You are on page 1of 9

P.

Artuso
e-mail: paola.artuso@uniroma1.it
F. Zuccari
A. DellEra
F. Orecchini
CIRPS-Interuniversity Research Centre for
Sustainable Development,
Sapienza-University of Rome,
Via delle Sette Sale 29,
00184 Rome, Italy
PV-Electrolyzer Plant: Models and
Optimization Procedure
The work focused on the analysis of the connection between a photovoltaic (PV) plant
and an electrolyzer for hydrogen production. On the basis of PV-plant and electrolyzer
experimental data, the effectiveness of the models adopted in the simulation program has
been veried in order to choose the best model and, eventually, modify some parameters.
By running the simulations, the procedure to optimize the PV-plant and the electrolyzer
combination has been established. In fact, the simulation results might be considered to
size an electrolyzer as small as possible, which is able to exploit up to the maximum
power actually produced by the PV-plant during a working year. This criterion allows
minimizing the overall plant costs. Furthermore, the possibility of deleting the maximum
power point tracker and the dc/dc converter has been analyzed. On the basis of the
obtained results, this opportunity is preferable to avoid the energy losses in the power
control system; and it is convenient even from an economic point of view, considering
that the electronic device costs are comparable with the PV-plant ones.
DOI: 10.1115/1.4001673
1 Introduction
The vast quantities of hydrogen produced today are consumed
in non-energy-related uses, for example, in ammonia production
for fertilizers or in renery processes, but further hydrogen appli-
cation is in energy eld, because it is heralded to be the energy
carrier of the future to run out by the petroleum era. Hydrogen
could be used as an energy vector 1,2 in a completely renewable
resources based system, because it can be produced from water, in
particular, by electrolysis process, when the source, for example,
sun radiation, exceeds the user electric energy demand while it
can be combusted in fuel cells when the source is lower than the
electrical demand. Following this scheme, the only by-product
obtained by the hydrogen production and used processes is water.
In other words, hydrogen production is a chemical method to store
the variable and insecure renewable energy. In fact, renewable
energy is used to dissociate the water molecule in its compounds;
thus, hydrogen contains most of the energy used in this process.
Then, when hydrogen reacts with oxygen in fuel cells or in an
ICE, most of the energy consumed to produce hydrogen is given
back during the combustion process. Although the hydrogen pro-
duction by renewable sources is actually possible in a technologi-
cal viewpoint 36, the costs are much higher than the gasoline
ones, per energy content. Thus, the optimal design of the hydro-
gen production plant could be the rst step to decrease the hydro-
gen costs and make it competitive with the other fuels, at least for
some applications.
The PV generator produces nonlinear I-V characteristics, and its
maximum power point MPP varies with solar isolation and tem-
perature. Thus, the PV-plant connection with the electrolyzer is
usually realized interposing a MPP tracker and a dc/dc converter
between the two systems. The MPP tracker is an electronic device,
which adjusts the PV panel output voltage to a value in order to
supply the maximum energy to the load; while the dc/dc converter
7 is an electronic system, which converts the PV output voltage
at MPP into the voltage required by the electrolyzer. This kind of
connection will be named as indirect. The opportunity to di-
rectly connect the PV-plant and the electrolyzer is intended to be
investigated and analyzed in comparison with the more usual in-
direct connection 8. To establish simple criteria in order to op-
timize a hydrogen production system consisting of a PV-plant
feeding an electrolyzer is the aim of this paper.
2 Simulation Program
The sun radiation is variable and insecure, but statistic values of
the cloudiness, temperature, and other important radiation param-
eters are well known and included in ofcial papers, such as the
Italian UNI 10349.
The PV-cell outputs are inuenced by a lot of conditioning
variables, which can be classied in ambient parameters daily
solar radiation, ambient temperature, cloudiness, date, and hour of
the day, PV-cell features usually provided by the producers
Table 1, and PV module parameters tilt, azimuth, latitude, and
series-parallel connections.
Among several kinds, the program simulates an alkaline ad-
vanced KOH electrolyzer on the basis of the ystein Ulleberg
model 9. It has been adopted using the empirical parameters in
Ref. 9, determined for a 26 kW advanced alkaline electrolyzer
consisting of 21 electrolytic cells electrically connected in a se-
ries.
In accordance with this model, the characteristic equation is
U = U
rev
+
r
A
I + s log
t
A
I + 1 1
Contributed by the Solar Energy Division of ASME for publication in the JOUR-
NAL OF SOLAR ENERGY ENGINEERING. Manuscript received September 24, 2009; nal
manuscript received March 26, 2010; published online June 29, 2010. Assoc. Editor:
Robert Palumbo.
Table 1 Average values for commercial PV cells considered in
program simulations
Quantity Symbol Value
Transparent coefcients of cell surface 0.9
Absorbing coefcients of cell surface 0.9
Normal operative cell temperature NOCT 50C
PV-cell efciency at STC
rif
0.121
Short-circuit current at STC I
sco
3 A
Open-circuit voltage V
oc
at STC V
oco
0.59 V
Constant of diode ideality N 1.3
I
sco
variation coefcient h
i
6.410
4
V
oco
variation coefcient h
v
3.710
3
Coefcient of PV-cell efciency variation B 0.0045
PV-cell area A
c
100 cm
2
Journal of Solar Energy Engineering AUGUST 2010, Vol. 132 / 031016-1 Copyright 2010 by ASME
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
Analyzing the data of the PHOEBUS electrolyzer, the conclusion
has been that the temperature does not inuence the parameter s.
The Ohmic resistance r is linearly dependent on the temperature,
and t is a function of the temperature Eqs. 2 and 3 as follows:
r = r
1
+ r
2
T 2
t = t
1
+
t
2
T
+
t
3
T
2
3
In the simulation program, the empirical parameters provided by
the paper 9 are considered. They are reported in Table 2.
It is considered to increase the electrolyzer size proportionally
to the cells number. The considered electrolytic cell properties are
summarized in Table 3. Operating electrolyzer conditions depend
on several parameters supplied power, number of electrolytic
cells, and temperature. Because of the complexity of the whole
hydrogen production system, a simulation program is considered a
useful instrument to consider all the working quantity variations
and to foresee the PV-plant electric production and the connected
electrolyzer outputs, in particular, the total hydrogen production.
3 Comparison Between the Simulated Data and the
Real Ones
The simulation program in LABVIEW 10,11 is implemented
using detailed models of radiation 12, PV-plant 1316, and
electrolyzer 9,14. A comparison between these models and the
experimental data has been carried out.
4 Photovoltaic Models
The photovoltaic cell varies the efciency when the tempera-
ture changes. In particular, the efciency enhances when the mod-
ule temperature decreases, in accordance with the formula
=
r

