improvement in the power system reliability. However,
there are the hidden faults which involve developing
and self extinguished/arcing faults which may not be
detected by the conventional protection system. In
particular, high resistance arcing faults are highly
random as viewed from their current waveforms. In
contrast to present protection techniques based on fault
transient detection, such as traveling wave based
protection techniques, the transient based techniques
are not limited by the bandwidth of the conventional
transducers, and are able to accurately separate and
extract the high frequency information required from
the dominant power frequency signals [4]. The
transient residual current and voltages are proposed to
be used for earth fault detection. Regarding to the
network simulation, it is very important to validate the
results from the simulation process and to be satisfied
that it can represent the actual realistic conditions.
Many research works have been done in this area in the
last few years, for earth fault detection in unearthed and
compensated neutral networks, but the goal is still not
fulfilled.
In Ref. [5], comparative investigations between real
earth fault field tests and the equivalent simulation data
have been presented. A simple powerful technique in
Ref. [6] is proposed to utilize the peak values of the
superimposed waveforms. In the same time, these
peaks can be used to synchronize the waveforms,
supported by a suitable secure manipulation within the
algorithm. Another efficient way is to use the
probabilistic analysis as proposed in Ref. [7]. The earth
fault characteristics, analysis and comparison, in
medium voltage distribution networks, 20 kV with high
impedance earthing, were presented in Ref. [8]; the
presented results are based on the evaluation of 465
real case data recording, obtained from substations of
distribution networks with an unearthed or a
compensated neutral during the observations period of
3 years; continued investigations were presented in Ref.
[9]. The earth fault location for different fault
conditions that in these networks, were also
investigated in Refs. [10-15] and a probabilistic-based
method is proposed in Ref. [16]. A residual equivalent
circuit for Peterson coil grounded distribution systems

83
have been presented in Ref. [17] and an extended zero
sequence equivalent circuit for compensated medium
voltage systems, with load current contribution
accounted, is presented in Ref. [18] and the detection
of single line-to-ground fault in compensated networks
has been investigated in Refs. [19, 20]. The transient
period features have been investigated in Ref. [21]. The
transient residual currents, of the network feeders, have
been utilized for earth fault detection and faulted feeder
selection in Ref. [22]; based on extensive
investigations on the residual current levels of the
faulty and healthy feeders during earth fault condition.
The effect of network configuration on the setting
values has been investigated.
The residual value is equal to the summation of the
instantaneous phase values (voltages or currents) as
given by Eqs. (1) and (2), which is equal to three times
the earth mode (zero mode of the modal components).
The residual voltage and currents are equal to zero in
normal operation and become meaningful in fault
condition. They are very sensitive for earth faults and
from practical point of view, it can be measured easily
by one sensor for each feeder; hence it is suggested to
use it for fault detection.
) ( ) ( ) ( ) ( t v t v t v t v
c b a r
(1)
) ( ) ( ) ( ) ( t i t i t i t i
c b a r
(2)
In this paper, a novel transient differential algorithm
for fault detection, using the transient residual currents,
will be presented. The proposed algorithm uses a
differential technique for fault indication, with unusual
manner. A differential percentage factor is proposed to
be used with a suitable setting. The validity of
algorithm is confirmed for both of unearthed (isolated)
and compensated neutral medium voltage (MV)
networks. Different fault conditions, inception angles,
fault locations and resistances were simulated to check
the validity of the algorithm.
2. The Simulated Network
Based on a typical 20 kV medium voltage
distribution network configuration, a suitable network
model will be proposed for the required investigations.
A realistic network of the arrangement shown in Fig. 1
is proposed for the study.
A primary (110/20 kV) substation feeds a rural
overhead network consisting of five feeders of many
tens of kilometers length. The mathematical model of
the network has been implemented using the
alternative transient program. The alternative transient
program ATP/EMTP is a popular simulation software
package mainly intended for transient analysis
applications [23, 24]. ATPDraw, a graphical
pre-processor to ATP, has been used to construct the
network elements using suitable graphical
blocks/symbols, to plot the required figures and to
write the required output data files in a suitable form
for further advanced analysis in Matlab [25]. The
transmission line frequency dependent model (JMarti)
is used for feeders simulation with sampling time of
100 s. The network data are given in the appendix.
3. The Proposed Algorithm
For the simulated network shown in Fig. 1, the
residual bus voltage and feeder currents are calculated
at substation (measuring points). Fig. 2 shows the
waveforms of the residual voltage at the bus and the
residual feeder currents. The waveforms for all healthy
feeders are approximately the same but situation is
different for faulted feeder as shown. The transient
period is the first few milliseconds directly after fault
incidence. The enlarged view of the transient period is
shown in Fig. 3.

Fault Point
66/20 kV, /Y
Substation
Transformer
Fault Indicator
Measuring Point
CB
Background Network
4 Feeders (143 km)
F5
F4
F3
F2
F1
Faulted Feeder (41 km)

Fig. 1 The simulated medium voltage distribution
network.

84

4 8 12 16 20 24 28
-50
-28
-6
16
38
60
Faulted
Current
(A)
Healthy
Current
(A)
Voltage (kV)

Fig. 2 The waveforms of the residual voltage and the
residual currents for faulted and healthy feeders, for one
period (20 ms for 50 Hz) after the fault incidence (8 ms
incidence time, 10 fault resistance and 80% fault
distance).

7 8 9 10 11 12
-40
-30
-20
-10
0
10
20
30
40
Faulted
Current (A)
Healthy
Currents (A)
Voltage (kV)
Polarity window

Fig. 3 The transient period (the first few milliseconds
directly after the fault incidence) for the waveforms shown
in Fig. 4 at 8 ms incidence time, 10 fault resistance and
80% fault distance.

From investigations of the simulation results it is
found that a suitable transient window of 2.5 ms is
adequate to cover all different fault conditions,
different fault inception angles, fault locations and fault
resistances. In transient period, normally we found a
half cycle of the residual voltage and a full cycle of the
residual currents. It means that the polarity is not
changed for voltage but changed for current. For the
first half cycle of the current, the voltage and current
polarities are reversed for faulted feeder and in the
same time are equal for healthy feeders as shown in
Fig. 3, which agrees with the polarity comparison
technique. Therefore, the polarity window covers the
first half cycle of the current earth mode, it is the most
sensitive transient window, in the transient period, to
fault incidence.
The window starts at beginning of transients, i.e.
when the values of the voltage and current earth modes
are meaningful. The window terminates at the zero
crossing of the current earth mode which normally
occurs at the maximum absolute value of voltage. The
period of the window is not constant; it varies
according to the fault characteristics, mainly depends
on the fault resistance and incidence angle [21]. The
residual fault current is composed of the residual
currents flowing through the earth capacitances of
background network [26], as presented in Fig. 4. The
other impedances of the network components are small
compared to those of the earth capacitance and can
hence be neglected.
The earth capacitance of the network depends on the
types and lengths of the lines connected in the same
part of the galvanic connected network. From Fig. 4, it
can be shown that the residual current for the faulted
feeder (I
r
) is equal to the residual current of the
background network (I
rb
); summation of residual
currents of other healthy feeders, but in case of a
healthy feeder, the summation of other feeders gives an
unequal result. Hence, the differential technique can be
used here for fault detection. The input and output
currents are the feeder residual current and background
network residual current. The proposed application of
the differential technique here is different from usual; if
the input and output currents are equal then the feeder
is faulted otherwise it is healthy.
The algorithm proposes to analyze the data in the
polarity window presented in Ref. [21]. The presented
technique in Ref. [21] is based on the transient
impedance which requires voltage and current signals
analysis but the proposed algorithm here is only based
on residual current signal analysis which leads to a
practical and powerful algorithm. The average values
of the residual currents will be calculated in polarity
window, and can be calculated from the discrete
samples as follows:
) (
1
,
feeder each for
N
i
I
N
k
k r
r
(3)

85
Faulted Feeder, 41 km
Measuring
Point
Fault
Point
Background
Network
143 km
66 kV
Supply
66/20 kV
/Y
Healthy Feeders
C
e
I
rb
I
r
Differential Technique

Fig. 4 The flow of the earth fault residual currents, during
a single-phase to ground fault, through the earth
capacitance of the background network.

where i
r,k
is the instantaneous residual current at sample
k calculated from Eq. (2) and N is the number of
samples in the polarity window.
The operating signal of the proposed differential
algorithm is equal to the absolute value of the
difference between the feeder residual current and the
background residual current; summation of other
feeder residual currents (the summation is equal to zero
in fault condition). To improve the performance of the
proposed differential algorithm, a percentage restraint
technique will be applied. The absolute of the
difference will be divided by the total absolute value of
all feeder residual currents (i.e. summation of the
absolute values of all feeder residual currents) as
follows:
100
,
,
total r
back r r
I
I I
K (4)
Where:
) (
,
feeders other for I I
r back r
(5)
) (
,
feeders all for I I
r total r
(6)
In faulty condition, the values of the differential
factor K are different for healthy and faulty feeder.
For faulty feeder the average value of K is always
equal to 100%. The polarities, in the polarity window,
are taken into account leading to strong algorithm;
which achieved by the minus sign in Eq. (4). For faulty
feeder, the polarity of the background residual current
will be opposite to the polarity of the feeder residual
current, and hence the result of the difference gives
value equal to two times the feeder residual current
value. Also, the total sum of the absolute values of the
residual current for all feeders gives value equal to two
times the feeder residual current value, and then the
value of K should equal to 100%. For healthy feeders,
the average value of K mainly depends on the number
of background network feeders and its lengths with
respect to the faulted feeder.
For the simulated system shown in Fig. 1, with
comparable feeder lengths, the average value of K is
equal to 25% but this value may change with different
network configurations. For example, for nine feeders
network, the average value of K will equal to 12.5%,
this is better for selectivity but on the other hand for
three feeders network the average value of K will equal
to 50%. The proposed algorithm is not working with
two feeders network which is not common in
distribution networks, in which the two average value
of K will equal to 100% (for both of the faulty and
healthy feeders).
The suitable setting value of the differential factor is
the average value between the healthy and faulty
conditions, which lies at the middle of the gap between
them. Therefore, the proposed setting value of the
differential factor (K) is 62.5%. For compensated
networks (Peterson coil/resonance grounded/earthing
networks), the impedance of the compensation coil is
relatively high at transient frequencies. Consequently,
the transients are about similar in both of unearthed and
compensated neutral networks. This is clearly
investigated from the simulation data that will be
presented in the following section.
4. Results and Setting
For the case shown in Fig. 4, the calculated values
are:
The faulted feeder residual current: I
rf
= 6.80 A;
The healthy feeder residual current: I
rh
-1.69 A;
The faulted feeder differential factor: K
f
= 99.71%;
The healthy feeder differential factor: K
h

25.15%.

86
Fig. 5 shows the variation of the differential factor,
for healthy and faulty conditions with the fault
incidence time, over the power frequency period (0-20
ms), Fig. 6 shows its variation with the fault resistance
and Fig. 7 shows its variation with the fault distance.
The values of the ratios for healthy and faulty
conditions for unearthed network are presented by
solid lines and for compensated network are presented
by dash lines. From the simulated results presented in
Figs. 5-7, it can be investigated that there is an
adequate gap between the healthy and faulty condition,
and no overlapping between the two conditions. This
gap can cover many of sources of error in the
measurements, that may give incorrect higher or lower
differential factors, and the proposed algorithm will
operate effectively.
The proposed technique is valid at higher values of
fault resistance (up to 1 M) and from Fig. 6, we can
investigate a higher error in the samples of small
amplitude at higher resistances after 30 k. In these
cases the limitation of application will mainly depend
on the sensitivity of the measuring devices due to very
low values of voltage and current samples. The
simulations were performed at sampling frequency of
100 kHz. It is found that lower sampling frequency
around 10 kHz can be used for data analysis without
affecting much in the sensitivity of operation, hence it
is suitable for practical implementation.
The proposed algorithm is very simple and requires
few calculations and does not need complicated
calculations or special signal processing. The proposed
algorithm uses the residual currents only and no need
for voltage signals.
The residual currents can be measured easily by one
sensor for each feeder. At normal operation, the
residual currents are zero or very small values. The
algorithm is proposed to work after transient detection
that is confirmed by all feeder currents. The feeder is
confirmed to be faulted if the magnitude of differential
factor (normally around 100%) is greater than the
setting value.
Faulty Condition
l
Healthy Condition
Setting Factor = 62.5%
Incidence time (ms)
D
i
f
f
e
r
e
n
t
i
a
l

F
a
c
t
o
r

0 2 4 6 8 10 12 14 16 18 20
0
25
50
75
100
125

Fig. 5 The variation of the differential factor for healthy
and faulty conditions with the fault incidence time, over the
power frequency period (0-20 ms) at 80 fault resistance
and 80% fault distance for unearthed and compensated
networks.

Healthy Condition
l
Faulty Condition
Fault resistance ()
D
i
f
f
e
r
e
n
t
i
a
l

F
a
c
t
o
r

1 10 100 1000 10000 100000 1000000
0
25
50
75
100
125

and faulty conditions with the fault resistance at 8 ms
incidence time and 80% fault distance.

Faulty Condition
Healthy Condition
Fault distance (%)
D
i
f
f
e
r
e
n
t
i
a
l

F
a
c
t
o
r

0 10 20 30 40 50 60 70 80 90 100
0
25
50
75
100
125

and faulty conditions with the fault distance at 8 ms
incidence time and 80 fault resistance.


87
The security of the algorithm can be increased by
using the polarity check to confirm the fault incidence.
This can be done by confirming that the polarity of the
faulted feeder current is opposite to the polarity of all
other healthy feeder currents. This technique is very
immune to any higher values of residual currents due to
unbalance operation, which is limited to specific level.
The proposed differential technique here supervises all
the network feeders, to discriminate between the
healthy and faulted feeders after transient detection,
which is not usual in normal differential algorithm that
protects only a specified zone such as one section of the
transmission line.
Fig. 8 presents the flowchart that summarizes the
main steps in the programming process for the
proposed transient current-based differential algorithm.

Start
Read:
i
a
(t), i
b
(t), i
c
(t)
(feeders 1-5)
Calculate:
I
r
(feeders 1-5)
(in polarity window)
Transient
Detection
Diffe. Factor
K > K
setting

Confirm Polarity
Feeder is faulted
Trip time setting unit
Suitable
Trip action
End
Yes
No
No
Yes

Fig. 8 The main steps in the programming process for the
proposed transient current differential algorithm.
5. Conclusions
A novel transient current differential algorithm is
proposed for both of unearthed and compensated
medium voltage networks. The algorithm uses only the
residual current signals for each feeder of the network.
A new application of differential protection is proposed
based on investigation of a phenomenon in the
unearthed and compensated networks. It is found that
the residual current for the faulted feeder is equal to the
summation of residual currents of all other healthy
feeders, hence it is used for fault detection which is
different from the usual application of differential
protection.
A percentage differential factor is proposed, with
polarity confirmation, to increase the selectivity and the
security of the algorithm. The proposed differential
algorithm supervises all network feeders, not only one
feeder. Different simulations, using ATP/EMTP
program, at different fault angles, fault resistances and
fault distances, have been done to confirm the algorithm
validity. Lower sampling rates around 10 kHz can be
used. From the simulation results it is found that the
performance of algorithm is accepted and suitable for
practical implementation.
References
[1] M. Lehtonen, T. Hakola, Neutral Earthing and Power
System Protection-Earthing Solutions and Protective
Relaying in Medium Voltage Distribution Networks, ABB
Transmit Oy, FIN-65101, Vassa, Finland, 1996.
[2] P. Imri, M. Lehtonen, Transient based earth fault location
in 110 kV subtransmission networks, in: 15th Power
Systems Computation Conference (PSCC 2005), Lige,
Belgium, 2005.
[3] S. Hnninen, M. Lehtonen, T. Hakola, E. Antila, J. Strm,
S. Ingman, Characteristics of earth faults in power systems
with a compensated or an unearthed neutral, in: 14th
International Conference and Exhibition on Electricity
Distribution (CIRED' 97), Birmingham, UK, 1997.
[4] Z.Q. Bo, F. Jiang, Z. Chen, X.Z. Dong, G. Weller, M.A.
Redfern, Transient based protection for power
transmission systems, in: IEEE Power Engineering
Society Winter Meeting (PESW), Vol. 3, 2000, pp.
1832-1837.

88
[5] M.F. Abdel-Fattah, M. Lehtonen, EMTP-ATP
investigations of earth-fault transient simulation in
realistic Finnish 21 kV overhead line medium voltage
network, in: EEUG Meeting 2009: European EMTP-ATP
Conference, Delft, The Netherlands, 2009.
[6] M.F. Abdel-Fattah, M. Lehtonen, A novel earth fault
detection technique based on extraction of transient
superimposed components from the phase currents, in:
EEUG Meeting 2009: European EMTP-ATP Conference,
Delft, The Netherlands, 2009.
[7] M.F. Abdel-Fattah, M. Lehtonen, A probabilistic-based
technique using transient RMS currents for earth fault
detection in medium voltage distribution networks, in: The
10th International Conference on Developments in Power
System Protection (DPSP 2010), Manchester, UK, 2010.
[8] S. Hnninen, M. Lehtonen, Characteristics of earth faults
in electrical distribution networks with high impedance
earthing, EPSR (Electric Power Systems Research) 44 (3)
(1998) 155-161.
[9] S. Hnninen, M. Lehtonen, T. Hakola, Earth faults and
related disturbances in distribution networks, in:
Proceedings of IEEE PES SM2001, Vancouver, Canada,
2001 [CD-ROM].
[10] G. Achleitner, C. Obkircher, L. Fickert, M. Sakulin, C.
Raunig, An earth fault distance location algorithm in
compensated networks with additional estimation of the fault
impedance and fault current, in: Power Quality and Supply
Reliability Conference (PQ 2008), 2008, pp. 193-198.
[11] S. Hnninen, M. Lehtonen, Method for detection and
location of very high resistive earth faults, European
Transactions on Electrical Power 9 (5) (1999) 285-291.
[12] J.B. Roberts, D. Hou, H. Altuve-Ferrer, Sensitive ground
fault detection system for use in compensated electric
power distribution networks, United States Patent,
6573726 B1 (2003).
[13] L.V. Bogdashova, V.E. Kachesov, Parametric on-line fault
location methods for distribution MV networks, in: IEEE
Power Tech 2005, Russia, 2005, pp. 1-7.
[14] E. Bjerkan, T. Venseth, Locating earth-faults in
compensated distribution networks by means of fault
indicators, in: International Conference in Power System
Transients (IPST05), Montreal, Canada, 2005, pp. 1-6.
[15] W. Cong, Z.C. Pan, G. Zheng, J. Hou, F. Zhang, Q.L.
Zhang, Study on single phase to ground fault site location
method based on injection signal and gsm short message,
in: IET 9th International Conference on Developments in
Power System Protection (DPSP 2008), 2008, pp.
365-369.
[16] S. Hnninen, M. Lehtonen, U. Pulkkinen, A probabilistic
method for detection and location of very high resistive
earth faults, Electric Power Systems Research 54 (3)
(2000) 199-206.
[17] R. Kaczmarek, W.Y. Huang, P. Bastard, Equivalent circuit
application to a phase to ground fault detection in
distribution networks without MV voltage measurements,
in: Eighth IEE International Conference on Developments
in Power System Protection, Vol. 2, 2004, pp. 481-485.
[18] W.Y. Huang, R. Kaczmarek, Equivalent circuits for an
SLG fault distance evaluation by curve fitting in
compensated distribution systems, IEEE Transactions on
Power Delivery 23 (2) (2008) 601-608.
[19] W.Y. Huang, R. Kaczmarek, SLG fault detection in
presence of strong capacitive currents in compensated
networks, IEEE Transactions on Power Delivery 22 (4)
(2007) 2132-2135.
[20] W.Y. Huang, R. Kaczmarek, Symmetric components for
transient regime application in MV systems, in:
International Conference in Power System Transients
(IPST07), Lyon, France, 2007, pp. 1-4.
[21] M.F. Abdel-Fattah, M. Lehtonen, A transient fault
detection technique with varying fault detection window
of earth modes in unearthed MV systems, in: 6th
International Conference on Power Quality and Supply
Reliability (PQ2008), Prnu, Estonia, 2008, pp. 181-186.
[22] M.F. Abdel-Fattah, M. Lehtonen. A simple
transient-based wide-area selectivity technique for
earth-faults in isolated and compensated neutral medium
voltage networks, in: The 4th International Conference on
Electrical and Control Technologies (ECT 2009), Kaunas,
Lithuania, 2009.
[23] EMTP Home Page, http://www.emtp.org/.
[24] L. Prikler, H. Hidalen, ATPDraw Users Manual, Sintef
Energy Research, TR A4790, 1998.
[25] Mathworks Home Page, http://www.mathworks.com/.
[26] R. Kaczmarek, W.Y. Huang, P. Bastard, Equivalent circuit
application to a phase to ground fault detection in
distribution networks without voltage measurements, in:
SUPELEC, France, 2004.
[27] Prysmian Cables & Systems Finland Home Page,
http://www.prysmian.fi/index.html.
Appendix
The simulated network data:
(1) Substation transformer:
U1/U2 = 66/20 kV
Delta/Star connection

89
S = 10 MVA, x = 10.0%
(2) Network:
Five 20 kV overhead line feeders
Total length = 184 km
Feeder 1 length = 41 km
Network positive sequence capacitance: 5.1 F
Network zero sequence capacitance: 9.6 F
Line average positive sequence inductance: 0.9 mH/km
(3) Distribution transformers
Five step-down transformers
U1/U2 = 20/0.4 kV
Delta/Star-earthed connection
(4) Loads:
Five 0.4 kV loads
Total load = 5.5 MW
Load 1 rating = 1.5 MW
(5) Overhead line configuration [27]:
MV wooden pole with steel cross-arm
Bare conductors ACSR 54/9 Raven
Aluminum Conductor Steel Reinforced (ACSR)
Current rating (in air at 80
o
C) = 280 A
Number of wires = 7 (6 Aluminum & 1 Steel)
Complete conductor diameter = 10.1 mm
Total cross-section area = 62.4 mm
2

DC resistance (at 20
o
C) = 0.536 /km
Conductor spacing = 1.1 m and 8.1 m height


Chaos Control and Modified Projective Synchronization
of Chaotic Dissipative Gyroscope Systems
Faezeh Farivar
1
, Mahdi Aliyari Shoorehdeli
2
, Mohammad Teshnehlab
3
and Mohammad Ali Nekoui
3
1. Department of Mechatronics Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
2. Department of Mechatronics Engineering, Faculty of Electrical Engineering, K. N. Toosi University of Technology, Tehran, Iran
3. Department of Control Engineering, Faculty of Electrical Engineering, K. N. Toosi University of Technology, Tehran, Iran

Received: November 28, 2010 / Accepted: April 29, 2011 / Published: January 31, 2012.

Abstract: This paper proposes the chaos control and the modified projective synchronization methods for chaotic dissipative
gyroscope systems. Because of the nonlinear terms of the gyroscope system, the system exhibits chaotic motions. Occasionally, the
extreme sensitivity to initial states in a system operating in chaotic mode can be very destructive to the system because of unpredictable
behavior. In order to improve the performance of a dynamic system or avoid the chaotic phenomena, it is necessary to control a chaotic
system with a periodic motion beneficial for working with a particular condition. As chaotic signals are usually broadband and noise
like, synchronized chaotic systems can be used as cipher generators for secure communication. This paper presents chaos
synchronization of two identical chaotic motions of symmetric gyroscopes. Using the variable structure control technique, control laws
are established which guarantees the chaos control and the modified projective synchronization. By Lyapunov stability theory, control
lows are proposed to ensure the stability of the controlled and synchronized system. Numerical simulations are presented to verify the
proposed control and the synchronization approach. This paper demonstrates that synchronization and anti-synchronization can coexist
in dissipative gyroscope systems via variable structure control.

Key words: Dissipative gyroscope, chaos control, modified projective synchronization, variable structure control.