pt


1
t
T
module
T
r

t
G

U
t
1
r

pt

4
The efciency can be calculated knowing the radiance and the
ambient temperature values, while the module temperature can be
evaluated knowing the ambient temperature by means of
T
module
= T
amb
+
G
U
t

r
G
U
t
5
The module temperature increases when the radiance rises. It also
depends on the module efciency, in particular, the higher the
temperature, the lower the efciency. The simulation program
uses Eq. 5 to compute the module temperature variation. The
model has been veried comparing the real values and the simu-
lated ones. The real modules inputs and outputs have been moni-
tored and collected during three working months of the PV-plant
located at the Engineering Faculty of the Sapienza-University of
Rome. The considered quantities are the date, the ambient tem-
perature, the PV module temperature, the radiance, the voltage,
and the generated power. The quantities are taken over each 10 s,
and the average of 60 values is computed each 10 min.
Using the real data to obtain the efciency and the ambient
temperature, the module temperature is computed to compare the
modeled values with the measured ones.
In Fig. 1, in particular, the temperature trends of ambient PV
cells and the simulated cell temperature values during a sample
day are plotted from 5:00 a.m. to 21:00 p.m. The comparison
between the graphs highlights the real cell temperatures that
match the simulated quantities with a small error, which is maxi-
mum equal to 19%, and the maximum error of the efciency was
estimated to be about 20%, as reported on a parity plot in Figs.
25.
If the module temperature is always overevaluated, it means
that using this model, in particular, Eq. 5, the computed ef-
ciency is lower than the real one, which is what has been inves-
tigated and reported in the following paragraph.
5 Efciency Model
The second step was to verify whether the simulated efciency
data match the real one. First, the efciency has been computed
knowing the cells features provided by the producers, the ambient
temperature trend, and the radiance variation by means of Eq. 2
14. The calculated data have been compared with the ltered
real data of the monitored Bologna PV-plant provided by ENEA
Fig. 1 One day temperature variation
Table 2 Overvoltage coefcients of Phoebus electrolyzer
Parameter Value
r
1
8.0510
5
m
2
r
2
2.510
7
m
2
C
1
s 0.185 V
t
1
1.002 A
1
m
2
t
2
8.424 A
1
m
2
C
t
3
247.3 A
1
m
2
C
2
Table 3 Electrolytic cells properties
Parameter Value
Size 1238 W
Electrolyzer temperature 80C
Area electrodes 0.25 m
2
Switch on current 100 A
Maximum voltage 1.90 V
Minimum voltage 1.43 V
031016-2 / Vol. 132, AUGUST 2010 Transactions of the ASME
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
Fig. 2. The data refer to an acquisition period from October 2000
to November 2002, and the quantities related with radiance lower
than 100 W/ m
2
are deleted because the quantities were affected
by too high measure errors due to very small quantities. In Figs.
35, the black line is the real data-tting curve. In those gures,
the triangles represent the real data, and the circles depict the
model efciency versus the radiance values. It has been observed
that for low radiance values, the real efciency data vary in a wide
range. This is due to the fact that in correspondence of certain
radiance, the module temperature can assume different values, in
particular, when the radiance suddenly changes; for example,
when small clouds are suddenly interposed between the sun and
the photovoltaic surface. Furthermore, the real tting curve is
characterized by a maximum: For low radiance values, the ef-
ciency features an increasing trend accordingly with the increas-
ing radiance, while for a high radiance quantity, the efciency
assumes a decreasing trend.
In particular, the efciency variation as a function of the radi-
ance has been analyzed, xing the temperature in a range of 1C.
The graphs in correspondence of three temperatures are reported
in Figs. 35.
Comparing the results, the conclusion has been that the model
matches the real data. In fact, the computed efciency has values
very similar to the real data-tting curves for each considered
temperature, and the relationship between the efciency and the
radiance is characterized by a maximum both for the computed
and real data.
6 Electrolyzer Model and Real Data
The possibility to apply the same model to an electrolyzer dif-
ferent from the Julich one has been analyzed. For this reason, a
comparison between the simulated data and a 5 kW alkaline KOH
electrolyzer characteristic curve provided by Casale Chemicals
Lugano, Switzerland, has been carried out. The ystein Ulleberg
procedure has been applied to compute the empirical coefcients.
In particular, the relationships among r, s, t, and temperature have
Fig. 2 a Comparison between model and real data efciency
versus the radiation values; b parity plot
Fig. 3 a Model and real data efciency versus the radiation
values. Ambient temperature range: 1516C; b parity plot.
Fig. 4 a Model and real data efciency versus the radiation
values. Ambient temperature range: 2526C; b parity plot.
Journal of Solar Energy Engineering AUGUST 2010, Vol. 132 / 031016-3
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
been determined. The data of the working voltage in correspon-
dence of three different power density and seven temperature val-
ues have been considered. In correspondence of each temperature,
Eq. 3 has been written for three power density values. A system
of three equations and three incognita r, s, and t has been ob-
tained in order to compute the r, s, and t values. In this way, the r,
s, and t parameters have been computed in correspondence of
each of the considered seven temperature values. In Figs. 68, the
points represent the parameters r, s, and t against the temperature.
The tting curves have been dened; thanks to the MATLAB
program support. Some of the equations and empirical parameters
are noticed to be different from the ones in Ref. 1. In fact, the
dened relationships are
r = r
1
+ r
2
T 6
s = s
1
+ s
2
T + s
3
T
2
7
t = t
1
+ t
2
T + t
3
T
2
+ t
4
T
3
8
The computed empirical parameters are reported in Table 4. After
all, the ystein Ulleberg model characteristic curve has been com-
puted, inserting the Table 4 empirical parameters. The obtained
values have been compared with the real ones, and a maximum
0.29% error is inferred. In conclusion, the Casale Chemicals elec-
trolyzer is veried to be well simulated by the ystein Ulleberg
model, xing nine empirical parameters instead of the six coef-
cients taken into account in the original equation.
7 Electrolyzer Simulation
In the simplied approach characterizing the simulation pro-
gram, a temperature of 80C has been considered even during the
start up phase, and any electrolyzer size is simulated considering
that the required power and voltage are proportional to the elec-
trolytic cell number. In fact, each electrolytic cell has been con-
sidered to be fed by the same power at the same voltage level. The
reference electrolyzer is 26 kW and consists of 21 cells, com-
pletely similar to the Julich one analyzed in Ref. 9.
If the total 26 kW power is equally shared into the electrolytic
cells, it implies that each cell is fed by power P
cell
equal to
P
cell
=
26,000 W
21
= 1238 W 9
Similarly, each electrolytic cell is considered to be characterized
by the same voltage value, so that the electrolyzer voltage is pro-
portional to the cell number. Being the working voltage range
3040 V, each electrolytic cell is considered to operate between
1.43 V and 1.90 V. On the basis of these considerations, in corre-
spondence of a chosen electrolyzer size, the electrolytic cells
Fig. 5 a Model and real data efciency versus the radiation
values. Ambient temperature range: 3536C; b parity plot.
Fig. 6 r parameter trend versus the temperature
Fig. 7 s parameter trend versus the temperature
Fig. 8 t parameter trend versus the temperature
Table 4 Overvoltage coefcients of Casale Chemicals
electrolyzer
Parameter Value
r
1
8.2410
5
m
2
r
2
4.1210
7
m
2
C
1
s
1
0.2393 V
s
2
0.002952 V C
1
s
3
1.5510
5
V C
2
t
1
0.6767 A
1
m
2
t
2
0.02711 A
1
m
2
C
1
t
3
0.0004856 A
1
m
2
C
2
t
4
2.6910
6
A
1
m
2
C
3
031016-4 / Vol. 132, AUGUST 2010 Transactions of the ASME
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
number and the corresponding required voltage could be calcu-
lated. The electrolyzer is composed of cells in a series, so that the
current supplied to each cell equals the total current characterizing
the entire hydrogen generator.
By exploiting the simulation program, we evaluated whether it
is more convenient to have a large number of small electrolytic
cells in comparison to the reference electrolytic cell. All the pro-
duced electrolyzers are characterized by a minimum operating
power, which corresponds to a minimum current value, which is
100 A in the reference electrolytic cell. This is the minimum
power level to guaranty suitable gas purity and proper safe work-
ing conditions. Being the electrolytic cells in a series in the sche-
matic system adopted, the switch on current is considered 100 A
no matter what the electrolytic cells number composing the elec-
trolyzer. Supposing to decrease the switch on power, the electro-
lyzer should exploit more energy produced by the PV-plant be-
cause it can operate at a current level lower than 100 A. The
minimum current depends on the size of the electrolytic cell, in
particular, on the active surface of the electrodes. In fact, the
current decreases proportionally to that area once the current den-