1. Introduction

Chaos in control systems and controlling chaos in
dynamical systems have both attracted much interest in
recent years. A chaotic system has complex dynamical
behaviors that possess some special features, such as
excessive sensitivity to initial conditions, broad
spectrums of Fourier transform, bounded and fractal
properties of the motion in the phase space, etc..
Chaotic phenomena can be found in many scientific
and engineering fields such as biological systems,
electronic circuits, power converters, chemical systems,
and etc. [1]. The pioneering work of Ott, Grebogi, and
Yorke [2] proposed the well-known OGY control

Corresponding author: Faezeh Farivar, Ph.D., research
fields: chaotic systems, intelligent control, nonlinear control.
E-mail: Faezeh_Farivar84@yahoo.com,
F.Farivar@srbiau.ac.ir.
method, where the control of chaotic systems has been
widely studied.
Chaos control can be mainly divided into two
categories [3]: one is the suppression of the chaotic
dynamical behavior and the other is to generate or
enhance chaos in nonlinear systems. Nowadays,
different techniques have been proposed to achieve
chaos control [4-6].
Also, since the synchronization of chaotic dynamical
systems has been observed by Pecora and Carroll [7] in
1990, chaos synchronization has become a topic of
great interest [8-10]. Synchronization phenomena have
been reported in the recent literature. Until now,
different types of synchronization have been found in
interacting chaotic systems, such as complete
synchronization [7], generalized synchronization [11],
phase synchronization [12] and anti-phase
D
DAVID PUBLISHING
Chaos Control and Modified Projective Synchronization of Chaotic Dissipative Gyroscope Systems

91
synchronization [13], etc.. In 1999, projective
synchronization has been first reported by Mainieri and
Rehacek [14] in partially linear systems that the master
and slave vectors synchronize up to a constant scaling
factor
o
(a proportional relation). Later, some
researchers have extended synchronization to a general
class of chaotic systems without the limitation of
partial-linearity, such as non-partially-linear systems
[15, 16]. After that, a new synchronization, called
generalized projective synchronization, has been
observed in the chaotic systems [17-19].
Recently, Li [20] considered a new synchronization
method, called modified projective synchronization
(MPS), where the responses of the synchronized
dynamical states synchronize up to a constant scaling
matrix. Therefore, the synchronization and
anti-synchronization can coexist in a chaos
synchronization problem.
The dynamics of a gyro is a very interesting
nonlinear problem in classical mechanics. The concept
of chaotic motion in a gyro was first presented in 1981
by Leipnik and Newton [21], showing the existence of
two strange attractors. Recently, many researches
address the stability, nonlinear phenomena and
technical applications of the gyroscopic system [22,
23]. Gyros for sensing angular motion are used in
airplane automatic pilots, rocket-vehicle
launch-guidance, space-vehicle attitude systems, ships
gyrocompasses and submarine inertial auto-navigators.
Some methods have been presented for the control and
synchronization of the nonlinear gyro system [24, 25].
The variable structure control (VSC) technique is a
discontinuous control strategy that involves, first,
selecting a sliding surface for the desired dynamics and,
secondly, designing a discontinuous control law such
that the system trajectory first reaches the surface and
then stays in it forever [26]. Very recently, several
researchers have used the sliding mode control
technique to control chaotic systems [27-31].
In this paper, the chaos control and modified
projective synchronization (MPS) for chaotic
dissipative gyroscope systems are considered. To
achieve these objectives, the VSC has been proposed.
Control laws for chaos control and MPS of gyroscope
are obtained by VSC, which is proved by the Lyapunov
stability theory.
This paper is organized as follows. In section 2, the
dynamics of a chaotic dissipative gyroscope system is
explained. The chaos control problem and VSC design
to achieve this goal are described in section 3. Also, a
lemma and a theorem for chaos control of dissipative
gyroscope system via VSC are proved in this section.
In section 4, the MPS problem and VSC design to
achieve this goal are described. Also, a theorem for
MPS of dissipative gyroscope systems via VSC is
proved in sections 4. Finally, to show the effectiveness
of the proposed control method for chaos control and
MPS of gyroscope systems, simulations are presented
in section 5. At the end, the paper is concluded in
section 6.
2. Nonlinear Dissipative Gyro Dynamics
The gyroscope contains a mechanical vibration
absorber in the interior in the form of a
spring-mass-dashpot. The absorber mass (m) is
centered on the z axis and position parallel to z axis.
The spring has constant k, and dashpot has damping
constant C. The geometry problem under consideration
is depicted in Fig. 1. The motion of symmetric
gyroscope mounted on a vibrating base can be
described by Eulers angels u , | and . The
Lagrangian can be expressed as [23]:
( )( ) ( )
( )
( )
2
2 2 2 2
1 3
2
2
0
1 1
sin cos
2 2
1
sin cos
2
sin cos ( )
2
g
g
L I mz I
mz M l l t
k
M z l t z l
u | u | u
e u
e u
= + + + +
+ +
+

(1)
where
1
I and
3
I are the polar and equatorial moments
of inertia of the typical gyroscope, g
M
is the gravity
force,
l
is the amplitude of the external excitation
disturbance,
e
is the frequency of the external
excitation disturbance, m is the mass of damper, and k

92

Fig. 1 A schematic diagram of a heavy symmetric
gyroscope.

is the spring constant. It is clear that coordinates
u

and
|
are cyclic, which provides the conjugate
momentum. The momentum integrals are
|
|
and

|
.
Let the state variables [ ]
T
x z p z u u =
, then the
dynamic equations can be rewritten as:
( )
( )
( ) ( )
( )
( )
( )
( ) ( )
( )
1 2
2
1
2
2 3
2
1
1 3
3 1
2
1 3
3 4
2 2
1
4 3 1
2 2
2
1
1 3
2
3 3 2 4
1 cos
sin
sin sin
1 cos
1 cos
sin
2
g g g
g
x x
x
x
x
I m x p
M l m p m x M m l t x
I m x p
x x
x
x x p g x
x
I m x p
k
x x p x cx
m
|
|
|
e
|
(
+ +
(
+ + + +
(

+
(
+ +
| |
| = + +
|
(
+ +
|
(
\ .
+ +
(2)
where p is
0
mg
l
k
. The gyroscope system, Eq. (2),

performs complex dynamics and has been extensively
studied by Ge [23] and Yau [24]. With specific values
set at
1
1 I = , 100 k = , 0.1 l = , 0.5 M = , 0.1 m = ,
0.1 p =
,
2
100
|
| =
, 2 e = ,
2 0.5 c =
and 5 l = in numeric
simulation, the dynamic behavior gyroscope system,
Eq. (2), exhibits an irregular motion as shown in Figs. 2
and 3 with initial conditions of (x
1
, x
2
, x
3
, x
4
) = (-1.2, 1,
-1.1, 0.00905).
Figs. 2 and 3 show that the gyroscope system
trajectories are in a state of chaotic motion.
0 2 4
-5
0
5
Time(s)
x
1
0 2 4
-50
0
50
Time(s)
x
2
0 2 4
-2
0
2
Time(s)
x
3
0 2 4
-50
0
50
Time(s)
x
4

Fig. 2 Time series of
1
x ,
2
x ,
3
x and
4
x .

-5 0 5
-40
-20
0
20
40
x
1
x
2
-2 0 2
-50
0
50
x
3
x
4

Fig. 3 Phase plane trajectories of a chaotic gyro.

In the next section, the chaos control problem of
dissipative gyroscope is described.
3. Chaos Control Problem of Dissipative
Gyros via Variable Structure Control
In the previous section, it has been shown that the
heavy symmetric gyro considered exhibits chaotic
motion. The extreme sensitivity to initial states in a
system operating in chaotic mode can be very
destructive to the system because of unpredictable
behavior. Sometimes, chaos is unwanted or
undesirable.
In order to improve the performance of a dynamic
system or avoid the chaotic phenomena, we need to
control a chaotic system with a periodic motion which
is beneficial for working with a particular condition.
It is thus of great practical importance to develop
suitable control methods. Many researchers have been
focused on this type of problem controlling chaos.
Anti-control of chaos is interesting, nontraditional, and
very challenging [32, 33].

93
For this purpose, in this section, variable structure
control is used as anti-control to chaos control of heavy
symmetric gyroscope.
3.1 Chaos Control Problem
Now, control inputs are introduced in the Eq. (2) for
the second and third states. Thus, the controlled
nonlinear gyro becomes as follows:
( )
( )
( ) ( )
( )
( )
( )
( ) ( )
( )
1 2
2
1
2
2 3
2
1
1 3
3 1
1
2
1 3
3 4
2 2
1
4 3 1
2 2
2
1
1 3
2
3 3 2 4 2
1 cos
sin
sin sin
( )
1 cos
1 cos
sin
2 ( )
g g g
g
x x
x
x
x
I m x p
u t
I m x p
x x
x
x x p g x
x
I m x p
k
x x p x cx u t
m
|
|
|
e
|
+ +
(

(
+ + + +
(

+ +
(
+ +
(

=
| |
|
| = + +
|
(
+ +
|
(
\ .
+ + +
(3)
where
1 2
, u u R e are the control inputs attached in the
nonlinear gyroscope system.
In order to simplify the following procedure, two
nonlinear functions are defined as follows:
( )
( )
( ) ( )
( )
2
1
1 1 3 2 3
2
1
1 3
3 1
2
1 3
1 cos
( , )
sin
sin sin
g g g g
x
f x x
x
I m x p
I m x p
|
|
e
=
(
+ +

(
+ + + +

+
(
+ +

(4)
( )
( )
( )
( ) ( )
2 2
1
2 1 2 3 4 3 2 2
2
1
1 3
2
1 3 3 2 4
1 cos
( , , , )
sin
1 cos 2
x
f x x x x x p
x
I m x p
k
g x x x p x cx
m
|
|
| |
|
= +
|
(
+ + |
\ .
+ + +
(5)
The control problem is to drive the system to track a
four-dimensional desired vector ( )
d
X t as follows:
1 2 3 _ 4
1 1 3 3
( ) [ , , , ]
[ , , , ]
T
d d d d d
T
d d d d
X t x x x x
x x x x

=
=
(6)
which belongs to a class of C function on
0
[ , ) t . Let
us define the tracking error as:
1 2 3 4
1 1 2 1 3 3 4 3
( ) [ ( ), ( ), ( ), ( )]
[ , , , ]
T
T
d d d d
E t e t e t e t e t
x x x x x x x x

=
=
(7)
Then, the error dynamics can be obtained from Eqs.
(8) and (3)-(5) as follow:
1 2
2 1 1 3 1 1
3 4
3 2 1 2 3 4 3 2
( , ) ( )
( , , , ) ( )
d
d
e e
e f x x x u t
e e
e f x x x x x u t
= +
= +

(8)
The control goal considered in this section is that for
any given target orbit ( )
d
X t , the controller is designed
such that the resulting tracking error vector satisfies:
( ) 0
t
E t
lim

(9)
where is the Euclidean norm of a vector.
3.2 Switching Surface and VSC Design
Using the VSC control method to control the chaotic
gyroscope system involves two basic steps:
(1) Selecting an appropriate sliding surface such that
the sliding motion on the sliding manifold is stable and
ensures
( ) 0
t
lim E t
=
.
(2) Establishing a robust control law which
guarantees the existence of the sliding
manifold ( ) 0 S t = . The sliding surfaces are defined as
[34]:
1 2 1 1
( ) ( ) ( ) S t e t e t o = + (10)
2 4 2 3
( ) ( ) ( ) S t e t e t o = + (11)
where ( 1,2)
( )
i
s t R
=
e
and o is a real positive constant.
Differentiating Eqs. (10) and (11) with respect to time
as follow:
1 2 1 1
1 1 2 1 1 1 2
( ) ( ) ( )
( , ) ( ) ( )
d
S t e t e t
f x x x u t e t
o
o
= +
= + +
(12)
2 4 2 3
2 1 2 3 4 3 2 2 4
( ) ( ) ( )
( , , , ) ( ) ( )
d
S t e t e t
f x x x x x u t e t
o
o
= +
= + +
(13)
The rate of convergence of the sliding surface is
governed by the value assigned to parameter
o
.
Having established appropriate sliding surfaces, the
next step is to design VSC to drive the error system
trajectories onto the sliding surfaces. Before stating the
scheme of the controller, the reaching condition of the
sliding mode is given below.
Lemma 1. The motions of the sliding surfaces (10),
(11) are asymptotically stable, if the following reaching
condition is satisfied:

94
( 1,2) ( 1,2)
( ) . ( ) 0
i i
S t S t
= =
(14)
Proof. Let us define the Lyapunov function as:
2 2
1 2
1 1
( ) ( ) ( )
2 2
V t S t S t = +
(15)
According to Lyapunov stability theory, condition
Eq. (14) ensures that
2
1
( ) ( ). ( ) 0
i i
i
V t s t s t
=
=

(16)
Then, ( 1,2)
( )
i
S t
= are the switching surfaces, Eqs. (10)
and (11) are asymptotically stable.
The current variable structure controller design is
described as follows:
Theorem.1: Consider the chaos control problem
represented by Eq. (8). If the control inputs
1
( ) u t and
2
( ) u t are suitably designed as:
1 1 1 1 1 2 1 1 2
( ) ( ( )) ( , ) ( )
d
u t sign S t f x x x e t q o
= + (17)
2 2 2 2 1 2 3 4 3 2 4
( ) ( ( )) ( , , , ) ( )
d
u t sign S t f x x x x x e t q o
= + (18)
where
1 2 1 1
( ) ( ) ( ) S t e t e t o = + ,
2 4 2 3
( ) ( ) ( ) S t e t e t o = + and
( 1,2) i
q
= , ( 1,2) i
o
= are positive constant parameters.
Then the hitting condition Eq. (14) of the sliding
mode is satisfied, and the trajectories of error dynamics
will converge to the sliding surfaces ( 1,2)
( ) 0
i
S t
=
=
.
Proof .1:
Define two sliding surfaces as:
1 2 1 1
( ) ( ) ( ) S t e t e t o = + (19)
2 4 2 3
( ) ( ) ( ) S t e t e t o = + (20)
Differentiating Eqs. (19) and (20) with respect to
time as follow:
1 2 1 1
1 1 2 1 1 1 2
( ) ( ) ( )
( , ) ( ) ( )
d
S t e t e t
f x x x u t e t
o
o
= +
= + +
(21)
2 4 2 3
2 1 2 3 4 3 2 2 4
( ) ( ) ( )
( , , , ) ( ) ( )
d
S t e t e t
f x x x x x u t e t
o
o
= +
= + +
(22)
Define a Lyapunov function as:
2 2
1 2
1 1
( ) ( ) ( )
2 2
V t S t S t = +
(23)
Differentiating Eq. (23) with respect to time, we
have:
1 1 2 2
( ) ( ) ( ) ( ) ( ) V t S t S t S t S t = +

(24)
Substituting Eqs. (21) and 22) into Eq. (24):
1 1 1 2 1 1 1 2
2 2 1 2 3 4 3 2 2 4
( ) ( ).[ ( , ) ( ) ( )]
( ).[ ( , , , ) ( ) ( )]
d
d
V t S t f x x x u t e t
S t f x x x x x u t e t
o
o
= + +
+ + +
(25)
Let
1 1 1 1 1 2 1 1 2
( ) ( ( )) ( , ) ( )
d
u t sign S t f x x x e t q o
= + (26)
and
2 2 2 2 1 2 3 4 3 2 4
( ) ( ( )) ( , , , ) ( )
d
u t sign S t f x x x x x e t q o
= + (27)
where ( 1,2) i
q
= are positive constant parameters. Then
1 1 2 2
( ) ( ) ( ) V t S t S t q q =
(28)
The reaching condition ( 0 SS <
) is always satisfied.
Furthermore, according to Lemma 1, ( 1,2)
( ) 0
i
S t
=

.
Thus, Theorem.1 is proven.
4. Chaos Synchronization Problem of
Dissipative Gyros via VSC
4.1 Modified Projective Synchronization Problem
Consider two coupled, chaotic dissipative gyro
systems, where the master and slave systems are denoted
by x and y , respectively. The master system is shown
in Eq. (2) and the slave system is shown in Eq. (3),
notice that in this equation x has to be denoted by y .
In order to simplify the following procedure, two
nonlinear functions are defined as follows:
( )
( )
( ) ( )
( )
2
2 3
2
1
2
1
1 cos
( , )
sin
sin sin
g g g
g
x
g x z
x
I m z p
M l m p m z M m l t x
I m z p
|
|
e
=
(
+ +

(
+ + + +

+
(
+ +
(

(29)
( )
( )
( )
( ) ( )
2 2
2 2
2
1
2
1 cos
( , , , )
sin
1 cos 2
x
h x y z v z p
x
I m z p
k
g x z z p y cv
m
|
|
| |
|
= + |
( |
+ + |
\ .
+ + +
(30)
Defining the synchronization errors between the
master and slave systems as follows:
1 1 1 1
2 2 1 2
3 3 2 3
4 4 2 4
0 0 0
0 0 0
0 0 0
0 0 0
e y x
e y x
e y x
e y x
o
o
o
o
( ( ( (
( ( ( (
( ( ( (
=
( ( ( (
( ( ( (

(31)
where
1 2
, R o o e are the scaling factors that define a
proportional relation between the synchronized
systems.
Obviously, the modified projective synchronization
is defined as the generalized projective synchronization
if
1
o is equal to
2
o .
The error dynamics can be obtained as:

95
( )
( )
1 2
2 1 1
3 4
4 2 2
1 ( , ) ( )
1 ( , , , ) ( )
e e
e g x z u t
e e
e h x y z v u t
o
o
= +
= +
(32)
The objective of the current control problem is to
design the appropriate control signals
1
( ) u t and
2
( ) u t such that for any initial conditions of the master
and slave systems, the synchronization errors converge
to zero such that the resulting synchronization error
vector satisfies Eq. (9).
In the next section, the controllers are designed to
achieve this objective.
4.2 Switching Surface and VSC Design for MPS
Using the VSC control method to MPS the chaotic
gyroscope system involves two basic steps described in
the section 3.2.
To solve this problem, let us to differentiate Eq. (10)
and (11) with respect to time as follow:
( )
1 2 1 1
1 1 1 2
( ) ( ) ( )
1 ( , ) ( ) ( )
S t e t e t
g x z u t e t
o
o o
= +
= + +

(33)
( )
2 4 2 3
2 2 2 4
( ) ( ) ( )
1 ( , , , ) ( ) ( )
S t e t e t
h x y z v u t e t
o
o o
= +
= + +

(34)
The following theorem shows the properties of the
MPS chaotic dissipative gyroscope systems via VSC.
Theorem.2: Consider the MPS problem represented
by Eq. (32). If the control inputs
1
( ) u t and
2
( ) u t are
suitably designed as:
( )
1 1 1 1 1 2
( ) ( ( )) 1 ( , ) ( ) u t sign S t g x z e t q o o = (35)
( )
2 2 2 2 2 4
( ) ( ( )) 1 ( , , , ) ( ) u t sign S t h x y z v e t q o o = (36)
where
1 2 1 1
( ) ( ) ( ) S t e t e t o = + ,
2 4 2 3
( ) ( ) ( ) S t e t e t o = + and
( 1,2) i
q
= ,
( 1,2) i
o
=
are positive constant parameters.
Then the hitting condition Eq. (14) of the sliding
mode is satisfied, and the trajectories of error dynamics
will converge to the sliding surfaces ( 1,2)
( ) 0
i
S t
=
=
.
Proof .2:
Define two sliding surfaces as:
1 2 1 1
( ) ( ) ( ) S t e t e t o = + (37)
2 4 2 3
( ) ( ) ( ) S t e t e t o = + (38)
Differentiating Eqs. (37) and (38) with respect to
time as follow:
( )
1 2 1 1
1 1 1 2
( ) ( ) ( )
1 ( , ) ( ) ( )
S t e t e t
g x z u t e t
o
o o
= +
= + +

(39)
( )
2 4 2 3
2 2 2 4
( ) ( ) ( )
1 ( , , , ) ( ) ( )
S t e t e t
h x y z v u t e t
o
o o
= +
= + +

(40)
Define a Lyapunov function as:
2 2
1 2
1 1
( ) ( ) ( )
2 2
V t S t S t = +
(41)
Differentiating Eq. (41) with respect to time, we
have:
1 1 2 2
( ) ( ) ( ) ( ) ( ) V t S t S t S t S t = +

(42)
Substituting Eqs. (39) and (40) into Eq. (42):
( )
( )
1 1 1 1 2
2 2 2 2 4
( ) ( ). 1 ( , ) ( ) ( )
( ). 1 ( , , , ) ( ) ( )
V t S t g x z u t e t
S t h x y z v u t e t
o o
o o
( = + +

( + + +

(43)
Let
( )
1 1 1 1 1 2
( ) ( ( )) 1 ( , ) ( ) u t sign S t g x z e t q o o = (44)
and
( )
2 2 2 2 2 4
( ) ( ( )) 1 ( , , , ) ( ) u t sign S t h x y z v e t q o o = (45)
where ( 1,2) i
q
= are positive constant parameters. Then
1 1 2 2
( ) ( ) ( ) V t S t S t q q =
(46)
The reaching condition ( 0 SS <
) is always satisfied.
Furthermore, according to Lemma.1, ( 1,2)
( ) 0
i
S t
=

.
Thus, Theorem.2 is proven.
5. Simulation Results
In this section, numerical simulations are given to
demonstrate chaos control and MPS of the chaotic
dissipative gyros via VSC. The parameters of
dissipative gyros systems are specified as follows:
1
1 I = , 100 k = , 0.1 l = , 0.5 M = , 0.1 m = ,
0.1 p =
,
2
100
|
| =
, 2 e = , 2 0.5 c = , 5 l = .
which, as shown in section 2, give rise to a chaotic state.
To reduce the system chattering, the sign functions in
VSC control are substituted with the saturation
functions.
5.1 Simulation Results of Chaos Control
Numerical simulations are given to demonstrate
tracking control of the dissipative gyro to two desired
trajectories: regular and periodic trajectories. Regular
trajectories are defined as follows:
_
( ) [ , 0, 0.5, 0] [1.047, 0, 0.5, 0]
3
T T
d Regular
X t
t
= =


96
And periodic trajectories are defined as follow:
1
_
2
cos( )
sin( )
( )
sin( )
cos( )
d Periodic
A wt d
Aw wt
X t
A wt d
Aw wt
+ (
(
(
=
( +
(

where 1 A = , 1 w = and
1 2
1.5 d d = = .
The initial conditions are defined as
1 2 3 4
[ (0) (0) (0) (0)] [ 1 1 1 1]
T T
x x x x = .
The time responses of the dissipative gyroscope
controlled to track regular trajectories are shown in
Fig. 4, and the corresponding error states converge
asymptotically to zero in Fig. 5. Also, the sliding
surfaces converge asymptotically to zero in Fig. 6.
The time responses of the dissipative gyroscope
controlled with track periodic trajectories are shown in
Fig. 7, where the corresponding error states converge
asymptotically to zero in Fig. 8. The sliding surfaces
converge asymptotically to zero in Fig. 9.
The simulation results of chaos control of dissipative
gyro via VSC have good performances and confirm that
the error states are asymptotically regulated to zero.

0 2 4 6 8 10
-1
0
1
2

X: 2.267
Y: 1.047
Time(s)
x
1 X
1
X
d1
0 2 4 6 8 10
0
1
2
Time(s)
x
2
0 2 4 6 8 10
-1
0
1
Time(s)
x
3

X
3
X
d3
0 2 4 6 8 10
0
0.5
1
1.5
Time(s)
x
4

Fig. 4 Time responses of state variables (tracking regular
trajectories): trajectories of state variables and desired
states are shown with solid and dashed line, respectively.
5.2 Simulation Results of MPS
Numerical simulations are given to demonstrate MPS
of the chaotic dissipative gyros via VSC.

0 2 4 6 8 10
-4
-2
0
Time(s)
e
1
0 2 4 6 8 10
0
1
2
Time(s)
e
2
0 2 4 6 8 10
-2
-1
0
Time(s)
e
3
0 2 4 6 8 10
0
0.5
1
1.5
Time(s)
e
4

Fig. 5 Time response of tracking error states (tracking
regular trajectories).

0 2 4 6 8 10
-6
-5
-4
-3
-2
-1
0
1
Time(s)
s
1
0 2 4 6 8 10
-0.8
-0.6
-0.4
-0.2
0
Time(s)
s
2

Fig. 6 The sliding surfaces (tracking regular trajectories).

97
0 2 4 6 8 10
-2
0
2
4
Time(s)
x
1

X
1
X
d1
0 2 4 6 8 10
-2
0
2
Time(s)
x
2
0 2 4 6 8 10
-2
0
2
4
Time(s)
x
3

X
3
X
d3
0 2 4 6 8 10
-2
0
2
4
Time(s)
x
4

Fig. 7 Time responses of state variables (tracking periodic
trajectories): trajectories of state variables and desired
states are shown with solid and dashed line, respectively.

0 2 4 6 8 10
-4
-2
0
2
Time(s)
e
1
0 2 4 6 8 10
-1
0
1
2
Time(s)
e
2
0 2 4 6 8 10
-5
0
5
Time(s)
e
3
0 2 4 6 8 10
-1
0
1
2
Time(s)
e
4

Fig. 8 Time response of tracking error states (tracking
periodic trajectories).
0 2 4 6 8 10
-200
-150
-100
-50
0
50
Time(s)
s
1
0 2 4 6 8 10
-150
-100
-50
0
50
Time(s)
s
2

Fig. 9 The sliding surfaces (tracking periodic trajectories).

The scaling factor matrix is specified as:
1
1
2
2
0 0 0 0.8 0 0 0
0 0 0 0 0.8 0 0
0 0 0 0 0 0.4 0
0 0 0 0 0 0 0.4
o
o
o
o
( (
( (
( (
=
( (
( (

The initial conditions are defined as:
1 2 3 4
[ (0) (0) (0) (0)] [ 1.2 1 1.1 0.000905]
T T
x x x x =
1 2 3 4
[ (0) (0) (0) (0)] [0.5 0.5 0.5 0.5]
T T
y y y y =
The time responses of controlled master-slave
chaotic gyros are shown in Fig. 10a and Fig. 10b.
Obviously, the MPS errors converge asymptotically
to zero in Fig. 11. Also, the sliding surfaces converge
asymptotically to zero in Fig. 12. The phase plane of the
master and the slave systems are shown in Fig. 13.
The simulation results of MPS via VSC have good
performances and confirm that the master and slave
systems achieve the modified projective synchronized
states. Also, these results demonstrate that the system
error states are asymptotically regulated to zero.
6. Conclusions
In this paper, the variable structure controls have
been designed to chaos control and modified projective
synchronization of chaotic dissipative gyroscope
systems.
Suppression of the chaos is presented so as to
improve the performance of a dynamical system.