sity is set. Therefore, the smaller the electrolytic cell, the lower
the minimum operating current. The possibility to consider a
small electrolyzer with a lower switch on current has been ana-
lyzed, but although it is assumed to connect an innite number of
very small electrolyzers, the switch on current should be, as a
limit, equal to 0 A. In this last case, the improvement of the annual
hydrogen production is evaluated to be only about 3% more than
the hydrogen produced by the reference electrolytic cell. In other
words, the switch on current does not cause an appreciable hydro-
gen production variation, so that it is not convenient to design a
plant with a large number of small electrolytic cells.
8 Direct and Indirect Connections
Two connection opportunities have been considered and ana-
lyzed. The PV-plant is indirectly connected to the electrolyzer
because of the interposition of a maximum power point tracker
MPPT and a dc/dc converter between the electric generator and
the hydrogen one, so that the two systems can work in the best
voltage and current conditions. In this case, the power supplied to
the electrolyzer is established to equal the power produced by the
PV in the MPP, minus the losses in the wires about 2%. The
electrolyzer operating point, which is dened by the current and
voltage values, is such a point on the electrolyzer characteristic
curve that the area of the rectangle in gray in Fig. 9 product of
voltage for current equals the computed power supplied to the
electrolyzer.
8.1 PV-Plant Directly Connected to the Electrolyzer. In
this case, the working conditions are determined by the cross
point of the two systems characteristic curves, as depicted in
Figs. 911, where the dashed line is the PV characteristic curve,
while the continuous line is the electrolyzer one.
In Fig. 10, it is shown how the PV curve changes, varying the
radiance G. Each characteristic is drawn in correspondence of a
constant G parameter. In particular, enhancing G, the curves shift
toward higher power values, while the voltage has a small varia-
tion. Therefore, by the graph, it is visible how the cross point
changes. Instead, in Fig. 11, the electrolyzer curve variation is
reported, changing the cells number; in other words, varying the
electrolyzer power-size. Increasing the electrolytic cells number,
the curve shifts toward higher voltage values in a proportional
way. Contemporary, the working point varies, moving toward
points more or less close to the point of maximum PV power
production, which is near the knee of the PV characteristic.
In the last picture Fig. 12, there is how the PV curve assumes
different shapes, varying the number of strings, once the total
PV-cell number is xed; in other words, x the PV-plant size. The
cross point is signicantly modied in the two considered cases.
In fact, in one case, the working point is close to the knee, which
is the maximum power point, while in the other case, the cross
point is far from the knee; thus, most of the PV potential power is
not exploited. The conclusion is that in the direct connection, it is
important to obtain an output PV voltage number of cells in a
series in order to realize a matching point close to the nominal
electrolyzer voltage.
The PV-cell number connected in a series is established in order
to produce an output voltage higher than the minimum voltage
required by the electrolyzer and in order to activate the electro-
Fig. 9 Direct connection
Fig. 10 PV characteristic curve variation, increasing or de-
creasing the radiance G
Fig. 11 Electrolyzer characteristic variation, increasing or de-
creasing the electrolytic cell number
Fig. 12 Electrolyzer characteristic variation, changing the
series-parallel connection
Journal of Solar Energy Engineering AUGUST 2010, Vol. 132 / 031016-5
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
lytic reaction in each cell as long as possible, in accordance with
the power supplied by the PV-plant. This step is really important
in direct connection because the hydrogen production is greatly
inuenced by the series-parallel combination of the modules.
When the program runs, the PV voltage output at the MPP is
computed and drawn on the front panel, as reported in the Fig.
13b graph. It is the typical shape of a sample day voltage varia-
tion. In the same gure, the a graph depicts the operating voltage
trend of the coupled electrolyzer, supposing the direct connection.
The electrolyzer voltage versus the time is not widely variable. By
the analysis of those plots, we assumed that the optimization pro-
cedure consists in nding the conditions in which the a curve is
as close as possible to the b curve in Fig. 13.
Considering that the output voltage at MPP does not vary sen-
sitively during the day, in spite of the radiation trend, the maxi-