98
0 1 2 3 4
-5
0
5
Time(s)
x
1
Master system
0 1 2 3 4
-5
0
5
Time(s)
y
1
Slave system
o
1
=0.8 and o
2
=0.4
0 1 2 3 4
-50
0
50
Time(s)
x
2
0 1 2 3 4
-50
0
50
Time(s)
y
2
0 1 2 3 4
-2
0
2
Time(s)
x
3
0 1 2 3 4
-2
0
2
Time(s)
y
3
0 1 2 3 4
-50
0
50
Time(s)
x
4
0 1 2 3 4
-20
0
20
Time(s)
y
4

(a)
0 0.5 1 1.5 2 2.5 3 3.5 4
-5
0
5
Time(s)
The Slave and The Proportional Master Systems

o
1
x
1
y
1
0 0.5 1 1.5 2 2.5 3 3.5 4
-50
0
50
Time(s)

0 0.5 1 1.5 2 2.5 3 3.5 4
-2
0
2
Time(s)

o
2
x
3
y
3
0 0.5 1 1.5 2 2.5 3 3.5 4
-20
0
20
Time(s)

o
2
x
4
y
4
o
1
x
2
y
2

(b)
Fig. 10 (a) Time response of the master and the slave
systems (with scaling factors
1 2
0.8, 0.4 o o = = ); (b) Time
response of the slave and the proportional master systems
(with scaling factors
1 2
0.8, 0.4 o o = = ). Trajectories of the
slave and the proportional master systems are shown with
solid and dashed line, respectively.
0 0.5 1 1.5 2 2.5 3 3.5 4
-2
0
2
Time(s)
e
1
0 0.5 1 1.5 2 2.5 3 3.5 4
-20
0
20
Time(s)
e
2
0 0.5 1 1.5 2 2.5 3 3.5 4
-1
0
1
Time(s)
e
3
0 0.5 1 1.5 2 2.5 3 3.5 4
-5
0
5
Time(s)
e
4

Fig. 11 Time response of MPS error states (with scaling
factors
1 2
0.8, 0.4 o o = = ).

0 0.5 1 1.5 2 2.5 3 3.5 4
-200
0
200
400
600
800
Time(s)
1
0 0.5 1 1.5 2 2.5 3 3.5 4
-100
0
100
200
300
400
Time(s)
2

Fig. 12 The sliding surfaces of MPS (with scaling
factors
1 2
0.8, 0.4 o o = = ).

To achieve MPS, it is clear that the proposed method
is capable to create a full range MPS of all state
variables in a proportional way. It also allows us to
arbitrarily adjust the desired scaling by controlling the
slave system.

99
-5 0 5
-50
0
50
x
1
x
2
Phase Plane of Master System
-2 0 2
-50
0
50
x
3
x
4
-2 0 2 4
-50
0
50
y
1
y
2
Phase Plane of Slave System
-1 0 1 2
-20
0
20
y
3
y
4

Fig. 13 Phase plane trajectories of the master and slave
chaotic gyros.

The advantages of this method can be summarized as
follows:
It is a systematic procedure for the control and
synchronization of dissipative gyroscope systems.
The controllers are easy to be implemented.
It is not necessary to calculate the Lyapunov
exponents and the Eigen values of the Jacobian matrix,
which makes it simple and convenient.
Synchronization and anti-synchronization can
coexist in dissipative gyroscope systems.
Simulations results have verified the effectiveness
of this method to control and synchronize chaotic
gyroscopes.
Since the gyro has been utilized to describe the
mode in navigational, aeronautical or space engineering,
the modified projective synchronization procedure in
this study may have practical applications in the future.
References
[1] G. Chen, Controlling Chaos and Bifurcations in
Engineering Systems, CRC Press, Boca Raton, FL, 1999.
[2] E. Ott, C. Grebogi, J.A. Yorke, Controlling chaos, Phys.
Rev. Lett. 64 (1990) 1196-1199.
[3] G. Chen, X. Dong, From Chaos to Order: Perspectives,
Methodologies and Applications, World Scientific Pub.
Co., Singapore, 1988.
[4] J. Feng, C. Xu, J. Tang, Controlling chens chaotic
attractor using two different techniques based on
parameter identification, Chaos, Solitons and Fractals 32
(2007) 1413-1418.
[5] A.M. Harb, A.A. Zahar, A.A. Al Qaisia, M.A. Zohdy,
Recursive backstepping control of chaotic doffing
oscillators, Chaos, Solitons and Fractals 34 (2007) 639-645.
[6] F. Farivar, M.A. Shoorehdeli, M.A. Nekoui, M.
Teshnehlab, Gaussian radial basis adaptive backstepping
control for a class of nonlinear systems, Journal of Applied
Sciences 9 (2) (2009) 248-257.
[7] L.M. Pecora, T.L. Carroll, Synchronization in chaotic
systems, Phys. Rev. Lett. 64 (1990) 821-824.
[8] J.Z. Yang, G. Hu, J.H. Xiao, Chaos Synchronization in
coupled chaotic oscillators with multiple positive
lyapunov exponents, Phys. Rev. Lett. 80 (1998) 496-499.
[9] E.M. Shahverdiev, Synchronization in systems with
multiple time delays, Phys. Rev. E 70 (2004) 067202.
[10] J.F. Xu, L.Q. Min, G.R. Chen, A chaotic communication
scheme based on generalized synchronization and hash
functions, Chin. Phys. Lett. 21 (2004) 1445-1448.
[11] N.F. Rulkov, M.M. Sushchik, L.S. Tsimring, Generalized
synchronization of chaos in directionally coupled chaotic
systems, Abarbanel HDL Phys. Rev. E 51 (1995) 980-994.
[12] M.G. Rosenblutn, A.S. Pikovsky, J. Kurths, Phase
synchronization of chaotic oscillators, Phys. Rev. Lett. 76
(1996) 1804-1807.
[13] L.Y. Cao, Y.C. Lai, Anti-phase synchronism in chaotic
systems, Phys. Rev. E 58 (1998) 382-386.
[14] R. Mainieri, J. Rehacek, Projective synchronization in
three-dimensional chaotic systems, Phys. Rev. Lett. 82
(1999) 3042-3045.
[15] G.L. Wen, D.L. Xu, Nonlinear observer control for
full-state projective synchronization in chaotic
continuous-time systems, Chaos, Solitons and Fractals 26
(2005) 71-77.
[16] G.L. Wen, D.L. Xu, Observer-based control for full-state
projective synchronization of a general class of chaotic
maps in any dimension, Phys. Lett. A 333 (2004) 420-425.
[17] J. Yan, C. Li, Generalized projective synchronization of a
unified chaotic system, Chaos, Solitons and Fractals 26
(2005) 1119-1124.
[18] J.P. Yan, C.P. Li, Generalized projective synchronization
for the chaotic Lorenz system and the chaotic Chen system,
J. Shanghai Univ. 10 (2006) 299-304.
Teshnehlab, Generalized projective synchronization for
chaotic systems via Gaussian radial basis adaptive
backstepping control, Chaos, Solitons and Fractals 42
(2009) 826-839.
[20] G.H. Li, Modified projective synchronization of chaotic
system, Chaos, Solitons and Fractals 32 (5) (2007)
1786-1790.
[21] R.B. Leipnik, T.A. Newton, Double strong attractors in
rigid body motion, Physics Letters A 86 (1981) 63-67.
[22] Z.M. Ge, C.L. Hsiao, Y.S. Chen, C.M. Chang, Chaos and
chaos control for a two degree of freedom heavy
symmetric gyroscope, International Journal of Nonlinear
Science and Numerical Simulation 8 (1) (2007) 89-100.

100
[23] H.K. Chen, Z.M. Ge, Bifurcations and chaos of a
two-degree-of-freedom dissipative gyroscope, Chaos,
Solitons and Fractals 24 (2005) 125-136.
[24] H.T. Yau, Synchronization and anti-synchronization
coexist in two-degree-of-freedom dissipative gyroscope
with nonlinear inputs, Journal of Nonlinear Analysis: Real
World Applications 9 (2008) 2253-2261.
Teshnehlab, Chaos synchronization of uncertain nonlinear
gyros via hybrid control, in: IEEE/ASME International
Conference on Advanced Intelligent Mechatronics, 2009,
pp. 1365-1370.
[26] Y.M. Hu, Variable Structure Control Theory and Its
Applications, Beijing Science Press, 2003.
[27] Q. Zhang, J. Lu, Chaos synchronization of a new chaotic
system via nonlinear control, Chaos, Solitons & Fractals
37 (2008) 175-179.
[28] M. Haeri, A.A. Emadzadeh, Synchronization different
chaotic systems using active sliding mode control, Chaos,
Solitons & Fractals 31 (2007) 119-129.
[29] J.J. Yan, M.L. Hung, T.Y. Chiang, Y.S. Yang, Robust
synchronization of chaotic systems via adaptive sliding
mode control, Physics Letters A 356 (2006) 220-225.
[30] J.J. Yan, Y.S. Yang, T.Y. Chiang, C.Y. Chen, Robust
synchronization of unified chaotic systems via sliding
mode control, Chaos, Solitons & Fractals 34 (2007)
947-954.
[31] H.T. Yang, Chaos synchronization of two uncertain
chaotic nonlinear gyros using fuzzy sliding mode control,
Mechanical Systems and Signal Processing 22 (2008)
408-418.
[32] Z.M. Ge, T.N. Lin, Chaos, Chaos control and
synchronization of electro-mechanical gyrostat system,
Journal of Sound and Vibration 259 (3) (2003) 585-603.
[33] B.A. Idowu, U.E. Vincent, A.N. Njah, Control and
synchronization of chaos in nonlinear gyros via
backstepping design, International Journal of Nonlinear
Science 5 (1) (2008) 11-19.
[34] J.E. Slotine, W. Li, Applied Nonlinear Control,
Prentice-Hall, Englewood Cliffs, New Jersey, 1991.


Investigation of Interline Dynamic Voltage Restorer with
Virtual Impedance Injection
Ahmed Elserougi
1
, Ahmed Hossam-Eldin
1
, Ahmed Massoud
2
and Shehab Ahmed
3

1. Electrical Engineering Department, Alexandria University, Alexandria 21526, Egypt
2. Electrical and Computer Engineering Department, Qatar University, Doha 23874, Qatar
3. Electrical and Computer Engineering Department, Texas A&M University at Qatar, Doha 23874, Qatar

Received: November 12, 2010 / Accepted: March 31, 2011 / Published: January 31, 2012.

Abstract: The inter-line dynamic voltage restorer (IDVR) consists of several voltage source inverters connected to different
independent distribution feeders with common dc bus. When one of the inverters compensates for voltage sag that appears in its feeder
(voltage control mode), the other inverters pump the required power into the dc bus (power control mode). Each inverter will have both
voltage and power controllers; only one controller is in use during the abnormal conditions according to its feeder state. The voltage
controller uses one of the dynamic voltage restoration techniques. In this paper, the in-phase technique is applied and two types of loads
are considered (constant impedance and three phase induction motor). Since the voltage restoration process may need real power
injection into the distribution system, the power controller injects this power via voltage injection. This voltage injection is simulated
by voltage drop across series virtual impedance. A new scheme is proposed to select the impedance value. The impedance value is
selected such that the power consumed by this impedance represents the required power to be transferred without perturbing the load
voltage. The performance of this system is also studied during voltage swell. A scheme for operation of multi-feeder IDVR system is
proposed in this paper. Simulation results substantiate the proposed concept.

Key words: Dynamic voltage restorers, IDVR, voltage sag, voltage swell, power quality.

1. Introduction

Dynamic voltage restorers (DVRs) are very effective
series-compensation devices for voltage sag mitigation.
Due to the increase in using equipments that are
sensitive to voltage variations, the voltage sag problem
becomes one of the important power quality problems.
The voltage sags are caused by sudden increase in
loading, faults or motors starting. Voltage sag
mitigation devices are classified into:
- Conventional solutions using tap changers. These
devices are heavy and bulky; that is why they are rarely
used.
- Uninterruptible power supplies: the main
disadvantage of this method is that the UPS has to carry

Corresponding author: Ahmed Elserougi, Ph.D. student,
research fields: voltage control, power control and voltage
restoration. E-mail: abbas_zone@yahoo.com.
the entire load without any energy contribution from
the grid which is very difficult in high power
applications.
- Static synchronous compensator (STATCOM): It
works by rebuilding the incoming voltage waveform
by switching back and forth from inductive to
capacitive load. If it is inductive, it will supply reactive
AC power. If it is capacitive, it will absorb reactive AC
power. Usually, a STATCOM is installed to support
networks that have a poor power factor and often poor
voltage regulation [1].
- Dynamic voltage restorers: It is a voltage source
converter (VSC) which is connected in series with the
grid. The basic operating principle of the DVR is to
inject an appropriate voltage in series with the supply
through injection transformer whenever voltage sags or
swells take place. The dynamic voltage restorer (DVR)
D
DAVID PUBLISHING
Investigation of Interline Dynamic Voltage Restorer with Virtual Impedance Injection

102
[2, 3] is the most technically advanced and economical
device for voltage-sag mitigation in distribution
systems.
Energy storage units in DVRs are responsible for
supplying the active power component needed during
voltage sag. If this energy is obtained from neighbor
feeder(s) it is called interline dynamic voltage restorer
(IDVR) [4-6]. The IDVR consists of several DVRs
connected to different distribution feeders in the power
system with common dc bus, as shown in Fig. 1. The
voltage-restoration process needs real-power and/or
reactive power injections into the distribution system,
so when one of the DVR compensates for voltage sag,
the other DVRs restore the voltage the common dc bus
via complementing the required energy [5]. Each
inverter will have both voltage and power controllers,
only one controller will be in use during the abnormal
conditions according to its feeder state. The inverter
will be switched to the voltage controller during
voltage sag. However, the inverter will be switched to
the power controller if its feeder voltage was normal
and real power is needed to compensate for sag in
another feeder. The main contribution in this paper can
be summarized in the following points:
- Applying the virtual impedance injection proposed
concept for the operation of power controller.

Fig. 1 Interline dynamic voltage restorer (IDVR) for
n-feeder.
- A scheme for operation of multi-feeder IDVR
system is proposed.
2. Power Controller with Virtual Impedance
Injection for Voltage Sag Mitigation
For inter-line dynamic voltage restorer with two
radial independent feeders, if voltage sag occurred at
one of the feeders, its inverter will compensate for the
voltage sag via injecting suitable voltage (voltage
control mode). Since the voltage-restoration process
needs real power injection into the distribution system,
this real power is decided by the type of
voltage-restoration method [6-8], namely, in-phase
injection, pre-sag supply voltage injection and energy
saving injection. In the first method, the DVR injects a
voltage in phase with the supply voltage. This method
features minimum DVR voltage injection but a phase
jump may introduce problems for loads sensitive to
supply voltage phase jumps. In the second method, the
DVR injects a voltage so as to maintain the load
voltage as the pre-sag supply voltage. In the third
method, the power injected by the inverter is
minimized.
The other inverter will be responsible for restoring
the energy absorbed from the dc bus via injecting
suitable virtual impedance (power control mode).
Assuming that the feeder is feeding a three phase
balanced load, Fig. 2 shows per phase equivalent
circuit of feeder with series virtual impedance injection.
Fig. 3 shows the corresponding phasor diagram.
Since the three phases load is balanced, all the
following analysis will be done for per phase circuit.
The voltage injection is simulated by a voltage drop
across series impedance. The impedance injection must
not perturb the load voltage magnitude. Moreover, the
power consumed by this impedance represents the
required power to be transferred.
To transfer power and to keep the load voltage
magnitude unchanged, the injected voltage must have
two components, namely V
r
which is in phase with
current phasor, and V
x
, which is perpendicular to the

103
P P
L
A +
L
P
P A

Fig. 2 Feeder circuit with virtual impedance injection.

| max
|
2 / ) ( | t

Fig. 3 Phasor diagram of feeder.

current phasor, as shown in Fig. 3. The V
r
component
absorbs active power (P) from source and V
x

component keeps the load voltage magnitude
unchanged. For a given transferred power P, the
corresponding value for the capacitive reactance x
should be estimated from the following relations to
ensure constant load voltage magnitude. The load
power is given by Eq. (1)
). cos( IV P
L
= (1)
The supply power is given by Eq. (2), as indicated in
the phasor diagram shown in Fig. 3.
) cos( IV P P
L
| A = + (2)
Hence
)
IV
P P
( cos
L 1
A
|
+
=

(3)
The maximum value for angle is the load power
factor angle where the supply input power factor
will be unity. Moreover, the maximum allowable P
corresponds to unity input power factor. For a given
load power factor, the maximum value for P is given
by Eq. (4)
)) cos( 1 ( IV P
max
A = (4)
The resistance r, which represents absorbed active
power from feeder, is given by
2
I
P
r
A
=
(5)
From the phasor diagram geometry, Fig. 3, the
capacitive reactance x is given by,
)) ( 5 . 0 tan( | t + + = r x (6)
Fig. 4 shows the relation between P (i.e. r) and x
for different values of load power factor assuming that
load voltage and current magnitudes are 1 p.u.
The locus of unity input pf is part of a circle, as
shown in Fig. 4. This can be proved as follows.
Z I V V L s + = (7)
Substituting with phasor values, yields
) jx r ( 0 1 1 0 1 Z + Z = Z (8)
Hence
Z = + 1 jx ) r 1 ( (9)
Eq. (9) represents a circle equation of radius 1 per
unit and centered at (1, 0) in the complex impedance
plane.
3. Proposed Control Scheme
For given P and measured signals (I, V, pf) a
voltage source inverter is used to inject a suitable
voltage in series with the grid; this injected voltage has
a reference magnitude is equal to (
2 2
. x r I +
) where r,
x calculated using Eqs. (5) and (6) and the injected
voltage phase will lag the supply voltage phase by
angle equal to {(t )/2} (as shown in Fig. 3). The
block diagram of power controller is shown in Fig. 5.
For supply voltage V
s
= 1 p.u, Z
load
= 1 p.u and load
power factors 0.8 and 0.9 lag, with reference power P
= 0.1 p.u for 0.1 < t < 0.2s due to voltage sag in another
feeder, the performance of power controller is shown in
Figs. 6 and 7.
At 0.8 lag load power factor, the virtual injected
resistance and capacitive reactance values will be 0.1
and 0.1641 p.u respectively which verifies the graph
shown in Fig. 4. From results shown in Fig. 6, the
supply power increased with P but the load voltage

104

Fig. 4 Relation between P (or r) and x for different values
of load power factor.

P A
|
2 / t

Fig. 5 Block diagram for proposed control scheme.

Fig. 6 Performance of power controller for 0.8 lag load
power factor; voltages and power in p.u.

Fig. 7 Performance of power controller for 0.9 lag load
power factor; voltages and power in p.u.

V
Load
and load power P
Load
does not change, this power
(P) is fed to the dc bus without affecting the load
voltage magnitude. At 0.9 lag load power factor, the
virtual injected resistance and capacitive reactance will
be 0.1 and 0.4314 p.u respectively which verifies the
graph shown in Fig. 4. The supply power P
s
reached 1
p.u i.e. unity input power factor that means it can not
contribute with more power if more power is needed.
4. Power Controller with Virtual Impedance
Injection for Voltage Swell Mitigation
For IDVR with two radial independent feeders, if
feeder 1 is subjected to voltage swell with normal
voltage at feeder 2, the inverter 1 will be switched to
voltage control mode to compensate for the voltage
swell and the extra power will be fed to the dc bus, the
inverter 2 must absorb this power to avoid the
increasing of dc bus voltage. It means that the P
command to the power controller will be negative, i.e.
virtual negative resistance will be injected (r = P/I
2
).
Inductive reactance must be injected in series with the
virtual negative resistance to guarantee same load
voltage magnitude. The inductive reactance will equal
to {-r.tan [(-+0.5(t-)]} which can be deduced from
the phasor diagram shown in Fig. 8, the angle of
injected voltage will lead the supply voltage by angle
equal to {(t )/2} where in this case (swell
mitigation) is given by Eq. (10).

105
L
P
P A

|

Fig. 8 Virtual impedance injection during voltage swell.

A
|
+
=

)
IV
P P
( cos
L 1
, P is negative (10)
The equations of power controller for sag condition
are given by Eqs. (11) and (12). In this case the P
input command is positive.
)
IV
P P
( cos
L 1
A
|
+
=

(11)
2 / ) ( t | = Z
inj
V
(12)
The equations of power controller for swell
condition are given by Eqs. (13) and (14). In this case
the P input command is negative.
)
IV
P P
( cos
L 1
A
|
+
+ =

(13)
2 / ) ( t | + = Z
inj
V
(14)
The general form of power controller which is valid
in both conditions (sag and swell) is given by Eq. (15).
) ( ). ( 5 . 0 P sign
inj
V A + = Z | t
Where
) ( )}. (
1
cos { P sign
IV
P
L
P
A
A +
= | (15)
The block diagram of general controller is shown in
Fig. 9.
P A
P A
|
2 / t

Fig. 9 General representation of power controller.
5. Case Studies
Case 1: sag condition at feeder 1 with normal
condition at feeder 2; the two feeders fed constant
impedance loads with Table 1.
The simulation results are shown in Fig. 10. The load
voltage and power of load 1 does not change due to
voltage sag in supply 1 because the inverter 1 injects
voltage V
inj1
to maintain constant voltage. The inverter
1 is switched to voltage controller during the sag; this
voltage injection consumes an active power P
1
from the
dc bus. This power is taken from feeder 2 which pumps
an active power P
2
to the dc bus via impedance
injection (P
2
= P
1
), which means that the inverter 2 is
switched to power controller and injects voltage V
inj2
to
pump the required power. The load voltage and power
of load 2 does not change during the compensation.
The supply power P
s1
decreased which means that the
injected voltage acts as voltage source which helps the
supply in feeding the load. The supply power P
s2

increased to feed the load and pump the required power
to the dc bus at the same time.
Case 2: switching between voltage controller and
power controller during different sag conditions; this
case tests the switching between the voltage control
mode and the power control mode at the same feeder;
each inverter will have both voltage and power
controller and according to its feeder state only one

106
Table 1 Simulation data for case 1.
Supply voltage Load (per-phase)
Feeder 1
220 V, with 20% voltage sag
for 0.1 < t < 0.2 s
10 + j10
Feeder 2 220 V 10 + j10

Fig. 10 Case 1 simulation results; voltages in volt and
powers in watt.

controller will be used. The inverter will be switched to
the voltage controller during sag condition. The
inverter will be switched to the power controller if its
feeder voltage was normal but real power is needed to
compensate voltage sag in another feeder. The two
radial independent feeders are shown in Table 2.
In the simulation results the absorbed power will have
positive sign and the pumped power will have negative
sign. The simulation results are shown in Fig. 11.
The switching between voltage and power
controllers during different intervals of the simulation
time can be summarized in Table 3.
Case 3: feeder 1 is feeding an induction motor
(Table 4), and feeder 2 is feeding a constant impedance
load of 10 + j10 ohm per phase.
If the motor starts at t = 0 and is subjected to voltage
sag for 0.3 < t < 0.7 s without any compensation, it will
give the results shown in Fig. 12.
Table 2 Simulation data for case 2.
Feeder 1
220 V, with 20% voltage
sag for 0.1 < t < 0.2 s
10 + j10
Feeder 2
sag for t > 0.3 s
10 + j10

Fig. 11 Case 2 simulation results; voltages in volt and
powers in watt.

Table 3 Summary of case 2 results.
t = 0-0.1 s Inverter(1): Bypassed Inverter(2): Bypassed
t = 0.1-0.2 s
Inverter(1): Switched
to voltage controller
(absorbing power)
Inverter(2): Switched to
power controller
( pumping power )
t = 0.2-0.3 s Inverter(1): Bypassed Inverter(2): Bypassed
t > 0.3 s
Inverter(1): Switched
to power controller
Inverter(2): Switched to
voltage controller

Table 4 Induction motor data.
Rated power (W) 4000
Rated voltage (Volt) 400
Rated frequency (Hz) 50
Rated speed (rpm) 1430
Load torque (Nm) T
Load
= 0.1736
m
Inertia (Kgm
2
) 0.0131
Number of pole pairs 2
Stator resistance (), inductance (mH) 1.4 , 5.839
Referred rotor resistance (), inductance
(mH)
1.4 , 5.839
Mutual inductance (H) 0.1722
Friction factor (Nms) 0.002985

In Fig. 12 the motor speed
m
has dropped during the
voltage sag, when the induction motor speed decreases,
the induction motor slip increases and the rotor current
frequency will increase. When compensation is applied,
the inverter connected to that feeder is switched to the
voltage control mode to inject suitable voltage V
inj
as

107

Fig. 12 Time response of induction motor with no
compensation.

shown in Fig. 13. From the simulation results the
inverter absorbs P
ex
to compensate the voltage sag at
the supply. The motor voltage is approximately
constant during the operation and has not been affected
by the sag i.e. constant speed operation with sag. The
inverter is bypassed during starting period to protect it
from huge starting current.
Case 4: swell condition at feeder 1 with normal
condition at feeder 2; in this case the inverter 1
switched to the voltage controller and inverter 2
switched to the power controller. So if a voltage swell
occurred at feeder 1, the inverter connected to feeder 1
will operate and inject voltage (V
inj1
) out of phase with
load voltage. Feeder 2 will also inject voltage which it
has two components, the first component is responsible
for absorbing the extra energy from dc bus. On the
other hand, the second component is responsible for
maintaining the load voltage within its reference level.
The simulation results are shown in Fig. 14.
6. Proposed Scheme for Multi-feeder System
If there are more than one feeder (assume m-feeder)
used to pump P power to the dc bus, each feeder will

Fig. 13 Time response of induction motor with
compensation.

Time, s Time, s
Fig. 14 Case 4 simulation results.

contribute with certain ratio according to input
command P
k
,k = 1, 2, 3.., m for each feeder. From
measured signals P
k(max)
of each feeder can be
estimated; it is the maximum power which the feeder
can contribute with the dc bus, where
) pf 1 ( V I P
k k k (max) k
= A (16)
P
k
can be calculated from the following proposed
scheme; first of all we must satisfy the following
condition;

108
<
m
1
(max) k
P P A A (17)
Then
=
m
1
(max) k
(max) k
k
P
P
P P
A
A
A A (18)
where k = 1, 2, 3,,m.
Assume 3 feeders connected to loads with 0.75, 0.8
and 0.9 lagging power factor. The maximum power can
be absorbed from the feeders that is equal to
{(1-0.75) + (1-0.8) + (1-0.9)} = 0.55 p.u
For P = 0.3 p.u (i.e. < 0.55) for 0.1 t 0.2 s, each
feeder will contribute with a certain ratio estimated
from Eq. (14). The simulation results are shown in Fig.
15. The simulation result shows the load power P
load
and the supply power P
s
of each feeder, the difference
between them is the pumped power from that feeder to
the dc bus. From simulation this values can be obtained
(P
1
= r
1
= 0.1364, x
1
= 0.1984), (P
2
= r
2
= 0.1091, x
2
= 0.1834) and (P
3
= r
3
= 0.0545, x
3
= 0.1378).
Assume IDVR with five feeders with the following
data:
Feeder 1
sag for 0.1 < t < 0.2 s
10 + j10
Feeder 2
sag for 0.1 < t < 0.2 s
10 + j10
Feeder 3 220 V 14.14 , 0.7 pf lag
Feeder 4 220 V 14.14 , 0.8 pf lag
Feeder 5 220 V 14.14 ,0.85 pf lag
It is obvious from the above mentioned data that,
when feeders 1 and 2 are subjected to voltage sag; it
will absorb power from P dc bus and feeders 3, 4 and
5 will share in pumping this power to the dc bus with
different ratios according to P
max
of each feeder. It is
easy to calculate the ratios from proposed multi-feeder
feeding scheme. The ratios are P
3
= 0.4556P, P
4
=
0.3111P and P
5
= 0.2333P. It is verified by
simulation results as shown in Fig. 16.
7. Conclusions
New control scheme is proposed for the power
controller in IDVR system. The main contribution in
this paper is the operation of power controller with

Fig. 15 Multi-feeder scheme simulation.

Fig. 16 Simulation output for multi-feeder system case
study, powers in watt and voltages in volt.

virtual impedance injection concept. When one of the
feeders is subjected to voltage sag, the voltage
restoration will need real power so the power controller
of the other feeder will inject the required power to
mitigate this sag via voltage injection; this voltage
injection is simulated by a voltage drop across series

109
virtual impedance. The proposed scheme selects the
suitable impedance value. The impedance value is
selected such that the power consumed by this
impedance represents the required power to be
transferred without perturbing the load voltage. The
performance of power controller is studied during
voltage sag and voltage swell, and it performed well in
both cases. Finally, for multi-feeder IDVR system,
when one of the feeders is subjected to voltage sag, the
voltage restoration will need real power so each power
controller of the other feeders will share in injecting the
required power to mitigate this sag. A new algorithm is
proposed which determines participation amount of
each power controller.
Acknowledgments
This publication was made possible by the support of
an NPRP grant from the Qatar National Research
Fund.
References
[1] B. Singh, R. Saha, A. Chandra, K. Al-Haddad, Static
synchronous compensators (STATCOM): A review, IET
Power Electronics 2 (4) (2009) 297-324.
[2] A. Ghosh, G. Ledwich, Compensation of distribution
system voltage using DVR, IEEE Trans. Power Del. 17 (4)
(2002) 1030-1036.
[3] S.S. Choi, B.H. Li, D.M. Vilathgamuwa, Dynamic voltage
restoration with minimum energy injection, IEEE Trans.
Power System 15 (1) (2000) 51-57.
[4] D.M. Vilathgamuwa, H.M. Wijekoon, S.S. Choi, Interline
dynamic voltage restorer: A novel and economical
approach for multi-line power quality compensation, in:
2003 IEEE Industry Applications Conference: 38th IAS
Annual Meeting, Vol. 2, 2003, pp. 833-840.
[5] H.M. Wijekoon, D.M. Vilathgamuwa, S.S. Choi, Interline
dynamic voltage restorer: An economical way to improve
interline power quality, IEE Proceedings Generation,
Transmission and Distribution 150 (5) (2003) 513-520.
[6] D. Mahinda, H.M. Wijekoon, S.S. Choi, A novel
technique to compensate voltage sags in multi-line
distribution system; the interline dynamic voltage restorer,
IEEE Transactions on Industrial Electronics 53 (5) (2006)
1603-1611.
[7] H.F. Ding, S.Y. Shu, X.Z. Duan, J. Gao, A novel dynamic
voltage restorer and its unbalanced control strategy based
on space vector PWM, Electrical Power and Energy
Systems 24 (2002) 693-699.
[8] A. Massoud, S. Ahmed, P. Enjeti, B.W. Williams,
Evaluation of a multilevel cascaded type dynamic voltage
restorer employing discontinuous space vector modulation,
IEEE Transactions on Industrial Electronics 57 (7) (2010)
2398-2410.


Innovative Renewable Energy-Load Management
Technology via Controlled Weight Motion
Mohammed A. El-Kady, Mamdooh Saud Thinyyan Al-Saud and Majeed A.S. Alkanhal

SEC Chair in Power System Reliability and Security, College of Engineering, King Saud University, Riyadh 11421, Saudi Arabia

Received: November 15, 2010 / Accepted: May 13, 2011 / Published: January 31, 2012.

Abstract: The paper presents an innovative renewable energy-load management technology, which is currently being developed as a
prototype, in preparation for potential large-scale implementation in power systems. The new technology employs controlled motion of
large weights, which are being maneuvered (up and down) through the system peak/off-peak demand pattern. The load management
scheme is similar to the pumping storage systems, but uses large weights instead of water and, therefore, is more suited to the operating
environment in Saudi Arabia, in which water is relatively scarce. A portion of the off-peak valley-filling energy is supplied from
wind-driven generators, which adds to the green energy component of the scheme. The controlled motion aspect of the new technology
is based on partial implementation of previously published ideas by the authors as well as a registered patent. The paper describes the
theoretical and analytical features of the introduced renewable energy-load management technology, and dwells on its practical
implementation features.

Key words: Power systems, renewable energy, load management, demand-supply balance.

1. Introduction
Electric power companies around the world have
come to the realization that they can no longer afford
the unlimited expansion for the generation and
transmission facilities in response to the overwhelming,
un-coordinated and ill-managed growth in electricity
demand [1]. Over the past two decades, power system
management has been shifted the focus of their prime
attention from traditional technical system operation
and planning matters to the more pressing issues of
better utilization of available energy resources and
improved quality of existing consumer electricity
services. The results of such a shift in attention and
redefined priorities on the part of electric companies
have been realized in the form of new topics and
subjects, which have emerged, such as load
management and alternate energy sources.

The most successful electric companies today are

Corresponding author: Mohammed A. El-Kady, professor,
research fields: design, operation, control and management of
large-scale power systems. E-mail: melkady@ksu.edu.sa.
those who have managed and adopted an integrated
approach, which combines both utility and consumer
related aspects into one set of coherent and
well-coordinated strategies.
Load management and energy conservation are
currently being considered and adopted by many power
utilities around the world in order to cope with the
increasingly common situation of limited capacity
additions facing, however, a continuously growing
demand [2-5]. On the other hand, renewable energy
technologies have been steadily forthcoming in order
to provide a clean alternative to fossil-fuel-derived
electricity.
In modern times electric power is regarded as one of
the most essential elements in economical and social
development in fast developing countries, especially
the Gulf countries, in which the demand reached
unexpected high level.
In Kingdom of Saudi Arabia the increasing demand
on power is related to several factors:
(1) Use of new building materials and civilization
D
DAVID PUBLISHING
Innovative Renewable Energy-Load Management Technology via Controlled Weight Motion

111
systems obtained from atmospheric and social
environment different from the local environment;
(2) The rich continuous support from the
government to electricity field presented in performing
large electricity projects and interconnections between
main networks;
(3) Increase power consumption levels according to
its availability with prices much lower than the actual
production cost due to government subsidies;
(4) The effect of high temperatures on the extensive
usage of air-conditioning devices during summer.
The increasing demand problem could be faced by
different methods such as:
(1) Adding generating units to face the highly
increasing peak load, especially between one oclock
and five oclock p.m. in summer;
(2) Reduction of electric power consumption by
guiding the customers to use less power according to
the need only. This will decrease customer
consumption bill, decrease operation and investment
costs, save consumed fuel in generating power stations
and decrease the environmental effects resulting from
oil products burning for electricity production;
(3) Electrical load management by shifting the
present electrical loads in maximum consuming time
out of this critical time as much as possible and benefit
from storage and non-electric instruments;
(4) Cut-off of some loads if necessary, even on a
periodic basis with some rewards for consumers.
In the Saudi interconnected electricity system, there
is a large continual increase in electric power
consumption with the maximum load reached during
the summer period. There is also a clear difference in
consumption levels among seasons. Also, difference in
demand levels appears daily according to atmospheric
effects, social conditions and human activities. The
increase and differences in consumption upon annual
and daily level create a style hard to deal with,
especially with the maximum load passes through 3-6
hours per day and the addition of new generating units
to deal with this limited-period load is obviously an
expensive solution and hard to achieve economically.
This paper presents an innovative renewable
energy-load management technology, which is
currently being developed as a prototype at King Saud
University (KSU), in preparation for potential
large-scale implementation in power systems. The new
technology employs controlled motion of large weights,
which are being maneuvered (up and down) through
the system peak/off-peak demand pattern. The load
management scheme is similar to the pumping storage
systems, but uses large weights instead of water and,
therefore, is more suited to the operating environment
in Saudi Arabia, in which water is relatively scarce.
A portion of the off-peak valley-filling energy is
supplied from wind-driven energy schemes [6-10],
which adds to the green energy component of the
scheme. The load management and controlled motion
aspect of the new technology is based on partial
implementation of load management techniques
[11, 12] as well as a registered patent [13]. The paper
describes the theoretical and analytical features of the
introduced technology setup, and dwells on its practical
implementation features.
2. Overall Technological Features
The renewable energy-load management technology
introduced in the paper constitutes the following salient
features:
(1) An efficient design for controlled motion of large
weights, which are being maneuvered, via successive
weight elevation and dropping, to follow certain
desired energy storage/release pattern. The controlled
motion design is currently being developed in the form
of a prototype in preparation for envisaged large-scale
implementation in the power system. The prototype
uses a 1.1 ton weight to be lifted during the off-peak
period via a 3-phase 380 V motor with partial
contribution from a wind-driven mechanism;
(2) A load management scheme, as depicted in Fig. 1,
which aims at reducing the peak load by lifting the
mass weights during the off-peak period and

112

Fig. 1 Illustration of peak shifting and valley filling.

using the stored potential energy to generate electricity
during the following peak period of the demand pattern
and, therefore, replacing the actual generation that
would otherwise have to be used from the power plants.
The load management analysis employs a set of
advanced simulation modules, which maps the
peak/off-peak demand pattern of the system over a
period of time and produces a projected estimate of the
benefits;
(3) A renewable energy scheme, using wind power
as depicted in Fig. 2 in order to supplement a portion of
the off-peak (valley filling) stored energy, which
otherwise would be drawn from the power grid. A

Fig. 2 Wind-driven setup at King Saud University.
hybrid setup involving both wind and solar energy
conversion is also being investigated. The wind energy
setup employs 2.2 m long fans with a 150-250 rpm
rotation speed and a directly applied gearbox.
Load management is undertaken by power utilities
to alter the load shape in order to achieve a better
balance (matching) between the customers cycle
demand and the utilities current and planned
generating as well as transmission and distribution
resources. In the present setup, load flattening is the
prime objective of the load management scheme.
Load flattening scheme transfers loads that would
occur on-peak to off-peak periods, thus combining
peak clipping and valley filling. This can be achieved
by implementing two distinctive load management
functions, namely peak clipping and valley filling.
These two functions are described as follows:
Peak Clipping: Peak clipping aims at reducing the
peak demand on a utility system by decreasing the
on-peak electricity consumption. Let P(t) denote the
system load (MW) as a function of time t (hours),
between O and T. The total saving in energy costs as a
result of peak clipping, to a new clipped power (MW)
of P
c
, is given by

where C
e
is the average cost of energy in Saudi Riyals
per MWh (SR/MWh), and the difference function Q(t)
is given by

In addition to the saving in energy costs as a result of
peak clipping, there is also a saving associated with
reducing system capacity requirement. The
capacity-type is given by

where C
p
is the cost capacity (SR/MW).
Valley Filling: Valley filling is designed to increase
load during off-peak periods. Such action is
appropriate to undertake when the incremental cost of
serving this load is lower than the average cost of
electricity. Let P(t) denote the system load (MW) as a
function of time t (hours), between O and T.
The total energy (MWh) added to increase the valley
portion of the load curve to a new power level of P
v

(MW) is given by

Hours
MW
{ }
Q t
for P t P
P t P for P t P
C
C C
( )
( )
( ) ( )
=
s
>
0
{ } | |
CP C Max P t P
sav P C
= ( )
E R t dt
add
T
=
}
[ ( ) ]
0
CE C Q t d t
sav e
T
=
}
[ ( ) ( )]
0

113
where R(t) is given by

In the present application of load flattening, the
efficiency of the energy reallocation (shift from/to
peak/off-peak) process plays an important role in
assessing the overall merits of the load management
scheme employed. The main objectives of the
application are as follows:
(1) Simulate the load flattening technique for Riyadh
average load curve during the next twenty years as a
case study;
(2) Prove that at the ideal case where the efficiency
of energy shift during the load flattening process is
unity, the optimal clipping point of the peak load at
which a maximum revenue in SR is achieved (due to
saving in future capacity additions) will be at the
average load level;
(3) Find this optimal clipping point in actual cases
where the efficiency of energy shift process is less than
one, which expected to be at load level between the
average and the maximum load depending on the ratio
of loosed energy due to flattening of the load curve
(value of the energy conservation efficiency).
Consider the load-flattening scheme illustrated in
Fig. 3 in which the peak energy E
1
is to be reallocated
to the two valley portions as E
2a
and E
2b
. The energy
reallocation process is assumed to take place with
certain efficiency.
The load flattening action can be simulated by a set
of expressions and relationships. The amount of shifted
energy is given by

or

In the above two expressions, distinction is made
between the case where the utility picks up the losses in
energy reallocation process (utility model) and the case
where the customer bears these losses (customer
model). In both cases, , denotes the efficiency of
energy shift process. The capital costs of generating

Fig. 3 Illustration of load flattening.

capacity according to the original (case 1) and the
shifted (case 2) curves are:

where Cp is the total cost of capacity (SR/MW). The
revenue in each case (SR) can be written as:

where Ce is the cost of the unit of energy in (SR/MWH)
and n is the number of years during the period study.
We now define the two functions F
1
and F
2
as:

where F
1
and F
2
denote the net benefit (revenue-cost) in
each case (SR). Hence, the difference between the two
cases could be written as:

which, according to the simple model considered, is an
indicative of the overall merit of carrying out the load
management program.
Our goal is to find the optimal clipping point (P
C
= P
2
)
at which the function f will reach its maximum positive
value for different values of energy shift efficiency. At
this point (P
C
) the high economical clipping benefit
from the scheme will be achieved.
3. Implementation Scenario
A demonstrative implementation case scenario of
the renewable energy-load management scheme has
been analyzed for a typical daily demand pattern (load
curve) of a portion of the Saudi electricity system.
Controlled motion data from the prototype setup at
KSU was used to obtain the projected range of
P1
P2
E
1

E
2 a
E
2 b
Time (hour)
Load
(MW)
{ }
R t
for P P t
P P t for P P t
V
V V
( )
( )
( ) ( )
=
s
>
0
E E
E
a b 2 2
1
+ =
,
( ) Utility model
E E E
a b 2 2 1
+ =

, ( ) Customer model
Cost C P
Cost C P
p
p
1 1
2 2
=
=
Re [ ] (8760 )
Re [ ] (8760 )
v C E E n
v C E E E n
e
e o a b
1 0 1
2 2 2
= +
= + +
F v Cost
F v Cost
1 1 1
2 2 2
=
=
Re
Re
f F F =
2 1

114
efficiency as well as the time-scale for the up and down
motions. Renewable energy sources in the form of
wind power generators are envisaged to account for
about 25% of the off-peak energy storage (valley filling)
in this implementation scenario.
A computerized simulation module was used to
simulate the average daily load curve of the considered
subsystem as shown in Fig. 4. The simulation module
was run at different values of energy shift efficiency
and the benefit in MSR (million Saudi Riyals) due to
load flattening of the load curve was found at certain
points specified by the software module. The module
scans the load curve starting from average load until
the max load is reached, with closed increment factor
of 0.05 MW. The relation between the benefit and
clipping level was then plotted for each value of energy
shift efficiency.
Two business models were used in the present load
management analysis depending on who (supplier of
customer) bears the cost of losses. Fig. 5 shows the
relation between the benefit and clipping level at
energy shift efficiency equal to 1 for the
customer-bear-losses scenario.
In this case the maximum benefit was achieved when
the load was clipped at the average load level. We can
therefore say that, at the ideal energy shift (efficiency =
1), P
C

optimal
= P
avg =
4767.7 MW. Fig. 6 shows the
relation between the benefit and clipping level at
energy shift efficiency equal to 0.9. In this case the
maximum benefit was achieved at P
C
= 4831.6 MW.
Fig. 7, on the other hand, shows the relation between
the benefit and clipping level at energy shift efficiency
equal to 0.8, where the maximum benefit was achieved
at P
C
= 4980.8 MW.
Both Figs. 6 and 7 are drawn for the
customer-bear-losses scenario. When the
utility-bear-losses scenario is considered, Figs. 8, 9
and 10 show the relation between the benefit and
clipping level at energy shift efficiency equal to 0.9,
0.85 and 0.8, respectively. It is to be noted that the
negative sign appearing in the benefit-clipping level
diagrams is not due to load flattening per se, but rather
due to the energy shift process itself.

0 5 10 15 20 25
4000
4200
4400
4600
4800
5000
5200
5400
Time (HOURS)
D
e
m
a
n
d

(
M
W
)

Fig. 4 Load curve of the implementation case scenario.

4700 4800 4900 5000 5100 5200
0
1
2
3
4
5
6
7
8
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

(
S
R
)

Fig. 5 Relation between benefit and clipping point at unity
shift efficiency (customer model).

4700 4800 4900 5000 5100 5200
0
0.5
1
1.5
2
2.5
3
3.5
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

Fig. 6 Relation between benefit and clipping point at 0.9

115
4700 4800 4900 5000 5100 5200
-1
-0.5
0
0.5
1
1.5
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

Fig. 7 Relation between benefit and clipping point at 0.8

4700 4800 4900 5000 5100 5200
0
0.5
1
1.5
2
2.5
3
3.5
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

(
S
R
)

Fig. 8 Relation between benefit and clipping level at 0.9
energy shift efficiency (utility model).

4700 4800 4900 5000 5100 5200
-1
-0.5
0
0.5
1
1.5
2
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

(
S
R
)

Fig. 9 The relation between benefit and clipping level at
0.85 energy shift efficiency (utility model).

The negative sign in the benefit-clipping level
diagrams simply indicates that the losses of benefits
due to energy losses are greater than the revenue to
load flattening. It starts appearing in Fig. 10 where the
4700 4800 4900 5000 5100 5200
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
x 10
6
Clipping Level (MW)
N
e
t

B
e
n
e
f
i
t

d
u
e

t
o

L
o
a
d

F
l
a
t
t
e
n
i
n
g

(
S
R
)

Fig. 10 The relation between benefit and clipping level at
0.8 energy shift efficiency (utility model).

efficiency of energy shift process equal to 0.8, and the
margin of negative benefit-clipping level increases as
the efficiency decreases. Because of this reason, the
margin of benefit due to the load flattening process is
very narrow (positive margin). Hence, for the specific
demand pattern analyzed, we can conclude that for
energy shift process efficiency less than 60%, the load
flattening process would not be beneficial.
The maximum net benefit corresponding to the
optimal clipping level for the chosen value of energy
shift efficiency is summarized in Table 1. It is to be
noted that the benefits are calculated for the particular
daily load curve under study.
The same analysis can easily be repeated for all other
days in the year using the associated daily demand
patterns in order to obtain an estimate of the yearly
savings. The clipping level P
C
is controlled in practice
by the intensity of applying the controlled weight
motion scheme during the up-motion (off-peak valley
filling) and down-motion (peak clipping). It is also
noted that the renewable energy portion of the off-peak
valley fillingif applied in the real-time
framewould normally depend on the availability of
the wind energy and, therefore, should be designed as
an optional complementary power. In this regard,
energy storage schemes could be examined for more
flexible usage of the renewable energy component of
the load management process.

116
Table 1 Maximum net benefit and clipping level
according to energy shift efficiency.
Energy shift
efficiency ef (%)
Clipping level
Pc (MW)
Maximum net benefit
fmax (MSR)
60 5130.1 0.3273
70 5066.1 0.666
80 4980.8 1.4328
90 4831.6 3.4367
100 4767.7 7.4418
4. Conclusions
This paper has presented an innovative renewable
energy-load management technology, which is currently
being developed as a prototype at King Saud University
(KSU), in preparation for potential large-scale
implementation in power systems. The new technology
employs controlled motion of large weights, which are
being maneuvered (up and down) through the system
peak/off-peak demand pattern. The load management
scheme is similar to the pumping storage systems, but
uses large weights instead of water and, therefore, is
more suited to the operating environment in Saudi
Arabia, in which water is relatively scarce. A portion of
the off-peak valley-filling energy is supplied from
wind-driven generators, which adds to the green energy
component of the overall load management scheme.
The implementation case scenario presented in the
paper for a typical daily demand pattern (load curve) of
a portion of the Saudi electricity system has
demonstrated that the controlled weight motion
technology can be employed as an effective load
management scheme in practical system operation.
Furthermore, there are two important factors which
can improve the net benefit achieved from the controlled
weight motion technology scheme:
(a) Increasing the efficiency of the energy shift process.
This can be achieved by improving the efficiency of the
individual controlled motion components, including the
gearbox, shaft-couplers, etc.;
(b) Operating at, or close tothe optimal clipping
level of the peak load at which the maximum benefit is
attained from the controlled weight motion technology.
Acknowledgments
This work was supported by the Saudi Electricity
Company (SEC) which funds the SEC Chair in Power
System Reliability and Security.
References
[1] M.A. El-Kady, Important considerations and overlooked
issues in developing and implementing energy conservation
programs, in: Proceeding of the 7th Saudi Engineering
Conference, Riyadh, Saudi Arabia, Nov. 26-28, 2007.
[2] C. Alvarez, R.P. Malhame, A. Gabaldon, A class of
models for load management application and evaluation
revisited, IEEE Transactions on Power Systems 7 (4)
(1992) 1435-1443.
[3] M.A. El-Kady, A.M. Shaalan, Impact of load management
and energy conservation strategies on the environment and
operating practices of power systems, in: Proc. of The 33rd
Intersociety Energy Conversion Engineering Conference
(IECEC), Colorado, USA, 1998.
[4] D. Ha, S. Ploix, E. Zamai, M. Jacomino, Real times
dynamic optimization for demand-side load management,
International Journal of Management Science and
Engineering Management 3 (4) (2008) 243-252.
[5] B. Thomas, Load Management Techniques, in:
Proceedings of the IEEE Southeastcon 2000, Nashville,
TN, USA, Apr. 7-9, 2000, pp. 139-145.
[6] T. Burton, D. Sharpe, N. Jenkins, E. Bossanyi, Wind Energy
Handbook, 1st ed., John Wiley & Sons, England, 2001.
[7] E.I. Baring-Gould, L. Flowers, P. Lundsager, L. Mott, M.
Shirazi, J. Zimmerman, Worldwide status of wind/diesel
applications, in: Proc. of the 2003 AWEA Conference,
Austin, TX, June, 2003.
[8] R. Howk, Utility plans bird point wind farm, Alaska Journal
of Commerce [Online], 27 (22) (2003), Web posted:
http://classic.alaskajournal.com/stories/060203/loc_2003060
2001.shtml.
[9] J.F. Manwell, J.G. McGowan, A.L. Rogers, Wind Energy
Explained: Theory, Design and Application, John Wiley &
Sons, Ltd., 2002.
[10] M.A. El-Kady, Overlooked issues in energy conservation
drives, in: Proc. of Energy Conservation Conference,
Ministry of Industry and Electricity, Riyadh, Kingdom of
Saudi Arabia, Nov. 14-15, 1999.
[11] S.A. Al-Rumaih, M.A. El-Kady, Simulation of load
management and energy conservation strategies in
interconnected power systems, in: Proc. of the IASTED
International Conference on Applied Simulation and
Modeling (ASM 2005), Benalmadena, Spain, June 15-17,
2005.
[12] M.S. Al-Saud, M.A. El-Kady, A.M. Shaalan, Design of
optimal load management strategies in electric power
systems, in: Proc. of the ASTED International Conference
on Applied Modelling and Simulation, Cairns, Queensland,
Australia, Sept. 1-3, 1999.
[13] M. Alkhamis, Energy generation system using differential
elevation, US Patent, 5873249 (1999).