mum voltage value can be taken as the indicative value of a day
voltage. By the simulation, the maximum PV voltage variation
during the day has been plotted, as reported in Fig. 14, by means
of which it is possible to point out that the minimum voltage
values are reached in correspondence of the summer months
JulyAugust, while the maximum voltage levels are recorded
during the winter months.
In Fig. 15, the PV-cell temperature variation both in a typical
winter and a summer day is reported. The maximum PV-cell tem-
perature varies from 64C during the hot months to 26C during
the cold season. Considering that the operating voltage produced
by the PV-plant is quite constant during the day Fig. 13 while it
is more variable during the year Fig. 14, the conclusion is that
the voltage is more sensitive to the temperature than to the radi-
ance, and, in particular, the voltage is inversely proportional to the
PV-cell temperature Fig. 15.
The plate electrolyzer voltage could be set equal to the typical
output voltage either in summer or in winter. In the rst case, the
electrolyzer works close to the nominal conditions during sum-
mer, but in winter, the operating voltage will be higher than the
characteristic plate value. In the second case, the electrolyzer
works close to the nominal conditions during winter, while during
summer, the operating voltage is lower than the nominal values.
The aim is to have the two characteristic curves cross point in
closeness of the MPP; in other words, the working electrolyzer
voltage should be near the voltage of the PV-plant MPP as long as
possible.
9 Simulation Description
The reference system considered in LABVIEW simulations con-
sists of 630 modules of 36 PV commercial polycrystalline cells,
characterized by a 12.1% reference efciency. It is considered to
be located in Rome 42 deg latitude, oriented perfectly at south 0
deg azimuth. Setting the modules number composing the PV-
plant, different cases have been analyzed considering different
tilts. The program has computed the maximum power actually
produced by the PV-plant during a working year in July. On the
basis of this value, and considering a 2.5% loss in the wires and
10% electrolyzer elasticity,
1
the electrolyzer size and the electro-
lytic cell number are deduced. Once the electrolytic cell number
has been established, the plate voltage required by the electrolyzer
is determined, and on the basis of this, the series-parallel module
1
The electrolyzer can work at a power 10% more than its nominal power. In fact,
all commercial electrolyzers have this property, in particular, if the high power level
is realized for short periods, as it actually happens when the electrolyzer is connected
to a PV-plant.
Fig. 13 a The coupling voltage and b the PV-plant MPP ver-
sus the hours of a sample day
Fig. 14 Voltage variation during the year
Fig. 15 Temperature variation in a sample day a in summer
and b in winter
031016-6 / Vol. 132, AUGUST 2010 Transactions of the ASME
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
connection is established in order to supply the voltage required
by the electrolyzer.
Both in the direct and indirect connections, the same electro-
lyzer is considered; in other words, the same properties and the
same number of electrolytic cells characterize the electrolyzer
chosen in the two analyzed cases. The hydrogen production ob-
tained by running the program has been written in Table 5.
The unexpected result has been that the maximum hydrogen
yield was reached with 60 deg tilted modules, although the best
radiance catchable during the year was obtained with 30 deg tilted
surfaces. The reason of the misunderstanding has been further
investigated. Since the intention was to exploit the full power that
PV can provide, with a tilt of 30 deg which maximizes the en-
ergy, the electrolyzer should consist of 20 cells instead of either
17 or 14. In this way, the voltage and power required by the
electrolyzer are higher than in the other cases. Therefore, in a 30
deg tilted module case, only two modules in a series cannot sup-
ply the required electrolyzer voltage, thereby three modules
strings have been considered. The consequence has been that the
voltage at a maximum power point is much higher than the volt-
age required by the electrolyzer. The crossing point voltage of the
two systems characteristic curves is much lower than the MPP
voltage. In this case, the MPPT and the chopper action guarantee
the best working conditions, even if they introduce losses esti-
mated around 2%. In the other cases, the maximum power pro-
duced by the PV-plant is lower; therefore the coupled electrolyz-
ers is smaller. The required voltage decreases proportionally to the
electrolyzer size, so that only two module strings can supply the