A Parameter Determination Method of Distribution
Voltage Regulators Considering Tap Change and
Voltage Profile
Yuji Hanai
1
, Yasuhiro Hayashi
2
, Junya Matsuki
1
, Yoshiaki Fuwa
3
and Kenjiro Mori
3
1. Department of Electrical and Electronics Engineering, University of Fukui, Fukui 910-8507, Japan
2. Graduate School of Advanced Science and Engineering, Waseda University, Tokyo 169-8050, Japan
3. Distribution Department, Power Network Division, Tokyo Electric Power Company, Tokyo, Japan


Abstract: This paper proposes a parameter determination method of distribution voltage regulators load ratio control transformers
(LRT) and step voltage regulators (SVR) considering the tap change and voltage profile. The method takes two procedures in order to
simplify the optimization problem and to reduce calculation time. One is to simultaneously determine the control parameters of LRT
and SVR minimizing voltage violations and voltage variations. The algorithm is based on particle swarm optimization (PSO), which is
one of non-linear optimization methods by using a concept of swarm intelligence. Another is to determine the dead-band width of LRT
and SVR on the basis of bi-evaluation of tap change and voltage margin. The concept of a Pareto optimal solution is used for the
decision of the best dead-band width. As the results of numerical simulations using distribution network model, the validity of the
proposed method has been affirmed.

Key words: Distribution network, voltage control, LRT, SVR, control parameter, deadband, PSO.

1. Introduction

Distribution line voltages have been mainly
controlled by load ratio control transformers (LRT) at
distribution substations and step voltage regulators
(SVR) on distribution lines. As a voltage control
method of these voltage regulators, the line drop
compensation method (LDC method) is generally used,
which adjusts automatically the sending voltage of
these voltage control regulators in proportion to the
load current. Recently, a number of distributed
generator (DG) such as photovoltaic system and wind
turbine generation system has been installed in a
distribution network. The complication of the voltage
distribution by reverse power flow from DG makes the

Corresponding author: Yuji Hanai, Ph.D., research fields:
operation and control of distribution system with distributed
generations. E-mail: hanai@u-fukui.ac.jp.
voltage control more difficult. In the past framework in
determining control parameters of voltage regulators,
the influence of connection of DG has not been
considered. Therefore, a development of a new
computation method to determine the control
parameters is needed to flexibly control voltage
distribution complicated by reverse power flow from
DG. On the other hand, sectionalizing switches with
sensor, which are called as IT switch, have been
developed and these are installed in a part of actual
distribution feeders. By the penetration of IT switch,
the complicated power flow in distribution network
with DG can be estimated accurately.
Many researches for the maintenance of a proper
voltage have been done [1-11]. The authors had also
proposed a computation method to determine the
control parameters of voltage regulators such as LRT
and SVR installed in the distribution network with DGs
D
DAVID PUBLISHING
A Parameter Determination Method of Distribution Voltage Regulators
Considering Tap Change and Voltage Profile

118
and IT switches [12]. The proposed algorithm can
determine simultaneously control parameters of LRT
and SVR based on particle swarm optimization (PSO),
which is one of non-linear optimization methods by
using a concept of swarm intelligence [13, 14]. The
validity of the proposed method is affirmed by both
simulations and experiments. However, the proposed
method does not target the reduction of daily tap
change of LRT and SVR although it is desirable for the
life extension. For reducing the tap change, an addition
of the dead-band width adjustment should be
considered. Since there is a trade-off problem between
reducing a tap change and keeping a voltage
margin on the determination of the dead-band width,
utility needs to suppress the tap change as much as
possible maintaining a proper voltage. Additionally, in
the parameter determination problem for LRT and
SVR, considering the tap change and the deadband
would yield an increase of the calculation time and the
decrease in the solution accuracy.
This paper proposes a parameter determination
method of LRT and SVR for reducing the daily tap
change and keeping voltage profile. The method takes
two procedures in order to simplify the optimization
problem and to reduce calculation time. One is to
simultaneously determine the control parameters of
LRT and SVR minimizing amount of voltage violation
and voltage variation. The algorithm is based on PSO
and optimizes voltage profile with the control
parameters but does not consider a reduction of the tap
change. Another is to determine the dead-band width of
LRT and SVR on the basis of bi-evaluation of tap
change and voltage deviation. The concept of a Pareto
optimal solution is used for the determination of the
best dead-band widths. By numerical simulations using
distribution network model, the validity of the
proposed method is affirmed.
2. Parameter Determination Method of
LRT/SVR
For stable operation and control of distribution
system, control parameters (LRT: I
MIN
, V
MIN
, I
MAX
,
V
MAX
, SVR: R
REF
, X
REF
, V
REF
) and deadband widths
(LRT:
LRT
, SVR:
SVR
), as shown in Fig. 1, must be
determined without voltage violation and the excess
tap change. The parameter determination problem for
LRT and SVR is a complex combination optimization
problem to determine the control parameters and the
deadband widths. Thus, it is necessary to find an
effective solution, which is a combination of the
control parameters and the deadband widths without
voltage violation and the excess tap change, at
practicable calculation time.
The proposed method has two procedures and
solves separately two determination problems for the
control parameters and the deadband widths. First, the
control parameters are optimized for improving
voltage profile by optimal power flow calculation
based on PSO. The deadband widths are set to the
minimum value, which is one-half of each tap voltage.
After that the deadband widths are optimized
considering voltage profile and tap change counts of
LRT and SVR with the optimized control parameters.
Tap change counts and voltage deviation of each
deadband width combinations are evaluated and the
optimum combination is determined using concept of a
Pareto optimal solution.
2.1 Optimizing Control Parameters of LRT/SVR
The control parameters of LRT and SVR are (I
MIN
,
V
MIN
, I
MAX
V
MAX
) and (R
REF
, X
REF
, V
REF
), respectively.
Since the voltage distribution of distribution feeder is
varied by the change of the control parameters, these
values must be determined without voltage violations.
The determination problem of control parameters of
LRT and SVR is defined as optimization problem
minimizing the amount of voltage violation and
deviation from voltage limit. The formulation is shown
as follows.
[Objective Function]
( ) ( ) ( ) Minimize E D F + = x x x
(1)
(Voltage deviation from upper and lower voltage limit)

119
Substation
Heavy Load Period
LRT SVR
Line Length
L
i
n
e

V
o
l
t
a
g
e

Control Characteristic
IMIN IMAX
VMAX
VMIN
IB
VS
Upper Limit
Lower Limit
Light Load Period
Control Characteristic
VREF
IB
VS
: Sectionalizing Switch
(Closed)
: Sectionalizing Switch
(Opened)
Gradient RREF+jXREF
LRT SVR

Fig. 1 Outline of distribution voltage control by LRT and
SVR.

( ) { ( ( ) ) ( ( ) ) }
2
,
2
,
1 1
j t n j t n
T
t
N
n
V V V V D + =
= =
x x x
(2)
(Voltage violation penalty)
( )
( ( ) ) ( ( ) )
( ( ) )
( ( ) ) ( ( ) )
>
s s
<
=
n t n j t n
n t n n
n t n j t n
V V V V
V V V
V V V V
E
x x
x
x x
x
,
2
, 2
,
,
2
, 1
0
o
o

(3)
(State variable)
| |
T
REF REF REF MAX MAX MIN MIN
V X R V I V I

= x
(4)
[Constrints]
MAX MIN V V <

(5)
MAX MIN
I I <

(6)
where, x: a set of control parameters of LRT and SVR
(discrete variable), I
MIN
, V
MIN
, I
MAX
, V
MAX
: control
parameters of LRT, R
REF
, X
REF
, V
REF
: control
parameters of SVR, D: deviation of voltage from upper
and lower limit, E: amount of voltage violation, V
n,t
:
node voltage at node n at time t, n n
V V ,
: upper and lower
limit of node voltage at node n,
1
,
2
: penalty
coefficient of voltage limit violation, N: total node
number ( N ne ), T: total time index ( T t e ).
Once certain control parameter vector x
i
* (= [I
MIN
*

V
MIN
*
I
MAX
*
V
MAX
*
R
REF
*
X
REF
*
V
REF
*
]
T
) of LRT and
SVR is given under the constraints of control
parameters range shown in Eqs. (5) and (6), the
objective function value represented in Eq. (1) can be
calculated by running the power flow calculation in all
time periods. Namely the voltage profiles controlled by
LRT and SVR can be calculated by using the
measurement data from IT switches in each time period.
In this study, the optimal control parameters of LRT
and SVR are searched by iterating the evaluation of
Eqs. (1)-(4) using particle swarm optimization (PSO),
which is one of non-linear optimization methods by
using a concept of swarm intelligence [13, 14]. PSO
simulates behavior that swarm of living matters such as
bird searches food sharing information. A particle
(agent) has the position vector and the velocity vector.
The i-th agent changes their position vector x
i
and
velocity vector v
i
based on two information pbest
i
and
gbest. Where, pbest
i
is the current best position vector
of the i-th agent, and gbest is the current best position
vector in the swarm of agents. The position vector is
related to a solution of the problem when we want to
solve an optimization problem. PSO has been applied
to various optimization problems so far. For iteration k,
present position vector x
i
(k)
of agent i is updated to the
next position vector x
i
(k+1)
by Eqs. (7)-(9). The weight
coefficient w
1
is agents inertia parameter and is
gradually decreased by Eq. (9).
( ) ( ) ( ) 1 1 + +
+ =
k
i
k
i
k
i
v x x

(7)
( ) ( )
(
( )
)
k
i i
k
i
k
i
rand w w x pbest v v + =
+
2 1
1

(
( )
)
k
i
rand w x gbest +
3

(8)
k
k
w w
w w
MAX
MIN MAX
MAX

=
1

(9)
where, x
i
: current position vector of the i-th agent, v
i
:
current velocity vector of the i-th agent, pbest
i
: current
best position vector of the i-th agent, gbest: current best
position vector in the swarm of agents, rand: uniform
random number from 0 to 1, w
1
, w
2
, w
3
: weight
coefficient of each term, w
MAX
, w
MIN
: coefficients of
inertia term, k: iteration count.
In this method, the various control parameter vector
x
i
* (= [I
MIN
*
V
MIN
*
I
MAX
*
V
MAX
*
V
REF
*
Z
REF
*
]
T
) is
determined by using modification of position vector of
PSO represented in Eqs. (7)-(9), which also should be
satisfied with the constraints of control parameters
range shown in Eqs. (5) and (6). And each agent of
PSO searches the optimal control parameter vector
(namely gbest used in PSO) in solution space with
dimensions of the numbers of control parameters. The
determination procedures of the optimal control
parameters of LRT and SVR are shown as follows.

120
[Step 1] Set initial value of position vector x
i
and
velocity vector v
i
of each agent i (I = 1, 2,, n ) at
random.
[Step 2] Set iteration count k to 1.
[Step 3] Set agent number i to 1.
[Step 4] Give control parameters of each voltage
regulator as discrete values after discretization of each
element of position vector x
i
.
[Step 5] Estimate each load power in all time
period in proportion to the contract demand by use of
the measurement data from IT switches.
[Step 6] Calculate voltages at all nodes in each
time period t by power flow calculation using the
estimated load power.
[Step 7] Calculate objective function value F(x
i
)
in each time period t by Eq. (1).
[Step 8] If objective function value F (x
i
) with
agent i is less than F(pbest
i
), save position vector x
i
as
pbest
i
. If the objective function value F(x
i
) with agent i
is less than F(gbest), save position vector x
i
as gbest.
[Step 9] Update position vector x
i
and velocity
vector v
i
by Eqs. (7)-(9).
[Step 10] If control parameters of all agent have
been evaluated based on Eq. (1), go to step 11. If or not,
update agent number i to i+1, and then return to step 4.
[Step 11] If iteration count k reaches the maximum
iteration count k
max
, go to step 12. If or not, update
iteration count k to k+1, and then return to step 3.
[Step 12] Determine gbest with minimum
objective function value F(x
i
) as the optimal control
parameters of voltage regulators (LRT and SVR).
2.2 Optimizing Deadband Width of LRT/SVR
In the proposed method, deadband widths (LRT:
LRT
, SVR:
SVR
) are determined with the dual aims of
reducing the tap change and keeping voltage margin.
The determination problem of deadband widths is
defined as optimization problem minimizing the total
tap change counts and the mean target voltage
deviation. The formulation is shown as follows.
[Objective Function]
( ) ( ) ( ) ( ) Minimize F w F w F 1 2 1 + = x x x

(10)
(Total tap change count of LRT and SVR)
( ) ( )( ( ) ( ) ) x x x x 1
0 0 0
1
= = =
=

t , p , m t , p , m t , p , m
T
t
P
p
M
m
c c c F

(11)
(Average target voltage deviation)
( ) ( ) ( ) 2
1
0 0
2 n n t , n
T
t
N
n
V V V
T N
F +
=

= =
x x

(12)
(State variable)
| |
T
M m

1 0
c c c c = x

(13)
[Constrints]
n t , n n V V V s s

(14)
Where, x: a set of deadband widths of LRT and SVR
(discrete variable), c
m,p,t
: connection state to tap
position p of voltage regulator m at time t (0: nonuse, 1:
use),
m
: deadband width of voltage regulator m, F
1
:
total tap change counts of LRT and SVR, F
2
: average
target voltage deviation, V
n,t
: node voltage at node n at
time t, n n
V V ,
: upper and lower limit of node voltage at
node n, N: total node number ( N ne ), M: total voltage
regulator number ( M me ), P: total tap number ( P pe ),
T: total time index ( T t e ).
As the formulation, different two physical
quantities are evaluated and optimized. However,
there is a relationship of trade-off between total tap
change counts and average target voltage deviation so
that it is impossible to minimize both these things.
Therefore, the determination problem of optimal
deadband width is multi-objective optimization
problem with the dual aims of minimizing total tap
change counts and average target voltage deviation. A
solution set, as shown in Fig. 2, that the judgment as to
the superiority or inferiority cannot be passed is called
a Pareto optimal solution, the concept is used on
determining the optimal deadband width. The
determination procedures of the optimal deadband
widths of LRT and SVR are shown as follows.
[Step 1] Input distribution network data and
optimized control parameters of LRT and SVR.
[Step 2] Create all the combinations x of
deadband widths of LRT and SVR with the actual
specification.

121
F1
F2
Non-dominated solution
(Pareto optimal solution)
Dominated solution
Pareto front

Fig. 2 An example of Pareto optimal solutions.

[Step 3] Arrange and number the deadband
combinations x
i
(I = 1, 2, , n ) in ascending order.
[Step 4] Set the combination number i to 1.
[Step 5] Calculate total tap change counts F
1
(x
i
*)
and average target voltage deviation F
2
(x
i
*) by power
flow calculation using the optimized control
parameters.
[Step 6] If x
i
* (= [
LRT
*
SVR
*
]
T
) is a
non-dominated solution, save x* as a Pareto optimal
solution.
[Step 7] If all the combinations are evaluated, go
to step 8. If or not, update the combination number n to
n+1, and then return to step 5.
[Step 8] The Pareto optimal solution suitable for
distribution operators needs (for example, norm
minimum, w = 1 or w = 0) is decided as an optimal
combination of deadband width of LRT and SVR.
3. Numerical Simulation
In order to affirm the validity of the proposed
parameter determination method for LRT and SVR,
the optimal control parameters and the deadband
widths are determined in a distribution network model.
In addition, when the principal objective of parameter
determination is the tap change reduction (w = 1) and
the voltage deviation reduction (w = 0), the profiles of
tap position and line voltage are compared.
3.1 Simulation Conditions
The distribution network model is shown in Fig. 3.
The distribution network model consists of 2
distribution feeders, 24 load connection nodes, 1 LRT
branch and 1 SVR branch. Daily load curve in each
distribution feeder is shown in Fig. 4. The typical load
profiles of industrial area and residential are applied to
the loads in feeder A (Node No.4~No.15) and feeder
B (Node No.17~No.28), respectively. The power
factor of all the loads is set as 1.0. The line length
between each node is 0.5 km in feeder A and 1.0 km in
feeder B. The impedance of distribution line between
each node is 0.252+ j0.348 /km. The proper range of
line voltage is from 6474 V-6726 V at all the load
connection nodes. The tap voltage of LRT and SVR is
30 V (21 taps). For simplicity, it is assumed that
sending voltage, line impedance, and load current
balance on three-phase circuit.
As to PSO parameters, maximum iteration count is
100, the number of searching agent is 20, and weight
coefficients are w
MIN
= 0.4, w
MAX
= 0.9, w
2
= 2.0, w
3
=
2.0. On the optimized calculation of the control
parameters, the penalty coefficients of voltage limit

0 1 2
3 4 5 6 7 8 9 10 11 12
14
15
16 17 18 19 20 21 22 23 24 25
28
LRT
Branch
(21 taps)
Slack
Node
(6600V)
13
26
27
SVR
Branch
(21 taps)
Load Connection Node
Feeder A
Feeder B

Fig. 3 A distribution network model.

122
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
P
o
w
e
r
[
p
u
]
Tim e [hour]
Feeder A
Feeder B

Fig. 4 Daily load curves of distribution feeders.

violation are
1
=
2
= 10
4
. Searching range of control
parameters and deadband widths of LRT and SVR is
shown in Table 1.
3.2 Determination of the Control Parameters
The optimal control parameters of LRT and SVR are
shown in Table 2. The objective function value is
small and the control parameters without voltage
violation have been obtained. The profile of objective
function value in iterative calculation by PSO is
shown in Fig. 5. A combination of control parameters
without voltage violation is found on 2nd searching,
and that of optimal control parameters is on 70th
searching. The calculation time is about 20 minutes on
using a personal PC (CPU: Intel Core2 Duo
Processor E8500 3.17 GHz, Memory: 2.00 GB).
3.3 Determination of the Deadband Widths
The Pareto optimal solutions of deadband widths of
LRT and SVR are shown in Fig. 6. As the simulation
result, 10 combinations of the deadband width are
obtained as the Pareto optimal solution. The
calculation time is about 38 seconds. When
distribution system utility attaches importance to a
reduction of the tap change of LRT and SVR, the
deadbands of LRT and SVR are set widely. In the
opposite, when a reduction of the voltage variation is
more important, the deadbands are set narrowly.
Using the minimum deadbands widths, the average
voltage deviation is minimum value. Therefore,
Table 1 Searching range of control parameters and
deadband width of LRT and SVR.
LRT
I
MIN
(A) 90 ~ 160
V
MIN
(V) 170 ~ 240
I
MAX
(A) 6450 ~ 6750
V
MAX
(V) 6600 ~ 6900
LRT
(%) 0.3 ~ 1.0
SVR
R
REF
(%) 1.0 ~ 24.0
X
REF
(%) 1.0 ~ 24.0
V
REF
(V) 6450 ~ 6750
SVR
(%) 0.3 ~ 3.0

Table 2 Calculation result of control parameters of LRT
and SVR.
LRT
I
MIN
(A) 99
V
MIN
(V) 185
I
MAX
(A) 6529
V
MAX
(V) 6696
SVR
R
REF
(%) 8.0
X
REF
(%) 7.0
V
REF
(V) 6600
Objective function value 0.04966

expanding the deadband of LRT and SVR involves a
decrease of the tap change and an increase of
distribution voltage variation. In this paper, simulation
results that the weight coefficient of objective function
is w = 1 and w = 0 are analyzed and compared.
When the principal objective of parameter
determination is reduction of tap change of LRT and
SVR, w = 1, the deadband widths of LRT and SVR
are 0.9% and 1.0%, respectively. The tap position
profiles and voltage distribution are shown in Fig. 7.
The total tap change count of LRT and SVR is 4 times
and the average target voltage deviation is 35.2 V. The
P
o
w
e
r

[
p
u
]


123

0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
O
b
j
e
c
t
i
v
e

F
u
n
c
t
i
o
n
:
F
g
b
e
s
t
U pdating C ount of Solution

Fig. 5 A profile of objective function value in iterative calculation by PSO.

20
22
24
26
28
30
32
34
36
38
40
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
A
v
e
r
a
g
e

T
a
r
g
e
t

V
o
l
t
a
g
e

E
r
r
o
r

[
V
]
Total Tap Change Count
F(0.7, 1.0)
F(0.9, 1.0)
F(0.7, 0.8)
F(0.4, 0.8)
F(0.3, 0.6)
F(0.3, 0.5)
F(0.3, 0.8)
F(0.4, 0.6)
F(0.3, 0.3)
F(0.3, 0.8)
F (LRT, SVR)

Fig. 6 Pareto optimal solutions of deadband widths of LRT and SVR.

6450
6500
6550
6600
6650
6700
6750
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
L
i
n
e

V
o
l
t
a
g
e

[
V
]
Node number

(a) Voltage distribution of each node
1
3
5
7
9
11
13
15
17
19
21
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
T
a
p

P
o
s
i
t
i
o
n

N
u
m
b
e
r
Time [hour]
LRT
SVR

(b) Tap position profiles of LRT and SVR
Fig. 7 Simulation analysis results when the principal objective of parameter determination is reduction of tap change of
LRT and SVR (w = 1).
O
b
j
e
c
t
i
v
e

F
u
n
c
t
i
o
n
:

F
g
b
e
s
t

Updating Count of Solution

124

tap change count is small but line voltages at end-side
node of each feeder approaches closely to the lower
voltage limit. On the other hand, when the principal
objective is reduction of voltage deviation, w = 0, the
deadband widths of LRT and SVR are 0.3%. The tap
position profiles and voltage distribution are shown in
Fig. 8. The total tap change count of LRT and SVR is
18 times and the average target voltage deviation is
27.2 V. Since the tap action of LRT and SVR follow
the load change, the voltage margin increases though
the tap change count increases. As these results, it is
affirmed for deadband width of LRT and SVR that
there is a relation of the trade-off between reduction
of the tap change and reduction of the voltage
deviation.
Through the numerical simulation, the effective
parameters of LRT and SVR considering the tap
change count and voltage profile could be obtained by
proposed method; the deadband widths are determined
using the policy of a Pareto optimal solution after the
control parameters are optimized for improving voltage
profile.
4. Conclusion
This paper proposes a parameter determination
method of LRT and SVR for reducing the daily tap
change and keeping voltage profile. The method takes
two procedures in order to simplify the optimization
problem and to reduce calculation time. One is to
simultaneously determine the control parameters of
LRT and SVR minimizing amount of voltage violation
and voltage variation. The algorithm is based on
particle swarm optimization (PSO), which is one of
non-linear optimization methods by using a concept of
swarm intelligence. Another is to determine the
dead-band widths of LRT and SVR on the basis of dual
evaluations of the tap change and voltage deviation.
The determination problem of deadband widths is
defined as optimization problem minimizing the total
tap change counts and the average target voltage

6450
6500
6550
6600
6650
6700
6750
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
L
i
n
e

V
o
l
t
a
g
e

[
V
]
Node number

(a) Voltage distribution of each node
1
3
5
7
9
11
13
15
17
19
21
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
T
a
p

P
o
s
i
t
i
o
n

N
u
m
b
e
r
Time [hour]
LRT
SVR

(b) Tap position profiles of LRT and SVR
Fig. 8 Simulation analysis results when the principal objective of parameter determination is to minimize voltage deviation (w
= 0).


125

deviation. The concept of a Pareto optimal solution is
used for solving the determination problem. Through
the numerical simulation using distribution network
model, the validity of the proposed method has been
affirmed; the effective parameters of LRT and SVR
considering the tap change count and voltage profile
could be obtained by proposed method.
References
[1] Electric Technology Research Association, Voltage
control and management of distribution lines, Electric
Technology Research 24 (4) (1968) 20-23. (in Japanese)
[2] S. Naka, S. Toune, T. Genji, T. Yura, S. Takayama, Y.
Fukuyama, Optimal setting for control equipment
considering interconnection of distributed generators,
T.IEE Japan 120B (12) (2000) 1558-1564. (in Japanese)
[3] T. Hirai, N. Fujihara, Technical issue concerning
interconnection due to increase of distributed generators,
IEEJ Journal 125 (3) (2005) 149-152. (in Japanese)
[4] Y. Liu, X. Qui, Optimal reactive power and voltage
control for radial distribution system, T.IEE Japan 116-B
(12) (1996) 1461-1469. (in Japanese)
[5] D. Villacci, G. Bontempi, A. Vaccaro, An adaptive local
learning-based methodology for voltage regulation in
distribution networks with dispersed generation, IEEE
Trans. Power Systems 21 (3) (2006) 1131-1140.
[6] A.S. Safigianni, G.J. Salis, Optimum voltage regulator
placement in a radial power distribution network, IEEE
[7] M.E. Baran, M.Y. IIsu, Volt/Var control at distribution
substations, IEEE Trans. Power Delivery 14 (1) (1999)
312-318.
[8] D. Basset, Control of tap change under load transformers
through the use of programmable logic controllers, IEEE
Trans. Power Delivery 8 (4) (1993) 1759-1765.
[9] Y. Kubota, T. Genji, K. Miyazato, N. Hayashi, H. Tokuda,
Y. Fukuyama, Verification of cooperative control method
for voltage control equipment on distribution network
simulator considering interconnection of wind power
generators, in: Proc. of IEEE/PES Transmission and
Distribution Conference, Vol. 2, 2004, pp. 1151-1156.
[10] P.C. Dugan, S.A. Thomas, D.T. Rizy, Integrating
dispersed storage and generation (DSG) with an
automated distribution systems, IEEE Trans. Power
Apparatus and Systems PAS103 (6) (1984) 1142-1146.
[11] Y. Malachi, S. Singer, A. Vaccaro, A genetic algorithm for
the corrective control of voltage and reactive power, IEEE
[12] T. Kasajima, R. Endo, Y. Wada, Y. Kudo, H. Kanawa,
The development of the advanced distribution automation
system with optical fiber network of Tokyo Electric Power
Co., Inc., in: IEEE Power Engineering Society General
Meeting, 2004.
[13] J. Kennedy, R. Eberhart, Particle swarm optimization, in:
Proc. of IEEE International Conference on Neural
Networks, Vol. 4, Perth, Australia, 1995, pp. 1942-1948.
[14] H. Yoshida, K. Kawata, Y. Fukuyama, S. Takayama, Y.
Nakanishi, A particle swarm optimization for reactive
power and voltage control considering voltage security
assessment, IEEE Trans. Power Systems 15 (4) (2000)
1232-1239.


The Development of the Generation Expansion Planning
System Using Multi-Criteria Decision Making Rule
Seokman Han
1
, Koohyung Chung
2
and Balho H. Kim
3
1. Research Institute of Science and Technology, Hongik University, Seoul 121-791, Korea
2. Electricity Industry Policy Research Center, Korea Electro-Technology Research Institute (KERI), Gyeonggi-do 437-808, Korea
3. Department of Electronic and Electrical Engineering, Hongik University, Seoul 121-791, Korea

Received: December 23, 2010 / Accepted: June 09, 2011 / Published: January 31, 2012.

Abstract: The power expansion planning is large and capital intensive capacity planning. In the past, the expansion planning was
established with the proper supply reliability in order to minimize social cost. However, such planning cannot be used in the power
markets with many market participants. This paper proposed the power expansion planning process in the power markets. This
system is composed of Regulator and Generation Company (GENCO)s model. Multi-criteria decision making rule is used for
regulator model and several scenarios for GENCO model are applied.

Key words: Power expansion planning, regulators model, multi-criteria decision making, procedure of planning, power market.

1. Introduction

Electric power we use every day is delivered to us
through generation, transmission and distribution
facilities. As for power, supply and consumption
happen at the same time, and its impossible to store.
In addition, facilities to supply and deliver power are
highly expensive and require long construction
periods, which mean they need long-term plans.
Power expansion plans should reflect those
characteristics of power facilities to make preparations
for future power demands [1].
In the past, vertically integrated electric utilities
owned, operated, and planned their own generation,
transmission, and distribution facilities and accordingly
set up power expansion plans in a way that they could
maintain the appropriate supply reliability and
minimize social costs. As shown in Fig. 1, their power
expansion plans were reflected in their rates systems
and thus guaranteed their profitability [1, 2].

Corresponding author: Balho H. Kim, professor, research
fields: power economics, optimal power flow and power
system planning. E-mail: bhkim@hongik.ac.kr.

Fig. 1 Power expansion planning system of vertically
integrated power company.

In todays power market where a number of
Generation Companies (GENCOs) are in the game,
however, all the responsibilities of constructing and
operating power plants are left to GENCOs, whose
investments into power plant construction are
essential. As shown in Fig. 2, GENCOs set up their
investment plans to maximize their profits by
reflecting the power prices and power market rules
and taking into account the construction strategies of
other GENCOs [3-5].
Such a market-driven model has an advantage of
making the market activated through free decision
D
DAVID PUBLISHING
The Development of the Generation Expansion Planning System Using
Multi-Criteria Decision Making Rule

127

Fig. 2 Generation investment planning system of GENCO.

making to maximize profits. But it cant secure
reliability definitely according to the business
conditions of electric power industry. Considering the
rapidly changing external and internal environments
of power industry and its characteristics, GENCOs
should endure immense investment risk and indefinite
reliability.
This study suggests a power expansion planning
system in which the government-driven model that
had vertically integrated electric utilities was
combined with the market-driven model that had a
competitive market. It also aimed to discuss how the
subjects (the government or the exchange) of power
expansion planning with the accountability of
securing reliability would set up their expansion plans.
2. Power Expansion Planning System
2.1 Current Power Expansion Planning System in
Korea
To examine the procedures and nature of the current
power expansion planning in Korea, the Minister of
Commerce, Industry, and Energy establishes and
announces a basic power expansion plan to promote
the stability of power expansion according to Article
25 of Electric Power Business Act every two years.
Then a service subcommittee is formed to review the
plan, which then goes through the screening process
by the Review Committee of Electric Power Policies
(Enforcement Ordinance 27~33 of Electric Power
Business Act) and a public hearing before being
confirmed and followed by the generator construction
by GENCOs [6].
The nature of power expansion planning has
changed with time. Before the restructuring in April,
2001, it was the Korea Electric Power Corporation
(KEPCO) that set up those plans to be reviewed and
confirmed by the government. Once plans were
confirmed, the sole GENCO, KEPCO, started to build
generators. Since it was the only power company that
had the monopoly back then, its investment plans
were long term. Its power expansion planning
approach was the same with the basic planning
approach in Fig. 1. After the restructuring in April,
2001, however, there was a conversion into an
approach of respecting a GENCOs own plans.
Furthermore, information was provided about the
predictions of power expansion to stabilize power
expansion and activate competition on the premise of
wholesale competition. Also, tendency for balance
between demand and supply under the competitive
structure was appeared.
The governments policies of new and renewable
energy and its sources were reflected separately.
Unlike in the past, they began to consider the
construction intentions of GENCOs and make
announcements that included the final versions of
power expansion plans whose construction schemes
were definite and finalized. That is, they set up power
expansion plans covering the size of generators,
composition ratio of power, and supply reliability by
the years based on the plans of GENCOs. In addition,
they set up basic plans and presented them as
reference data. As presented at Fig. 3, todays power
expansion plans undergo the following process: a plan
of minimum costs is prepared using a computer
model; then the concerned criteria are provided; and
expansion plans of power generation facilities are set
up based on the survey of GENCOs intentions [6].

128

Fig. 3 Current power expansion planning system in
Korea.

The current planning system of power generation
facilities is based on the regulators scheme of
minimum costs and considers a GENCOs intentions.
But there can be a gap between the regulators scheme
of minimum costs and a GENCOs intended plan.
This study suggested a system to narrow the gap.
2.2 Proposed Power Expansion Planning System
The proposed system is a combination of the
government- and market-driven model. The regulator
offers a basic plan to GENCOs using a multi-attribute
model, and GENCOs use the plan to estimate power
prices and set up their investment plans. The regulator
employs the multi-criteria decision making model to
reflect GENCOs investment intentions to the power
market in addition to a cost minimization objective
function. If the regulators plan and a GENCOs plan
dont converge, they can make them converge by
revising the power market rules such as power prices
and capacity payments. Fig. 4 describes such process.
3. Power Expansion Planning System
Simulation Model
In this study, computer based model was developed
to simulate the proposed power expansion planning
system. The flow chart of this model is presented at
Fig. 5. The investigator compared the regulators plan
and a GENCOs plan, modified the former if there
were any differences between them, and adopted an
algorithm to revise the latter according to the modified
former. If the two plans fail to converge after a series
of those measures, the author changed capacity
payment, one of the power market rules, to revise both
the plans.

Fig. 4 Proposed power expansion planning system.

Load Data
Fixed System Data
New System Data
Emission Data etc.
Multiple Criteria
Decision Making
Regulators
Planning Model
Expansion plan
(Year, Fuel Types)
Availability factor
of each generators
Gencoss
Scenario
Check
Fuel Mix of Final Year
Is it possible
to adjust attributes?
Print
Adjusting
Attributes
Yes
STOP
Yes
Adjusting
Capacity Payment
Gencoss
Expansion plan
No
No

Fig. 5 Simulation algorithm of power expansion planning
system.

3.1 Power Expansion Planning Model of Regulator
When designing the model, the author applied the
multi-criteria decision making rule. In the proposed
model, the author defined a cost minimization
objective function as the attribute of a
government-driven expansion plan and an objective
function of maximizing the reflection of investment

129
intentions as that of a market-driven expansion plan.
And as for preference to apply preference order
dynamic programming, the author resolved the issues
of composite measures using a distance measure
consisting of each attribute [7-11].
( ) ) ( ), (
1
t
Mar
t t
Gov
t
T
t
t
y F x F P MAX

=
(1)
. .t s t C x LOLP
t t t
s , ) ( (2)
t C x Emission CO
CO
t t t
s , ) (
2
2
(3)
t u x x
t t t
+ =

,
1
(4)
t u
t
> , 0 (5)
t t t t
x u x y c c ,
w here,
t : year;
T : total planning period;
t
x : operating generators in period t ;
t
y : new constructed generators by period t ;
) (
t
P : preference function in period t ;
) (
Gov
t
F : government-driven attribute (cost);
) (
Mar
t
F : market-driven attribute (investment
intension, generation of new capacity);
t
LOLP : loss of load probability in period t ;
t
C : reliability criteria in period t ;
t
Emission CO
2
: CO
2
emission in period t ;
2
CO
t
C : CO
2
emission criteria in period t ;
t
u : new constructed generators in period t .
Eq. (1) is the objective function to maximize overall
preference during the planning period. The function
contains the cost minimization objective function to
reflect policy functions and the objective function of
maximizing the reflection of investment intentions to
reflect market functions. Eqs. (2) and (3) present the
standard of supply reliability and the restriction of
greenhouse gas emission, respectively. Eq. (4)
presents the facility expansion process, and Eq. (5)
means that the capacity of a new generation facility is
bigger than 0 all the time [12, 13].
The cost minimization objective function is
comprised of the parts to calculate construction and
operation costs as in Eq. (6).
) ( ) (
t t t t
Gov
t
x G u J F + = (6)
where,
t
J : construction cost function in period t ;
t
G : operation cost function in period t .
Eq. (7) of maximization of investment intentions
was put into a formula based on the assumption that
GENCOs would maximize the generation capacity of
a new facility in order to maximize their profits.
) (
,
j
t
j
t
Mar
j t
y H F = (7)
where,
j
t
H : generation power of type j in period t ,
{ } Nuclear Coal Oil LNG j , , , e .
Thus preference can be defined with cost, a
government-driven attribute, and electricity generation,
a market-driven attribute. Each attribute can be
measured with a distance measure in Eq. (8).
{ }
2 / 1
2
) (
(
=

i
t i
i
i t
F d w P (8)
where,
i
w : weighting factor of attribute i ;
{ } Mar Gov i , e .
The relative deviation
i
d decides the current
function values position between the best and the
worst value in Eq. (9). If attribute i has the best
attribute value among alternative f , 1 =
i
d . And if
it has the worst attribute value, 0 =
i
d .
i i
i i
i
worst best
worst f
d
= (9)
Using the weight of each objective function, one
can determine how much the concerned attribute will
be reflected.
3.2 Generation Investment Model of GENCO
In the study, the author assumed the following
simple format for the GENCO model.
IR UF UF =
min
(10)
where,

130
min
UF : minimum utilization factor taking
investment;
UF : utilization factor;
IR : increasing rate of capacity payment price.
GENCOs have investment alternatives for each
facility utilization rate and make actual investments
only when the rate is bigger than the minimal
utilization rate for possible investment calculated
through MCDM. Let us assume that a GENCO has the
following investment alternative:
-Utilization factor 60% ---------------- 3 units
-Utilization factor 30% ---------------- 2 units
-Utilization factor 10% ---------------- 1 unit
If the minimal utilization rate of possible
investment through MCDM is 40%, they will
construct only three generators that have 60%. Under
the same conditions, if capacity payment increases by
17%, the rate drops to 23%, which means that they
will construct five generators (3+2).
3.3 Adjusting Attributes
Whats the most important in a power expansion
planning model is to decides w, the attribute to use
in the regulators MCDM. The part can be put into a
formula as follows to minimize the gap between the
result of the regulators model and that of the GENCO
model:
w
MIN
G R
X X (11)
. .t s Regulator Model (12)
GENCO Model (13)
1
0
s s w w (14)
0 , 1 0 = s s i w
i
(15)
1 =
i
i
w (16)
where,
R
X : plans of regulator;
G
X : plans of GENCOs;
i
w : weighting factor of attribute i ;
0
w : weighting factor of cost;
w: minimum value of weighting factor of cost.
Eq. (11) is an objective function to converge the
two plans. Eqs. (14) and (15) present the scope of cost
attributes and other attributes, respectively. Eq. (16)
tells that the addition of weights is 1.
R
X and
G
X
that can obtain with Eqs. (12) and (13).
4. Case Study
For calculation convenience, the author made the
following assumptions:
-The review period is 15 years;
-There are two kinds (GENCO) of new generation
facilities;
-The increase rate of capacity payments will be 3%
when the objective function value of the optimization
model Eqs.(11)-(16) is not 0.
The GENCOs investment alternatives are
presented at Table 1.
The simulation results are presented at Table 2.

Table 1 GENCOs investment scenario.
GENCO
Minimum utilization factor taking
investment (%)
Units
A
73 1
68 1
63 1
B
81 8
78 2
71 1

Table 2 Simulation result using minimization cost.
) 0 . 0 , 0 . 0 , 0 . 1 (
2 1 0
= = = w w w

Year GENCO A GENCO B
2008 8
2009
2010 1
2011
2012
2013 1
2014
2015
2016
2017
2018 1
2019
2020
2021
2022 1
Cumulative units 3 9
Utilization factor at final year 69.5 % 79.2 %
Final increasing rate of capacity payment price = 6%.

131
After MCDM, the regulator wants GENCO A and B
to build two and ten generators, respectively. However,
GENCO A and B want to construct two and ten
generators, respectively, according to their investment
plans. After regulating the attribute weights and market
rules, the convergence between the regulators plan
and the GENCOs plans were appeared.
5. Conclusions
The study discussed power expansion planning
under various systems, especially under the cost-based
pool system. It also suggested a power expansion
planning system to reflect the positions of both the
regulator and GENCOs. The proposed system has
expandability to utilize the government-driven model,
the market-driven model, and the halfway model. It can
also develop and apply other attributes in addition to
cost minimization by adopting the multi-criteria
decision making model. After developing a computer
model to simulate those systems, the author tested
them through a case study.
In future, there should be efforts to enhance the
reliability of the computer model, to increase
convergence speed, and to compliment it to apply it to
large-scale practice. In addition, researches should be
conducted on the development of attributes to reflect
diverse conditions. And the GENCO investment model
briefly applied in the study needs further improvements
to accommodate reality.
Acknowledgments
This research was supported by Basic Science
Research Program through the National Research
Foundation of Korea (NRF) funded by the Ministry of
Education, Science and Technology (No.
2010-0022556).
References
[1] Y.C. Kim, Generation Capacity Investment Theory,
Aegyeong M&B, Korea, Vol. 3, 2006. (in Korean)
[2] H.G. Stoll, Least Cost Electric Utility Planning, John
Wiley & Sons, 1989.
[3] Y.H. Song, X.F. Wang, Operation of Market-oriented
Power Systems, Springer, 2003.
[4] D. Kirschen, G. Strbac, Fundamentals of Power System
Economics, John Wiley & sons, 2004.
[5] S. Stoft, Power System Economics: Designing Markets
for Electricity, Wiley, 2002.
[6] KPX, 4th Power Balancing Base Planning, 2008.
[7] S.M. Han, B.H. Kim, A study on the power expansion
planning model using multi-criteria decision making rule,
The Trans. KIEE 58 (3) (2009) 462-466.
[8] KEPCO, WASP-II Users Manual, 1993.
[9] IAEA, WASP-IV Users Manual, 2001.
[10] M. Zeleny, Multiple Criteria Decision Making,
McGraw-Hill Books Company, 1982.
[11] L.G. Mitten, Preference order dynamic programming,
Management Science 21 (1) (1974) 43-46.
[12] Y.C. Kim, Multi-criteria generation expansion planning
with global environmental considerations, Ph.D. Thesis,
Dept. Management Science, KAIST, 1992.
[13] J. Vardi, B.A. Itzhak, Electric Energy Generation
Economics, Reliability, and Rates, The MIT Press, 1981.


Study on the Photovoltaic System with MPT and
Supercapacitors
Jan Leuchter
1
, Pavel Bauer
2
and Vladimir Rerucha
1
1. Department of Electrical Engineering, University of Defence, Brno 66210, Czech Republic
2. Electrical Power Processing Department, Delft University of Technology, Delft, The Netherlands


Abstract: The paper makes survey of operating principle of power management of photovoltaic system with supercapacitors and a
maximum power tracker (MPT) control to achieve the maximum efficiency. Supercapacitors are well suited to replace classical
batteries and conventional capacitors in photovoltaic applications to improve a dynamic behaviors and life-time. It describes the
advantages of a supercapacitors solution and shows the results of comparison with convenient batteries. The paper presents different
topologies of basic photovoltaic concepts with dc-dc converters and supercapacitors according to the power dimension. The
mathematical equations describing the photovoltaic cells, temperature analyses and mathematical solution of MPT are shown in the
paper and supported by measurements. The mathematical models were applied in the design of 50 W and 6 kW photovoltaic sources
with MPT and supercapacitors. Both models were verified using experimental measurements on the photovoltaic setup.

Key words: Maximum power tracker, photovoltaic system, supercapacitor, dc-dc converters.

1. Introduction
The photovoltaic (PV) power system is very

attractive source of electricity. The civilization of
human beings is solely dependent on energy: Starting
from the simplest energy consumption in the form of
fire from wood to different types of energy sources
these days. As the population of the world grew larger
and larger and conventional resources such as coal or
oil getting depleted, alternative energy sources that are
environmentally friendly and at the same time that can
fully satisfy the need of the population in general
become inevitable. Besides to other renewable energy
sources, the solar energy is one to be relied upon. The
solar power can be utilized in the form of heat or
electrical energy. Considering the case of harnessing
the energy of the sun in the form of solar electric
energy, the high cost and low efficiency of the

Corresponding author: Jan Leuchter, associate professor,
research fields: power processing, bi-directional dc-to-dc
converters and power sources. E-mail: jan.leuchter@unob.cz.
photovoltaic cells, that convert the sun light to direct
current electricity, stands as a major constraint for its
full utilization [1, 2].
By making use of the photovoltaic cells, arranging
them in modules and arrays, we get dc voltage.
Individual PV cells with cell areas ranging from small
to approximately 225 cm
2
generally produce only a
few watts or less at about half a volt. Thus, to produce
large amounts of power, cells must be connected in
series-parallel configurations. Such a configuration is
called a module. The modules can be further connected
to form larger units called arrays. When cells are
connected together, normally they are incorporated
into PV modules, which often combine as many as 40
cells in series to produce voltages in the range of 20 V
and currents of several amperes. When voltages and
currents beyond the capability of an individual module
are desired, the modules can be connected into the
arrays that will produce higher voltages and higher
currents. Although most cells produce only a few watts,
and most modules produce 10 to 300 W, most arrays
D
DAVID PUBLISHING
Study on the Photovoltaic System with MPT and Supercapacitors

133
produce a few thousand watts. The simplest
configuration of photovoltaic system is shown in Fig. 1.
There is a 50 W photovoltaic array shown and a DC
load in a form of a light bulb.
The output I-V characteristic of such a 50 W array as
a function of sun intensity G is shown in Fig. 2. It is
possible to see dependences between input and output
photovoltaic array parameters. The photovoltaic
experimental module (M-S63-53) is described by
following parameters [3]:
-peak power 53 W;
-open circuit voltage 21.5 V;
-short circuit current 3.42 A;
-optimum voltage 17.2 V;
-dimensions 1005 453 34 mm;
-weight 5.9 kg.
Fig. 2b shows on curve 1 the experimental results of
a measurement of the first generation photovoltaic cells
output I-A characteristics (note: another photovoltaic
cells than cells from Fig. 2a) and on the curve 2 the
second plastic photovoltaic generation cells (flexi cells)
output I-A characteristics. Experimental measurement
shows that classical photovoltaic cells achieve better
efficiency (15%) with comparison to flexi cells (6%).
This is caused by internal value of resistor, which
deforms the shape of output characteristics of flexi cell.
Stand-alone photovoltaic systems may be as simple
as a module connected to an electrical load mentioned
above in Fig. 1. The efficiency of such a system is
usually poor. The next level of complexity involves
usage of energy storage. These systems also generally
include an electronic controller to prevent battery
overcharge and another controller to prevent complete
discharge of the batteries. Many charge controllers
perform both functions. The configuration of
photovoltaic system with battery is shown in Fig. 3.
The output voltage and current of photovoltaic array
are variable and the task of controller is to set up
constant value of DC voltage on the load.
Charge and discharge controller can be solved as
dc-dc bi-directional converters. The bi-directional
DC load

Fig. 1 Simplest configuration of photovoltaic system [1].

0,0
0,5
1,0
1,5
2,0
2,5
3,0
0 5 10 15 20 25 30
U [V]
I

[
A
]
G
Voltage (V)
C
u
r
r
e
n
t

(
A
)

*

(a)
0
10
20
30
40
50
60
70
80
0 5 10 15 20 25
U [V]
I

[
m
A
]
1G
2G
Voltage (V)
C
u
r
r
e
n
t

(
m
A
)

(b)
Fig. 2 I-V characteristics of photovoltaic array: (a) first
generation and (b) second generation of PV [1].

converter interfaces the battery voltage to the DC link
of photovoltaic array and DC output of photovoltaic
source. When the power flowing from the photovoltaic
array is higher than required by the load, the
bi-directional converter operates in the charge mode. If
the delivered power is lower, converter operates in
discharge mode and energy is pushed from battery. The
energy flow from and to the buttery must be controlled.

134
C
h
a
r
g
e
/
D
i
s
c
h
a
r
g
e

C
o
n
t
r
o
l
l
e
r

+ +
+
- -
-
U2=const
I2=f(Load)
U1=var
I1=f(Sun and Load)
I3=f(Sun and Load)

Fig. 3 System with battery backup.

Based on the intended application, the voltage can be
used as DC for some types of stand alone applications,
or used to charge a battery that can be used as energy
storage. But in most of the cases, it is converted to AC,
amplified into a desired voltage and frequency level
and then connected to the grid or alternatively used to
drive a machine that uses an AC source. If loads are AC,
an inverter can be incorporated into the system to
convert the DC from the photovoltaic array to AC. A
wide range of inverter designs is available with a wide
range of output waveforms ranging from square waves
to relatively well-approximated sine waves with
minimal harmonic distortion. Most good inverters are
capable of operating at conversion efficiencies greater
than 90% over most of their output power range.
2. Photovoltaic System with Supercapacitors
With the development of new technologies in
transportation, vehicles, renewable energy sources,
UPS, mobile electrical energy generating sets and in
other branches the accumulation of electrical energy,
its transformation and transportation represents one
common problem. For this purposes various types of
electrical energy buffering methods and converters
including buffers are used. As will be shown,
supercapacitors are well suited to replace classical
batteries and conventional capacitors in photovoltaic
applications, as in Ref. [4].
The supercapacitor energy storage instead of
convenient batteries (Lead-Acid, Li-Ion, Ni-Mh, Ni-Cd)
can offer high energy density and long life, as in Ref.
[5]. Supercapacitors provide extended power
availability and it can relieve batteries from burst
power function, thereby reduction costs and
maximizing space and energy efficiency. Generally, it
can be stated that supercapacitors can be charged and
discharged almost an unlimited number of times, can
be charged and discharged in terms of miliseconds,
seconds or minutes, have very high power density, do
not release any thermal heat during discharge due to
negligible internal resistence, cannot be overcharged,
are not affected by deep discharges as are chemical
batteries, have a long lifetime (80% of capacity after 10
years, lifetime up to 20 years), the DC/DC round trip
efficiency 80% to 95% in most applications, operating
temperature range between -50 C and 85 C, are
environment friendly, as in Ref. [6]. The figure shown
below (Fig. 4) illustrates the comparison of different
energy storage batteries.
The comparison of Lead-Acid, Li-Ion, Ni-Mh,
Ni-Cd and Supercaps units in relation to the price,
weight and value of stored energy is shown in Fig. 4.
Convenient batteries can bring very good relationship
between stored energy and price, nevertheless brings
weight increasing. Supercapacitors bring better
relationship between weight and price [5-7].