necessary voltage when directly connected to the hydrogen gen-
erator. The direct connection is more convenient because the
working points are nearer the MPP during the year, for the reason
explained before. In this case, the advantage is to avoid the losses
related to the MPP tracker and the dc/dc converter, deleting these
two elements.
When used, the MPPT and the dc/dc converter introduce losses,
although they are really low. For this reason, in most cases, the
hydrogen production in the direct connection is higher than in the
indirect connection. If the system is well designed because of the
little voltage variation in both working condition systems, the
crossing point of the two characteristic curves can be near the
maximum power point, even without the MPPT and the dc/dc
converter. In this case, the PV-cell number in a series has to be
established in order to produce an output voltage higher than the
minimum voltage required by the electrolyzer, which is the mini-
mum voltage to activate the electrolytic reaction in each cell. In
other words, the PV-plant can be designed in such a way that the
MPPs in several radiance and temperature conditions are close to
the electrolyzer characteristic curve 8.
10 The Optimizing Procedure
The optimization phase consists of determining the PV-cell
number in a series cells of the PV module in order to supply an
output voltage higher than the minimum value required by the
electrolyzer in order to activate the reaction in each cell.
A procedure to compute the PV cells connected in a series in
order to optimize the system in a hydrogen production point of
view has been established. In this case, the PV-cell number in a
series is not xed, taking into account the commercial modules,
which are typically composed of 36 or 72 cells. The PV cells in a
series are established in order to obtain the maximum hydrogen
production directly connecting the PV-plant and the electrolyzer
without using the MPPT and the dc/dc converter. Once the xed
PV-plant size the total number of PV cells is tilted 30 deg, it
produces a maximum power of 24.8 kW in Rome. The electro-
lytic cells number can be calculated, considering that each one
requires 1.24 kW, so that the electolyzer is properly connected at
24.8 kW, consisting of 20 cells characterized by a maximum volt-
age of 38 V. Dividing this quantity for the voltage value that
characterizes the PV cell during summer in accordance with the
maximum power point, the PV-cell number in a series is obtained,
which, in this case, equals to 86.
In this way, the crossing point between the PV and electrolyzer
characteristic curves is very close to the maximum power point
when the electricity production is the highest of the year. Further-
more, the working voltage of both electrolyzer and PV-plant does
not vary signicantly; therefore, the working conditions are al-
ways near the maximum power point, even in a direct connection.
The total cell number of the established PV-plant is n
cell
as
follows:
n
cell
= 630 36 = 22,680
Dividing the total cell number of the PV-plant for 86, the number
of modules connected in parallel is calculated.
Therefore, following, step by step, the block diagram in Fig. 16,
the PV-electrolyzer system has been optimized, and its features
are summarized in Table 6.
Therefore, when the cells number connected in a series of a
commercial module is xed, the matching point between the PV
and electolyzer characteristic curves can be located far from the
maximum power point, as the 30 deg tilted case discussed in this
paper Table 5 and Fig. 12. The hydrogen production is actually
the highest obtained by the xed considered number of PV cells.
11 Conclusions
The PV-electrolyzer system optimization has been analyzed
considering both the direct connection and the interposition of
MPPT, plus a dc/dc converter between the PV-plant and the elec-
trolyzer.
The design procedure to optimize the whole hydrogen produc-
tion system can be schematized in ve steps.
1. Establish the PV-plant size.
2. The PV-plant maximum power actually realized is computed
in 1 year simulation.
3. The electrolyzer size is computed on the basis of the maxi-
mum PV-plant power output, considering the losses in the
wires and the electrolyzer elasticity.
4. The PV-cell number in a series is established dividing the
maximum voltage required by the electrolyzer for the PV-
cell voltage output during summer.
5. Dividing the total amount of PV cells for the PV cells in a
series is how the string number is established.
The considered plant located in Rome designed on the basis of
this procedure produces the highest hydrogen quantity in the di-
rect connection, in correspondence of 30 deg tilted modules. In
Table 5 The PV-plant data considering 36 cell commercial modules
Tilt
deg Series Parallel
Power Electrolyzer
kW n
cell
H
2dir
N m
3

H
2ind
N m
3

0 2 315 21 17 8863.6 8063.8


30 3 210 24.8 20 8858.9 9110.1
60 2 315 21 17 9269.1 8437.6
90 2 315 17.3 14 6290.0 6276.5
Journal of Solar Energy Engineering AUGUST 2010, Vol. 132 / 031016-7
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
Table 7, the results obtained in correspondence of the 30 deg tilted
commercial modules 36 cells, 60 deg tilted commercial panels
36 cells, and 30 deg tilted optimizing modules 86 cells in se-
ries are reported in order to compare the maximum hydrogen
production and the whole system efciency in the most interesting
cases. The efciency is the ratio between the corresponding en-
ergy of the hydrogen produced in 1 year, which is the hydrogen
quantity multiplied for the LHV, and the maximum total radiance
energy upon the PV-plant surface during the same period, which is
obtained in correspondence of the 30 deg tilt, and it equals
1804.4 kWh/ m
2
.
The hydrogen production in the optimized plant 30 deg opti-
mization is higher than the hydrogen obtained with the commer-
cial modules, and it is even higher than the hydrogen quantity
generated using MPPT and the dc/dc converter, allowing savings
of about 2500 /kW, which is roughly the electronic device price.
In fact, because of the relatively smooth voltage variation in both
the PV output and electrolyzer required input, the above described
procedure allows designing the whole system in order to have
working conditions close to the MPP even in a direct connection
during the year.
The coupling voltage has been noticed to be the key quantity
for the optimization of the whole system in order to obtain the
highest hydrogen production and efciency.
The inverted procedure can be adopted to establish the PV-plant
size on the basis of the xed hydrogen production.
The future work will focus on verifying that the simulation
results match the real data of a PV-electrolyzer system in demon-
strative projects. In particular, the possibility of deleting the dc/dc
converter between the two systems has to be veried in real con-
ditions because the sudden variation in PV-plant outputs could
create problems in the whole working production system. The
chopper, in fact, is not only a voltage converter, but also an elec-
tric protection against dangerous transient working conditions.
Nomenclature
dc direct current
A area of electrode, m
2
I current, A
n
cells
number of electrolyzer cells
r parameter related to Ohmic resistance of elec-
trolyte, W m
2
s coefcient for overvoltage on electrodes, V
t coefcient for overvoltage on electrodes,
A
1
m
T temperature
T
a
ambient temperature, C
T
r
array reference temperature, 28C for commer-
cial polycrystalline PV modules
T
module
PV module temperature, C
G radiance, W/ m
2
U voltage, V
U
rev
reversible voltage; for water, it equals to 1228
V at 6 bars and 80C
U
t
coefcient of thermal exchange between cell
and ambient =800/ NOCT20, W/ m
2
K
t transparent coefcient
a adsorption coefcient
NOCT normal operative cell temperature, which is the
PV-cell temperature when T
a
=20 deg and G
=800 W/ m
2
efciency