0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Specific Energy
(Wh/Kg)
Power density
(W/Kg)
Total Weight
(Kg)
Price ($/kWh) life (cycles) charging
efficiency (%)
Lead Acid NiCd NiMH Li-ion Supercapacitors (Maxwell)

Fig. 4 Comparison of energy buffers [4].

135
For photovoltaic applications supercapacitors take
less space and weight than conventional batteries, but
they posses excellent cold weather capabilities. The
photovoltaic concept with supercapacitor backup is
shown in Fig. 5. In this concept supercapacitor can be
designed for a low value of voltage. The first dc-dc
converter can be designed as simple step-up/down (cuk,
sepic) converters. Such a concept of the photovoltaic
power sources is very common for very low power,
where it is very important to save size, weight,
maintenance cost and for photovoltaic application,
where it is common to use the maximum power tracker
(MPT) control to achieve the maximum efficiency.
Such power sources of electrical energy are common
for military use respect to the purpose, place, sort of
appliance, type of supplied weapon systems.
3. PV System with Maximum Power Tracker
The main idea of photovoltaic system with a
maximum power tracker is to achieve the maximum
available power every time. A typical I-V characteristic
of a solar cell was shown in Fig. 2a. The open-circuit
voltage decreases only slightly of the load which is
between 0 and 70% of maximum current of every curve.
Under constant temperature, the locus of the maximum
power points is an almost vertical line, sloping slightly
toward the origin. Fig. 6 shows one of the PV
characteristics from Fig. 2a in detail (mark by *) as an
effect of power.
It is possible to see that e.g. power of the operating
point of A is 1.6 W and if we set up the system by the
controlled current to the B, we will be able to archive
4.4 W which is 70% more. This increase is not
negligible.
Next cause, which leads to use a MPT, is a
temperature effect of the photovoltaic cells. Fig. 7
shows the measurement workplace of the photovoltaic
cells and results of I-V characteristics as a function of
temperature of photovoltaic cells second generation.
Temperature effect has an influence on the shape of
characteristics. If the temperature increases then the
+
-
U
1,
I
1

D
C
-
D
C

c
o
n
v
e
r
t
e
r

U
3

D
1
D
2

I
opt

D
C
-
D
C

c
o
n
v
e
r
t
e
r

U
2

1
2

Fig. 5 System with supercapacitors backup [1].

0,0
0,4
2,9
4,4
4,2
3,6
2,9
1,6
0,2
0,0 0,0
0,1
0,1
0,2
0,2
0,3
0,3
0,4
0,4
0,5
0 5 10 15 20
U [V]
I

[
A
]
0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
5
P

[
W
]
I [A] P [W] Voltage (V)
C
u
r
r
e
n
t

(
m
A
)

P
o
w
e
r

(
W
)

Fig. 6 I-V and P-V characteristics of photovoltaic array [8].

load voltages decrease. The measurement shows the
temperature gradient 0.24% per C. On the contrary,
the short current increases 0.185 per C [8].
According to the results which were mentioned
above (Figs. 5 and 6) it is necessary to apply MPT
techniques to achieve the maximum efficiency of
photovoltaic systems. A variety of techniques has been
developed in recent years to determine an operating
point of MPT. The results of the mathematical
modeling of a variety method of MPT by used
Matlab-Simulink are shown below.
The measurements from Fig. 2a indicate that the
optimal operating voltage of a photovoltaic module is
very close to a fixed percentage of the open-circuit
voltage. This implies that MPT could simply use the
open-circuit voltage to predict the optimal operating
condition. This is called voltage-based MPT [9-12].
The results of the mathematical equations describing
the I-V characteristics with an open-voltage MPT are
shown in Fig. 8. This figure shows the optimum
voltage which is necessary to set to achieve the
maximum power of photovoltaic modules.


136
Temperature
stabilization
Cells (flexi)

(a)
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
0 5 10 15 20 25 30
U [V]
I

[
m
A
]
30 C 50 C 80 C 105 C

(b)
Fig. 7 (a) Test workplace; (b) I-V characteristics as a
temperature function.

0 5 10 15 20 25
0
1
2
3
4
Voltage (V)
C
u
r
r
e
n
t

(
A
)

Fig. 8 MPT estimation (open-voltage MPT technique).

The area of an optimum voltage as a function of sun
intensity is depicted by the solid line in the Fig. 8. An
open-voltage MPT technique is not absolutely accurate,
regardless of its accuracy is very often used since the
method is very simple and can be very easily set. The
fault of the method is marked another line. Estimating
an optimum point is not really accurate when one takes
the temperature effect into consideration. Next
problem is the degradation of solar cell by means of
ageing, where the interne resistance is changing too.
The next MPT method is based on the approximation
of a linear relationship between the maximum power
point (MPP) and the short-circuit current [9]. The
method estimates the operating point with better
reliability in comparison to the previous method.
Nevertheless the problem of temperature affects
cells ageing. Fig. 9 shows the MPT method located as a
short-circuit current. Another technique was developed
according to a linear approximation between the
maximum output power and the optimal operating
current. Such a technique is based on the optimum
point where the derivation of current to voltage is equal
to negative value of internal conductivity of
photovoltaic cell. The results of modeling are shown in
Fig. 10. The optimum points to achieve the maximum
power are pictured by star-point.
A fourth algorithm [9] measures and compares the
output power increase of modules to track the MPT.
The principle of setting MPT based on the dynamic
setting is shown in Fig. 11. The operating points A is
given by the loads and system operates with low
efficiency. To improve efficiency, the point B have to
be set. Its design involves the selection and
optimization of main control topology, current and
voltage measurement and estimation of maximum
power. If actual power is lower than maximum power
(estimation) then the controller of dc-dc1 converter
changing the duty cycle D
1
(Fig. 5) step-by-step to
achieved a maximum power. Naturally, the minimum
steps of duty setting is required. The results of MPT
system modeling to show MPT setting and maximum
power achieving is shown in Fig. 12.
Various other techniques [13-19] are also available
to estimate the characteristics of photovoltaic output
and track the MPP by i.e. way of some specific
mechanisms such as a sliding-mode observer and
neural network. According to the analyses of
techniques MPT which was shown above, the
maximum power can be archived by feed back control
of MPT point. The general scheme of the photovoltaic

137
0 1 2 3 4
0
5
10
15
20
25
30
Current (A)
V
o
l
t
a
g
e

(
V
)

Fig. 9 MPT estimation (short-circuit current MPT
technique).

12 14 16 18 20 22 24
0
0.5
1
1.5
2
2.5
3
3.5

Voltage (V)
C
u
r
r
e
n
t

(
A
)

Fig. 10 MPT estimation as a linear approximation of
optimal operating point.

I
U
Iopt
Iload
G
Uopt Uload
I
A
B
2
1
3
4
Voltage (V)
C
u
r
r
e
n
t

(
A
)

Fig. 11 Principle method of MPT with dynamic setting.

Voltage (V)
P
o
w
e
r

(
W
)

0 5 10 15 20 25
0
5
10
15
20
25
30
35
40
45
50

Fig. 12 Results of MPT dynamic setting verification.
system using a maximum power tracker and
supercapacitor battery was shown in Fig. 5. Dc-dc
*1

converter sets it to the maximum power tracker point
by means of current controlled to achieve a maximal
efficiency of the photovoltaic panel and dc-dc
*2

converter sets up a constant output voltage.
The result of analyses determines the optimum
power point to find a proper structure and parameters
of the feedback control. The optimum can be expressed
as a value of an optimal current that is a function of sun
intensity G. The basic method of control is shown in
Fig. 13a. In this case, the output PV voltage and sun
intensity G is measured and MPT is achieved by the
MPT method from Fig. 8 or Fig. 9.
Fig. 13b shows the alternating method of dynamic
control which was described above (Fig. 11). Both
systems can be used, but the alternating method is more
accurate. The main disadvantage is extra current sensor
must be designed for wide current range and responsive
for high current harmonics, because the switching
frequency of dc-dc1 is up to 100 kHz [8].
4. Experimental PV System with MPT and
Supercapacitor
According to the previous analyses and Fig. 14, the
concept of photovoltaic source with supercapacitor
includes two dc-dc converters, where first one operates
in a current mode to set the optimum current during
charging supercapacitor according to MPT and the
second dc-dc converter operates in a voltage mode to
achieve the constant output voltage 12 V. A photo of
the experimental converter is shown in Fig. 15. The
topology of SEPIC converter is very common [20]. The
SEPIC dc-dc converters produce a lower or higher
output voltage than the dc input voltage and an
advantage of this converter is that SEPIC topology
suffer with lower power losses than e.g. CUK converter.
The experimental verifications of the laboratory model
are shown in the Fig. 16. The first oscilloscope records
are verifications of MPT, the second one are
verifications of PID of output voltage during a change
of sun (input voltage) and the third one is PID output

138
opt
U U I
G
FVM
) U ( I
opt

opt opt
I I

(a)

opt
P I U

I
U

FVM ) , ( U I P
opt
opt
P

(b)
Fig. 13 Feed-back control.

+
-
U
1,
I
1

Maximum
Power Tracker
D
C
-
D
C

c
o
n
v
e
r
t
e
r

U
3

D1 D2
U1
I1

U3
I
opt

D
C
-
D
C

c
o
n
v
e
r
t
e
r

PID
U
2

U
out

U
1

U
2

Fig. 14 PV system with MPT and supercapacitors [1].

FPGA
SEPIC
1

SEPIC
2
U1
U2
Uout
Iopt
Supercaps

Fig. 15 The photo of photovoltaic workplace with
maximum power tracker and supercaps with dc-dc
converters (SEPIC) [1].

verification during load changing. As figures show,
changing duty of dc-dc*1 have effect on the power
output of the photovoltaic panel. The results of
Figs. 16b and 16c show the output of concept with a
supercapacitor. It has a constant output voltage on the

D
1
Voltage
Current
Power

(a)
U
1

U
out

(b)
I
out

U
out

(c)
Fig. 16 Experimental results of (a) MPT setting; (b) input
and output voltage of PV system with supercapacitor; (c)
input and output current of PV system with supercapacitor.

output during changing the sun intensity and load
changing.
5. Conclusions
The paper presented experimental photovoltaic
system with maximum power tracker and
supercapacitor.
The concept of the photovoltaic with a
supercapacitor as the buffer power peak delivery was
experimentally verified. Such a concept has the same
advantage as e.g. dynamic behavior.
Mathematical modeling and experimental results
were shown.

139
Acknowledgments
The research work is supported by the Grant Agency
of the Czech Republic (project No. 102/09/0013).
References
[1] J. Leuchter, P. Bauer, V. Rerucha, P. Bojda, DC-DC
converters with FPGA control for photovoltaic system, in:
13th Power Electronics and Motion Control Conference
(EPE-PEMC 2008), Poznan, Poland, 2008, pp. 422-427.
[2] M. Meinnhard, G. Cramer, Past, present and future of grid
connected photovoltaic and hybrid power systems, in:
IEEE Power Engineering Society Summer Meeting, Vol. 2,
2000, pp.1283-1288.
[3] Solartec Home Page, www.solartec.cz.
[4] J. Leuchter, P. Bauer, V. Stekly, Battery-supercapacitors
mixed as electrical power buffers, in: Proceedings of the
5th IET International Conference on Power Electronics,
Machines and Drives (PEMD 2010), Brighton, UK, Apr.
19-21, 2010, pp. 1-6.
[5] M. Piemontesi, Alternative energy storage systems for
UPS, in: Proc. PCIM Conference 2006, Nuremberg,
Germany, 2006.
[6] Maxwell technologies Home Page,
http://www.maxwell.com/.
[7] J. Leuchter, P. Bauer, A.F. Zobaa, Power electronics and
energy management of hybrid power sources with
supercapacitors, in: Proceedings of the 5th IET
International Conference on Power Electronics, Machines
and Drives (PEMD 2010), Brighton, UK, Apr. 19-21,
2010, pp. 1-6.
[8] J. Leuchter, V. Rerucha, A.F. Zobaa, Mathematical
modeling of photovoltaic systems, in: 14th Power
Electronics and Motion Control Conference (EPE-PEMC
2010), Ohrid, Macedonia, 2010, pp. 422-427.
[9] F. Chimento, S. Musumeci, A. Raciti, C.A. Sapuppo,
Control algorithm for power converter in the field of
photovoltaic application, in: EPE Conference 2007,
Aalborg, 2007.
[10] R. Diver, C. Andraka, K. Rawlinson, T. Moss, V.
Goldberg, G. Thomas, Status of the advanced dish
development system project, in: Proceedings of the ASME
International Solar Energy Conference, Kohala Coast,
Hawaii, Mar. 15-18, 2003.
[11] http://www.blueskyenergyinc.com/uploads/pdf/BSE_Wh
at_is_MPPT.pdf
[12] H. Patel, V. Agarwal, Maximum power point tracking
scheme for pv systems operating under partially shaded
conditions, IEEE Transactions on Industrial Electronics
55 (4) (2008) 1689-1698.
[13] V.P. Benjamin, L. Chong, D. Abbas, Modeling and control
of a bidirectional converter for a standalone photovoltaick
power plant, in: EPE Conference 2007, Aalborg, 2007.
[14] S. Kazmi, H. Goto, O. Ichinokura, G. Hai-Jiao, An
improved and very efficient MPPT controller for PV
systems subjected to rapidly varying atmospheric
conditions and partial shading, in: Australasian
Universities Power Engineering Conference, (AUPEC
2009), Sept. 27-30, 2009, pp.1-6.
[15] X. Weidong, W.G. Dunford, A. Capel, A novel modeling
method for photovoltaic cells, in: 35th Power Electronics
Specialists Conference (PESC 2004), Aachen, Germany,
2004, pp. 1950-1956.
[16] S. Yuvarajan, J. Shoeb, A fast and accurate maximum
power point tracker for PV systems, in: Proceedings of the
23th Applied Power Electronics Conference and
Exposition (APEC 2008), Austin, 2008, pp. 167-172.
[17] Y.J. Ou, C.H. Wang, F. Hong, Variable step maximum
power point tracking method using taylor mean value
theorem, in: Proceedings of the Power and Energy
Engineering Conference (APPEEC 2010), Chengdu,
China, 2010, pp. 1-4.
[18] N. Femia, G. Petrone, M. Spagnuolo, M. Vitelli,
Optimization of perturb and observe maximum power
point tracking methods, IEEE Transactions on Industrial
Electronics 20 (4) (2005) 963-973.
[19] S.L. Brunton, C.W. Rowley, S.R. Kulkarni, C. Clarkson,
Maximum power point tracking for photovoltaic
optimization using extremum seeking, in: Photovoltaic
Specialists Conference (PVSC 2009), Philadelphie, USA,
2009, pp. 13-16.
[20] R. Mohan, Power Electronics, John Wiley and Sons, New
York, 1995.

An Experience of On-site PD Testing for Condition
Monitoring of an 11 kV PILC Cable Insulation System
Xiaosheng Peng, Chengke Zhou and Xiaodi Song
School of Engineering and Computing, Glasgow Caledonian University, Glasgow G4 0BA, UK

Received: November 02, 2010 / Accepted: February 12, 2011 / Published: January 31, 2012.

Abstract: A cable circuit of a substation in the United Kingdom showed high level of PD activities during a survey using hand hold PD
testing equipment. The authors were invited to carry out on-site PD testing experiment to further diagnose and locate the potential
problem of the cable system. This paper presents the experience of the present authors carrying out the cable test. Following a brief
introduction to the experiment equipments and physical connections, the paper analyses the data collected from the testing, including
PD pulse shape analysis, frequency spectrum analysis and phase resolved PD pattern analysis. Associated with PD propagation
direction identification, PD source diagnosis and localisation was made. Four different types of sensors, which were adapted during the
testing, are shown to have different frequency bandwidths and performed differently. After comparing the parameters of the sensor and
the PD signals detected by individual sensor, optimal PD monitoring bandwidth for cable system is suggested.

Key words: PD identification, on-site testing, PILC cable, switchgear box, insulation defect, PD propagation direction, sensor.

1. Introduction

A cable circuit of a substation in the United
Kingdom showed high level of PD activities during a
routine survey using hand hold PD testing equipment.
Due to low denoising capability and low signal
localisation capability of the hand hold equipment,
advanced tests of the cable were desired. The authors
were invited to apply their PD testing techniques to the
cables, based on their previous research in this area,
having attracted two EPSRC funded projects in the area
of on-line PD monitoring and in signal processing and
denoising of on-line PD signals [1, 2].
Partial discharge is localized electrical discharge that
only partially bridges the insulation between
conductors and which can or can not occur adjacent to
conductor [3].
During the last 70 years, PD has been reported to be
detected in power cables, power and instrument

Corresponding author: Xiaosheng Peng, Ph.D. student,
research field: power plant condition monitoring. E-mail:
xiaosheng.peng@gcu.ac.uk.
transformers, switchgears and rotating machines [4].
For on-line or on-site PD based diagnosis, one
challenge is to distinguish PD from various
interference sources because many HV apparatuses are
connected to each other. PD and pulse shaped signals
emanating from connected electronic equipments can
propagate between different HV components which are
also susceptible to radio frequency interference.
Due to the different sources of PD, characteristics of
PD signals in different HV apparatuses are different.
PD activities in different types of cables, e.g., paper
insulated polyethylene (PILC) cables, cross-linked
polyethylene (XLPE) cables, ethylene-propylene (EPR)
cables, manifest differently.
Paper insulated MV cable were originally not
designed to be complete free of partial discharge [5].
The laminated insulation and oil filled accessories can
withstand PD activities over a long time, often more
than 40 to 50 years [5].
Most of the PD occurrences in XLPE cables are
related to accessories due to bad workmanship [5]. In the
XLPE insulation itself normally no PD should be
D
DAVID PUBLISHING
An Experience of On-site PD Testing for Condition Monitoring of
an 11 kV PILC Cable Insulation System

141
present because XLPE is a homogenous material and the
cable well routine tested at the manufacturer site [5].
In contrast, detecting problems in terminations and
joints of XLPE and EPR power cables are considered
useful, since the PD tends to exist for much longer
periods of time prior to failure in these accessories [3].
PD signals come from switchgear interfere strongly
with PD based diagnosis for cables, because there are a
number of failure processes that produce PD or
discharges in switchgear, such as electrical treeing or
tracking of epoxy spacers or control mechanisms,
metallic particles that may bounce in the switchgear due
to electromagnetic forces, poor electrical connections,
either on the HV or ground conductors [3].
2. PD Detection Tests
Physical connection of the experimental equipment
is shown in Fig. 1.
Three high frequency current transformers (HFCTs)
are attached to the earthing strap of the 3-phase cables,
which are used to capture signals from earthing system
and cables. A transient earth voltage (TEV) sensor is
attached to the switchgear box as shown, to assist with
detection of PD signals generated inside the switchgear
box.
The technical specifications of the three HFCTs and
one TEV sensors deployed in this investigation are
given in Table 1 and Table 2, respectively.
3. Data Analysis and Signal Identification
One set of raw data collected by HFCT is shown in
Fig. 2. The sampling rate of the testing is 100 MS/s.
The data acquisition time length is 20 ms, which is one
cycle of 50 Hz sinusoid signal of power supply. The
trigger mode is set to be LINE trigger mode, meaning
that data acquisition will be triggered when the
sinusoid signal of the power supply is zero-crossing
from negative to positive.
In Fig. 2, two pulses with high magnitude, more than
1 volt, appear at the 2.1 ms and 12.9 ms of the 20 ms
raw data. Ranging from 1.8 ms to 7.5 ms, one cluster of

Fig. 1 Physical connections of experimental equipments.

Table 1 Parameters of the HFCT sensor.
Item Parameters
Transfer function 5.0 V/A
Frequency response 50 kHz-20 MHz
Internal diameter 50 mm
External diameter 110 mm
Rec. load impedance 50
Output connector BNC Female
Weight 0.4 kg

Table 2 Parameters of TEV sensor.
Item Parameters
Frequency response 1 MHz-50 MHz
Typical rise time response Better than 5nSec
Sensor head dimensions 60 50 25 mm
Equivalent capacitance 150 pF (pico Farad)
Output connector BNC Female
Coupling Permanent magnets

Fig. 2 Raw data collected from the testing.

signals with low magnitude exist which are less than
0.5 volt. From 11.6 ms to 16.5 ms, there is one cluster
of signals with low magnitude which are less than 0.5
volt. The two clusters of signals are with high
possibility to be from one source and the two big pulses
are with high possibility to be from another source.

142
3.1 PD Pulse Shape
The big pulse appearing at 2.145 ms is shown in
Fig. 3a. Two small pulses which appear at 3.365 ms
and 4.995 ms are shown in Figs. 3b and 3c. The three
pulses are with similar pulse shape, and have the
highest magnitude at the beginning of the pulse and
with a long oscillation tail. The peak values of the three
pulses are all positive value.
The big pulse occuring at 12.883 ms is shown in
Fig. 3d. Two small pulses which appear at 12.064 ms
and 15.471 ms are shown in Figs. 3e and 3f. The three
pulses are of similar pulse shape, and have the highest
magnitude appearing at the beginning of the pulse and
a long oscillation tail. The peak values of the three
pulses are all negative value.
3.2 PD Pulse Frequency Spectrum
The frequency spectrum of the pulses shown in
Fig. 3a is shown in Fig. 4a. The main frequency is
782.8 kHz. The main energy of the signal distributes
from 500 kHz to 4 MHz.
The frequency spectrum of the pulses shown in
Fig. 3d is shown in Fig. 4b. The main frequency is
782.8 kHz. The main energy of the signal distributes
from 500 kHz to 5 MHz.
There is no evident difference between the
frequency spectrums of the positive and negative
pulses.
3.3 Phase Resolved PD Pattern
Phase resolved pattern (PRP), recognised as the most
effective method to describe the distributive
characteristics of PD signals, is used to classify
different types of interference signals and PD signals. 6
different types of PRP patterns, which are shown in
Fig. 5, are found during the tests.
PRP type 1, a random distribution, is a typical
pattern of interference signals. PRP type 2, a regular
distribution, signifies those signals emanating from
periodic interference source, such as electronic signals
from certain equipment. PRP type 3 is typical of the
2.144 2.152
-1
0
1
t (ms)
3.367 3.375
-0.2
0
0.2
t (ms)
4.994 5.002
-0.4
0
0.4
t (ms)
(a) (b) (c)

Fig. 3 Individual PD pulse.

|
Y
(
f
)
|
|
Y
(
f
)
|

Fig. 4 Frequency spectrum of individual PD pulses.

pattern expected of corona, which are usually
generated at a sharp point in a high voltage termination,
discharging into surrounding air. When corona
inception voltage is reached, the corona display will be
distributed close to the negative voltage peak [6]. PRP
type 4 illustrates a typical discharge between external
metal and dielectric surface [6]. PRP type 5 is typical
internal insulation PD pattern for one phase, whilst
type 6 shows the pattern expected of sources influenced
by two of the three phase voltages.
From Fig. 2, PRP of the tested data is not clear, as
its one set of data. Fig. 6 shows the PRP of 7 sets of
data. Comparing with PRP shown in Fig. 5, conclusion
can be made that PRP of the testing is PRP type 6.
Sources of the defect are influenced by two of the three
phase voltages.
3.4 PD Source Localisation
PD propagation direction is identified by three
HFCTs installed in the earth strap as shown in Fig. 7.
After identifying the propagation direction of the PD
signals, It is found that the source of the PD signals is
from the cable of concern. Signal propagation direction

143

Fig. 5 Different types of PRP pattern.

Fig. 6 PRP of 7 sets of data.

Fig. 7 Signal propagation direction.

is shown in Fig. 7, which is indicated by arrows. PD
signals are propagating from eternal earth strap to the
cable earth strap.
Further PD source localisation was carried out by
installing TEV sensor at different place of the
substation. As TEV sensor installed nearby switchgear
box, PD signals increased rapidly. Finally, the source is
identified to be from the switchgear box, which is
behind the cable, as shown in Fig. 7.
Strong interference signals are detected near a fault
current transformer (FCT) which is used for measuring
transient surge current. One set of data is shown in
Fig. 8a. One original pulse shape is shown in Fig. 8b.
The magnitude of the interference signal is up to 1.6
volt. Signal propagation analysis was carried out, the
result of which is shown in Fig. 9a. According to the
propagation direction, source of the interference
signals is FCT.
The basic principle of the FCT is shown in Fig. 9b.
The voltage of earth strap is V1. The current of earth
strap is I1. The voltage of CT is V2. The current of CT
is I2. Relationship between these parameters is V1/V2
= I2/I1 = N1/N2. N1 and N2 are the numbers of
windings of CT. N1 is 1. N2 is 1000. V1/V2 =
I2/I1=1/1000. I1=1000*I2. Interference signals in CT
will be amplified 1000 times after propagating through
CT. Thats why there are strong interference signals
nearby CT.
4. Switchgear Failures after 4 Months
Due to the accident happened in the substation,
substation managers considered the switchgear box,
which present PD activities, was in high risk.
Maintenance of the switchgear box was carried out. PD
source was found inside the switchgear box, which
proved the conclusion of the PD testing.
5. Comparison of 4 Types of Sensor
4 different sensors were used during the testing.
Parameters of the 4 types of sensors are shown in
Table 3.
Four sets of data for 4 sensors are shown in Fig. 10a,
10b, 10c, and 10d, 4 sensors were triggered
synchronously.
From Fig. 10 one conclusion can be made that HFCT
type 1 is the most effective sensor for PD based cable
testing.
One pulse with high magnitude is detected at 8.655
ms with four sensors. Pulse shapes are shown in Figs.
11a, 11b, 11c, and 11d. Corresponding frequency
spectrums of the four pulses are shown in Figs. 12 a,
12b, 12c, and 12d.


144

Fig. 8 Interference signals detected nearby CT.

(a)

(b)
Fig. 9 (a) Signal propagation direction in FCT; (b) Basic
principle of CT.

Table 3 Parameters of 4 types of sensors.
Sensor type Frequency response
HFCT type1 50 kHz-20 MHz
UHF sensor 20 MHz-1500 MHz
TEV sensor 1 MHz-50 MHz
HFCT type2 30 MHz-100 MHz

From the pulse shape and frequency spectrum of the
four pulses, two conclusions can be made: Firstly,
pulse shapes of signals are different when using
different sensors.UHF senor and TEV sensor is more
sensitive of signals at high frequency, HFCT can
couple signals with low frequency and high frequency.
Secondly, optimal frequency bandwidth should be
chosen for PD based the requirement of cable diagnosis.
Recommended frequency bandwidth of PD based cable
diagnosis is 50 kHz to 50 MHz.

Fig. 10 (a) Data collected from HFCT typel; (b) data
collected from UHF sensor; (c) data collected from sensor
HFCT type2; (d) data collected from TEV sensor.

Fig. 11 (a) Data collected from HFCT type1; (b) data
collected from UHF sensor; (c) data collected from sensor
HFCT type2; (d) data collected from TEV sensor.

Fig. 12 (a) Frequency spectrum of pulse in Fig. 11(a); (b)
Frequency spectrum of pulse in Fig. 11(b); (c) Frequency
spectrum of pulse in Fig. 11(c); (d) Frequency spectrum of
pulse in Fig. 11(d).

145
6. Conclusions
In summary the following conclusions can be drawn:
(a) Interference signals come from difference
sources seriously infect PD based cable insulation
diagnosis, effective denoising is key to success of cable
PD tests;
(b) Further identification of sources of difference
interference signals will contribute to the PD based
insulation diagnosis;
(c) It is important to distinguish PD from switchgear
box and that from cable itself;
(d) TEV is one kind of effective sensor which is
sensitive to couple PD inside switchgear;
(e) Optimal frequency bandwidth of PD based cable
insulation diagnosis system is 50 kHz to 50 MHz.
Acknowledgments
This paper is supported by EPSRC (UK),
EP/G028397/1.
References
[1] C. Zhou, X. Zhou, B. Stewart, A. Nesbitt, D. Hepburn, D.
Guo, Comparisons of digital filter, matched filter and
wavelet transform in PD detection, Recommended by
CIGRE UK panel to CIGRE 2006.
[2] X. Song, C. Zhou, D.M. Hepburn, M. Michel, Second
generation wavelet transform in PD measurement
denoising, IEEE Trans. Dielectrics and Electrical
Insulation 14 (6) (2007) 1531-1537
[3] IEC Standard 60270, High-Voltage Test Techniques
Partial Discharge Measurements, 2000.
[4] G.C. Stone, Partial discharge diagnostics and electrical
equipment insulation condition assessment, IEEE
Transactions on Dielectrics and Electrical Insulation 12 (5)
(2005) 891-904.
[5] F. Petzold, H. Schlapp, E. Gulski, P.P. Seitz, B. Quak,
Advanced solution for on-site diagnosis of distribution
power cables, in: 20th International Conference on
Electricity Distribution, Prague, June 8-11, 2009.
[6] D.A. Nattrass, Partial discharge measurement and
interpretation, IEEE Insulation Magazine 4 (3) (1988)
10-23.


Research on IPv6 Transition Evolvement and Security
Architecture of Smart Distribution Grid Data
Communication System
Xin Miao and Xi Chen
China Electric Power Research Institute, Beijing 100192, China


Abstract: Smart distribution grid needs data communication systems as a support to complete their important functions. The smart
distribution grid of the data and information are increasingly adopting internet protocol and Ethernet technology. The IP addresses
are more and more important for the smart distribution grid equipment. The current IPv4 protocol occupies a dominant position;
therefore, the challenges of the evolution to IPv6 and network security are faced by data communication systems of the smart
distribution grid. The importance of data communications network and its main bearer of business were described. The data
communications network from IPv4 to IPv6 evolution of the five processes and four stages of the transition were analyzed. The smart
distribution grid data communications network security and types of their offensive and defensive were discussed. And the data
communications network security architecture was established. It covers three dimensions, the security level, the communications
network security engineering and the communications network security management. The security architecture safeguards the
evolution to IPv6 for the smart distribution grid data communication systems.

Key words: Smart distribution grid, data communication, network, IPv6, transition evolvement, safety threaten, security architecture.

1. Introduction

Since January, 2009, the Barack Obama Government
has promoted Smart Grid for the US national strategy.
Smart Grid has become a popular word.
The State grid corporation of China proposed
strategy development targets of strong and smart grid
for the first time in May 21-22, 2009, UHV 2009, in
Beijing. The characters of strong and smart grid are as
follows. A Strong and Smart Grid is based on a strong
grid network, supported by information and
communication platform, with smart control over all
aspects of power system including generation,
transmission, distribution, customer service and power
dispatch, and covers all voltages. It is an integration of

Corresponding author: Xin Miao, senior engineer, research
fields: optical fiber communications technology, emergency,
disaster relief communications, information and
communications system of smart grid. E-mail:
miaoxin2001@hotmail.com.
power flow, information flow and work flow, a strong,
reliable, economical, efficient, clean, transparent and
interactive modern power grid [1-4].
The smart distribution grid needs the data
communication system to support to complete its
important responsibility, for example, economical
operating, intelligent early warning, auxiliary
decision-making, safety controlling, devices
management, electrical energy quality, large-scale
stored energy, and so on.
The data message of smart distribution gird more
and more uses the Internet protocol and the Ethernet
technology. And equipments of smart distribution grid
will more and more have the IP address.
The Internet Protocol is core protocol of the Internet.
At present, the Internet Protocol Version 4 (IPv4)
which was designed in the late-1970s is widely used.
The IPv4 is already difficult to adapt the request of the
D
DAVID PUBLISHING
Research on IPv6 Transition Evolvement and Security Architecture of Smart
Distribution Grid Data Communication System

147
computer development, the Internet scale and the
network transmission speed. The IPv4 address total
quantity is 2
32
, namely 4,294,967,296. But 32 bits
IPv4 addresses have far from enough, the expert
estimated that the global IPv4 address resources will
exhaust in 2012 [4-11].
The 128 bits IPv6 (Internet Protocol Version 6) has
the obvious improvement, of address capacity, security,
network management, mobility as well as quality of
service. But IPv4 to the IPv6 transition will be the long
process. The process needs to consider that the existing
data communication network to the IPv6 evolution as
well as the security system construction question, which
supports the smart distribution grid's construction.
2. Ipv6 Transition Evolvement
2.1 Data Communication Network
The data communication network is power grid
important basis of realizing the informationization,
automation and interaction goal. It is composed
together by all layers of data communication network,
like the national backbone, the area, the province, and
the city.
Integration data service, namely management
regulation service and auxiliary decision-making
service, besides dispatch control service, are main
load services.
Management regulation services are including
service of production, management, management,
coordination work and so on.
Auxiliary decision-making services are including
strategy and plan management,
production/management/operation service auxiliary
decision system and so on.
In each kind of data traffic, the IPv4 protocol
occupied the dominant position. Therefore, it is facing
the challenge of the IPv6 evolution.
2.2 Ipv6 Transition Evolvement of Data
Communication Network
The evolution process from the IPv4 to the IPv6 of
the data communication network will include the
migration, the transition, the integration, the
interoperation, and the long-term co-existence
processes [4, 11].
The transition from the IPv4 to the IPv6 can be
divided into four stages. The first is, take the IPv4
network as the main body, the IPv6 network only
constitutes the middle and small scale network in the
part, namely IPv6 isolated island, and IPv4 sea; The
2nd is, the IPv4 network and the IPv6 network parallel;
The third-order section is, take the IPv6 network as
the main body, the IPv4 network only constitutes the
middle and small scale network in the part, namely
IPv4 isolated island, and IPv6 sea; The fourth-order
section is, as monopolizes take the IPv6 network. The
IPv6 network evolution is shown in Fig. 1.
3. Security Architecture
3.1 Security Threat of the Data Communication
Network
The IP technology reduces the software and
hardware cost for the smart distribution grid data
communication network, brings convenience to R&D
(research and development) and O&M (operation and
maintenance) at the same time, also has introduced the
extra security risk. The security threat that the smart
distribution grid data communication network faces
mainly comes from the following 4 aspects.
The first is from security threat of transport network
(low layer of the data communication network). For
example, the transport network is multi-service optical
fiber transmission network, which carries the smart
distribution grid data communication network.
The second is threat from its management plane.
The third is threat from massively use
general-purpose computer. In order to reduce the cost,
the general-purpose computer equipment will be more
and more used in the smart distribution grid data
communication network. But it can bring various
influence of security like reliability reducing, hidden
security dangers.

148

Fig. 1 IPv6 evolvement sketch map of smart distribution
grid data communications network.

The fourth is from the IPv4 to the IPv6 evolution
process, the security crack or the hidden danger which
IPv4 and the IPv6 integration network technology
itself exists.
There are three attacks, regarding controllability
attack (i.e. illegal use), regarding confidentiality attack
(i.e. secret detection), and regarding usability attack
(i.e. malicious destruction) for the smart distribution
grid data communication network.
There are four kinds of methods of defending the
attack possible, the system mechanism defense, realizes
the technical defense, the engineering application
defense, and the O&M management defense.
The security of the smart distribution grid data
communication network includes environment
security, equipment security, nodes security, link
security, topology security, and system security.
3.2 Security Architecture of Smart Distribution Grid
Data Communications Network
The security architecture of smart distribution grid
data communications network includes three
dimensionality, security layers, communications
network security engineering, and communication
network security management.
Security layers include physical environment
security, equipment security, network security,
application security, information security and
information content security of their six layers.
Communications network security engineering
includes health inspection and investigation &
demand survey, evaluation analysis, planning design,
implementation and O&M (operation and
maintenance) guarantee of five process.
Communication network security management
includes the assets, staff, organization, regulations and
operation and maintenance of five elements as shown
in Fig. 2.
The security that strong support for smart
distribution grid, must be ensure from the viewpoint of
the integrity of the smart distribution grid
communication system. Concretely speaking, there are
six layers, physical environment security, equipment
security, network security, application security,
information security and information content security.
The physical environment security, equipment security
and network security belong to the category of
communication network security. And application
security, information security and information content
security belong to the category of information security.
Smart distribution grid communication system from
the perspective of security can be divided into three
categories: network security, terminal security and
application security.
There four levels of the network management,
environmental security, equipment security, network
management information security, and network
management information content security. The
network management information content security
involves junk, viruses information filtering and
inspection. See Fig. 3.
4. Conclusions
Including illegal uses, the secret detection and the
malicious destruction are attacks of the smart
distribution grid data communication network. There

149

Fig. 2 Security architecture of smart distribution grid
data communications network.

Fig. 3 Safety research hierarchy of smart distribution grid
data communications network.

four kinds of defense, including the system
mechanism defense, to realize the technical defense,
the engineering application defense, and the O&M
management defense.
Security layers include physical environment
security, equipment security, network security,
application security, information security and
information content security of their six layers.
Communications network security engineering
includes five process health inspection and
investigation & demand survey, evaluation analysis,
planning design, implementation and O&M (operation
and maintenance) guarantee. Communication network
security management includes the assets, staff,
organization, regulations and operation and
maintenance of five elements. The security layers,
communications network security engineering and
communication network security management, which
three dimensions compose the smart distribution grid
data communication system security framework.
From the network security, terminal security and
application security, three parts, develop research of
environmental security, equipment security, network
management information security, and network
management information content security. That is to
safeguards the smart distribution grid data
communication system to the IPv6 smooth evolution.
References
[1] X. Miao, Y.F. Zhang, J. Liu, Chinese road of smart grid,
State Grid 7 (2009) 54-56. (in Chinese)
[2] X. Miao, K. Zhang, X. Chen, Development
countermeasure of constructing smart grid, Electric
Power Construction 30 (6) (2009) 6-10. (in Chinese)
[3] X. Miao, K. Zhang, S.M. Tian, Information
communication system supporting smart grid, Power
System Technology 33 (17) (2009) 8-13. (in Chinese)
[4] X. Miao, IPv6 transition technology and strategy
supporting distribution network and demand side of smart
grid, Electric Power Construction 31 (2) (2010) 1-7. (in
Chinese)
[5] ITU-T, General overview of IPv6-based NGN, Y.2051,
2008.
[6] ITU-T, Framework of multi-homing in IPv6-based NGN,
Y.2052, 2008.
[7] ITU-T, Functional requirements for IPv6 migration in
NGN, Y.2053, 2008.
[8] ITU-T, Framework to support signaling for IPv6-based
NGN, Y.2054, 2008.
[9] IETF, Handling of overlapping IPv6 fragments, RFC5722,
2009.
[10] IETF, IPv6 host-to-router load sharing, RFC4311, 2005.
[11] IETF, Connection of IPv6 domains via IPv4 clouds,
RFC3056, 2001.


Operational Reliability in Transmission Power Grid
Daniel Morar
1
, Basarab Guzun
2
and Ioan Rodean
1
1. Department of M &O of T. PG., NPGC TranselectricaSA, Bucharest 030786, Romania
2. Department of Power Engineering, University Politehnica of Buchares-PUB, Bucharest 060042, Romania

Received: November 24, 2010 / Accepted: May 13, 2011 / Published: January 31, 2012.

Abstract: The paper presents the methods proposed by the authors to be applied in operation of substations. The substations taken in
consideration were rehabilitated by replacing the existing equipments by new equipments, the last generation. The methods consist in
study of equipments failures, study of equipments behaviour, study of substation architecture and study of systems operation. Paper
presents new equipments on their architecture and performances. The parts of method exposed in this paper are based on the operational
reliability. The mathematical methods and new analyses methods are given for resolving the issues on operation of substations. The
results of this study will be used to obtain the best solution for operation of the substations. The results of the analysis were applied to
some substation from the Transmission Power Grid.

Key words: Power grid, rehabilitate, reliability.

1. Introduction

We studied the operation of several substations from
the Transmission Power Grid. The old equipments no
more correspond to the requirement on the parameters
from the Transmission Power Grid: safety, economics
and quality. Therefore, we promote the rehabilitation
of the substations.
New methods for the operation of substations must
be applied to the rehabilitated substations and at the
new equipments. But no experience about the
equipments behaviour exists. Therefore is needed
finding the solutions to using those types the
equipments to obtain the best efficiency. Few papers
about the experience on using the new equipments
were published to help the operational activities [1-6].
This is not enough and is hard to apply to different
conditions of functioning, like the Power Grid which
was taken in consideration.
In this paper we will present our proposal for the
operational high voltage substation mode from the

Corresponding author: Daniel Morar, Ph.D., research
fields: transmission and distribution electrical engineering.
E-mail: contul2006@yahoo.it.
Power Grid. The behaviour of new equipments,
assembled in substations, will be presented in this
paper, like a conclusion of our study.
We make comparison between different types of
equipments assembled in substations. We present the
assumptions about how to use the equipments
operating in specific conditions for each substation
from the Power Grid. We studied the behaviour of new
equipments in different types of architecture of
substations and operational way in each from those
types. We accomplished the analysis about the new
equipments behaviour by using the methods of the
operational reliability.
The results of this analysis will be presented in this
paper.
2. Optimum Operational Method for the
EquipmentProbabilistic Model
Many factors become useful in the transport of the
electric energy which to fulfil the duty to maintain the
safety condition of the Power Grid, the quality of the
electric energy and the economic factors. Those factors
influence the electric energy transmission process. We
D
DAVID PUBLISHING

151
have considered the following factors in our analysis:
mode to use the equipments, operational scheme,
supervision and behaviour analysis the new
equipments, implementation methods, remote
command of the substations, the substations
supervisory and control.
This study was made on the equipments from the
high voltage substations, 220 kV and 400 kV.
Optimization of operational methods of equipments
can be achieved by different ways, such as: application
of probabilistic models based on factory tests,
achievement tests in design process for new
equipments, achievement factory tests before the
delivery of equipments, behaviour in operational
time-monitoring the behaviour of equipments during
operation will be achieved in different time periods
such as guaranty period, limit period in use, life cycle
and other selected periods [1-6].
The authors have chosen the method base on the
probabilistic model [7]. This model is based on
behaviour analysis of the rehabilitated equipments and
subsequent cataloguing results on criteria such as
equipment type, period of the events, the equipment
function at that time. The chosen method uses data
from substation. Data was obtained as a result of the
supervisory and operational substation with new
equipments.
The number of failures of each equipment, studied
by the authors, are shown in Table 1 (the number writes
down with symbol F and it is in percents).
The equipments from the table belong to substation
in Power Grid and there was supervisory seven years
along. For the equipments older than seven years, with
life time remaining less than 10%, we considered three
years after half of life cycle.
The equipments, which have been studied, are made
in Romania and abroad, too.
In conclusion the equipments with the number of
failures satisfactory are:
-Circuit Breaker with SF6 insulated and spring type
drive;
-Disconnector vertical type with slip guide part, in
air;
-Capacitive voltage transformers with oil and
connection by optic fibre;
-Current transformers with SF6 and connection by
optic fibre;
-Digital protection;
-OHTL 400 kV;
-Glass insulators of OHTL.
Other conclusions:
-OHTL 220 kV has more number of failures than
OHTL 400 kV;
-Composite insulators have more number of failures
than glass insulators;
-In most cases, the OHTL failures are caused by
clamps and insulators.
Equipment-Failure Method
This study was developed based on data from
equipments, throughout their life cycle, which ended
with conclusions about their evolution.
The equipments were in different stages of operation,
from the design until the ending of their life cycle.
During operation of the equipments occurred several
failures. The causes of those failures can be:
-Design errorwrong calculus, wrong data expose
on plans or data wrong collected from substation and
which are included in calculus;
-Other errors: missing data, insufficient data or
wrong data about the equipments (such as wrong
location, technical issue, incorrect installation, errors in
operation and maintenance, lack quality
system/environmental plan/geological plan and other
data errors);
-Setting errordate error about place, equipment
parameters, error communication, and error demand;
-Fitting error;
-Operational error (overload, wrong manoeuvre,
external cause, natural phenomenon, human error,
planning error).


152

Table 1 Example for the number of failures of equipment [8].
Tip of equipments Failures Tip of equipments Failures Tip of equipments Failures
for Circuit Breaker with:
-Oil insulated
-Drive-oil-pneumatic type
-Use at 220 kV
F = 0.762
for Disconnector with:
-In air
-Drive-electric type
-Use at 220 kV and 400 kV
F = 0.57
for Digital Protection
-Use for 220 kV and 400
kV installation
F = 0.39

The differences between demand and offer for
equipments acquisitioned appear in stage demand
and in stage marketing. This error has the following
causes:
-Demand contains parameters which can't come true;
-Different requirements by substation needs;
-Incomplete requests;
-The type of equipment is lack on the market;
-Economic situation.
The probability of failures must be known by the
decisional staff to take appropriate measures for
improved the operation of the substation. Failures,
which can appear, and must be estimated, are:
-The break of the equipment;
-operation failure;
-loss of load;
-failures removed later or non selectively;
-breaker failure operation;
-Long time for maintenance;
-Loss the auxiliary transformers or auxiliary bus bar
stays without voltage;
-AR operation;
-OHTL is broken;
-wrong operation of the equipment.
In Table 2 are presented the deviation from ideal
case (d symbol) and the reliability decrease during
operation time of the equipment (F symbol and M
symbol) [8, 9].
The reliability over all life time of the equipment is
present in three steps:
-First 5 years;
-Next 15 years;
-Last 5 years.
The values and evolution of the equipments, which
were taken in consideration, on its life cycle is
presented in Fig. 1.
Table 2 The deviation from ideal case and the reliability
decrease during operation time of the equipment.
Action Deviation Failures
On demand by costumer D = 0.20 F = 0.0120
On market D = 0.20 F = 0.0130
Design D = 0.14 F = 0.0900
In factory - F = 0.0001
At mounting - F = 0.2000
In operation - F = 0.2600
Maintenance - M = 0.8700

Fig. 1 The deviation from ideal all over life time of the
equipment.

The conclusions of the study about the reliability
over all life time of the equipment are: the reliability
during operation is about half the deviation from the
ideal (from 1, which means ideal, to 0.53).
At the ending the life time the equipment was still
functional. The equipment had a good behaviour all
over its life time.
The equipment that was chosen had an average
evolution of all study equipment.

153
High Voltage Systems
The equipments, used in calculus the operational
reliability, are part of a system with different
architectures.
We studied how can be use several types of the
equipments in the systems with different architectures.
The study presents the types of architectures of
systems that are most profitable.
The data from calculus of the operational reliability
of equipments, which was presented above, were used
to calculus the reliability of the system. In calculus of
operational reliability of the systems were used only
the best results of the reliability of equipments.
We achieved an analysis about the solutions offered
of each system, with different architectures, from
Power Grid.
We have taken in consideration the behaviour of the
system to each kind of failure.
For calculate the reliability and the elasticity for each
of the systems architecture, we have taken in
consideration the following factors:
-The safety of the scheme;
-The architecture type of scheme;
-The equipments manoeuvre action;
-The equipments type and function;
-The equipments reliability;
-The operational scheme;
-The spare capacity of the scheme;
-The operational mode of the equipment and the
scheme (use, supervisor, manoeuvre, monitoring).
The results of the behaviour of different systems,
having over 5 years of operation, are presented in Table
3.
The conclusions of the analysis of the systems types
are:
-The equipment cant function for long time on the
by-pass bus bar;
-The highest values of the reliability were recorded
of the following systems, in descending order:
-1.5 circuit breaker system
-Double bus bar system
-Hexagonal system
-Most convenient systems for proper operation are:
double bus bar with by-pass bus
bar-equipment-double bus bar with by-pass bus bar"
and 1.5 circuit breaker-equipment-double bus bar
with by-pass bus bar;
-Most reliable system is 1.5 circuit breaker system.
5. Conclusion
The study shows the following results:
(1) It notes that the larger number of failures is due to
actions taken before the commissioning of the
equipment, the maintaining in operation of the old
equipment or equipments without maintenance at the
time.
The solution proposed in this paper is:
-Identification of specific causes of failure by an
efficient analysis, both graphically and statistically;
-Collaboration between the operating
unit-production and service units;
-Operating the business organization to enhance its
performance.
We have obtained the following result: we increased
the ratio of time to use the Transmission Power Grid,
we reduced the energy costs and we reduced the
number of accidental interventions.
(2) Values of probabilities of failure show that over
two thirds of equipments must be replaced after 40
years of operation and 90% of equipments cannot be
used over all normal life cycle.
The solution proposed in this paper is:
-Enable maintenance intervals that yield efficiency;
-Enable maintenance based on failure analysis
results and specific defects;
-Enable maintenance by applying advanced
technologies, LW technologies;
-Purchase of equipment performance.
We have obtained the following result: we increased
the ratio of time to use the Transmission Power Grid
and we reduced the maintenance costs of equipments

154

Table 3 Example for the reliability of the systems (R symbol).
System type R (%) System type R (%) System type R (%)
Simple bus bare-OHTL-simple bus
bare
0.36
1.5 circuit breaker-transformer
-1.5 circuit breaker
0.44
H system-2 OHTL with 1
transformer
0.18
1.5 circuit breaker-OHTL-double
bus bare
0.40 Hsystem-cross function 0.38 OHTL -transformer in H system 0.42

References
[1] G.M.D. Poenaru, Comportarea n Exploatare a
Echipamentelor Termoenergetice, Danubius Publishing
House, Bucharest, 2000.
[2] V. Dua, V. Vaida, Comanda i Controlul Funcionrii
Reelelor Electrice, Technique Publishing House,
Bucharest, 2001.
[3] I. Felea, N. Coroiu, Fiabilitatea i Mentenana
Echipamentelor Electrice, Technique Publishing House,
Bucharest, 2001.
[4] N.F. Pinte, TIS. Teoria Speciilor Informaionale, Irecson
Publishing House, Bucharest, 2007.
[5] J. McDonald, Electric Power Substations Engineering,
CRC Press and IEEE Press, 2001.
[6] R.L. Giles, Layout of EHV Substations, IEEE Press, 2001.
[7] W. Mendenhall, T. Sincich, Statistics for Engineering and
the Sciences, 4th ed., Englewood Cliffs, New Jersey,
Prentice Hall, 1995.
[8] D. Morar, The methods about the reliability of the
operation substation and operation power grid on high
voltage, Ph.D. paper, University Politehnica of Bucharest,
2007.
[9] D. Morar, The operation of the high voltage equipments
in the power grid substations, Ph.D. paper, University
Politehnica of Bucharest, 2003.

Journal of Energy and Power Engineering-2012.01

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Journal of Energy and Power Engineering-2012.01

Uploaded by

Copyright:

Available Formats

Volume 6, Number 1, January 2012 (Serial Number 50)

. The gyroscope system, Eq. (2),

The photovoltaic (PV) power system is very

You might also like