r
array reference efciency

pt
efciency of power tracking equipment

rif
cell efciency at standard condition
b
t
temperature coefcient of efciency; it is as-
sumed to be a constant, and for silicon cells,
the range is 0.0040.006 per C.
I
cco
short circuit current at standard condition
V
oco
open circuit voltage at standard condition
N constant of diode ideality
h
i
I
cco
coefcient of variance with the
temperature
Fig. 16 Optimizing procedure diagram block
Table 6 The PV-plant data considering 86 cell modules
PV-plant
Cells number in series 86
Parallel number 264
Maximum power 27 kW
Electrolyzer
Size 24.8 kW
Cells number 20
Produced hydrogen 10,133.4 N m
3
Working hours 3485
Coupling maximum power 25.3 kW
Table 7 PV-hydrogen plant results
Tilt 30 deg 60 deg 30 deg optimization
PV cells in series 108 72 86
Modules in parallel 210 315 264
PV surface m
2
226.8 226.8 227.04
Power Electrolyzer kW 24.8 21 24.8
H
2dir
N m
3
8858.9 9269.1 10,133.4
H
2ind
N m
3
9110.1 8437.59 9170.95
Global maximum efciency 0.0665 0.0677 0.0739
031016-8 / Vol. 132, AUGUST 2010 Transactions of the ASME
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms
h
v
coefcient of variance with the temperature
V
oco
B coefcient of variance with the temperature
A
c
PV-cell area
P
cell
power feeding each electrolytic cell, equal to
1238 W
ICE internal combustion engine
LHV lower heating value
References
1 Orecchini, F., 2006, The Era of Energy Vectors, Int. J. Hydrogen Energy,
31, pp. 19511954.
2 Orecchini, F., and Naso, V., 2003, La societ NO OIL, Orme Editori, Milan,
Italy.
3 Meurer, C., Barthels, H., Brocke, W. A., Emonts, B., and Groehn, H. G., 1999,
PHOEBUSAn Autonomous Supply System With Renewable Energy: Six
Years of Operational Experience and Advanced Concepts, Sol. Energy, 67,
pp. 131138.
4 Ghosh, P. C., Emonts, B., Janen, H., Mergel, J., and Stolten, D., 2003, Ten
Years of Operational Experience With a Hydrogen-Based Renewable Energy
Supply System, Sol. Energy, 75, pp. 469478.
5 Cacciola, G., Orecchini, F., and Vellone, R., 2003, IdrogenoEcoenergie,
ISES Italia, Rome, Italy.
6 Ronchetti, M., and Iacobazzi, A., 2002, Celle a CombustibileStato di Svi-
luppo e Prospettive Della Tecnologia, ENEA, Rome, Italy.
7 Mohan, N., Undeland, T. M., and Robbins, W. P., 1989, Power Electronics:
Converts, Applications, Design, Wiley, New York.
8 Zuccari, F., Santiangeli, A., Artuso, P., and DellEra, A., 2008, Proceedings of
the 63 Congresso Nazionale ATI, 63 National Congress ATI, Sept. 2326.
9 ystein, U., 2003, Modeling of Advanced Alkaline Electrolyzers: A System
Simulation Approach, Int. J. Hydrogen Energy, 28, pp. 2133.
10 2003, Getting Started With LabVIEW, National Instruments Corporation, Aus-
tin, TX.
11 2003, LabVIEW User Manual, National Instruments Corporation, Austin, TX.
12 Tani, T., Sekiguchi, N., Sakai, M., and Ohta, D., 2000, Optimization of Solar
Hydrogen Systems Based on Hydrogen Production Cost, Sol. Energy, 68, pp.
143149.
13 Shakya, B. D., and Aye, L., and Musgrave, P.,2005, Technical Feasibility and
Financial Analysis of Hybrid WindPhotovoltaic System With Hydrogen
Storage for Cooma, Int. J. Hydrogen Energy, 30, pp. 920.
14 Cucumo, M. A., Marinelli, V., and Oliveti, G., 1994, Ingegneria Solare. Prin-
cipi ed Applicazioni, Pitagora, Bologna, Italy.
15 Ferro, P., 2004, I Sistemi Fotovoltaici: Progettazione Tecnico-Architettonica,
ISES Italia, Rome, Italy.
16 Bassoli, R., Messana, C., and Vigotti, R., 1992, Energia dal Sole, Hypothesis,
ISES Italia, Rome, Italy.
Journal of Solar Energy Engineering AUGUST 2010, Vol. 132 / 031016-9
Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/01/2014 Terms of Use: http://asme.org/terms

You might also like