INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH

D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

MATLAB Simscape Model of An Alkaline

Electrolyser and Its Simulation with A Directly
Coupled PV Module

David Martinez and Ramon Zamora*

Electrical and Electronic Engineering Department, Auckland University of Technology, Auckland 1010, New Zealand
(david.martinez@outlook.co.nz, ramon.zamora@aut.ac.nz)

*Corresponding Author; Dr. Ramon Zamora, Electrical and Electronic Engineering Department, Auckland University of
Technology, Auckland 1010, New Zealand, Tel: +64-9-921-9999, ramon.zamora@aut.ac.nz

Received: 22.11.2017 Accepted:28.01.2018

Abstract- This paper discusses a MATLAB/Simulink/Simscape power system implementation of an alkaline electrolyser
directly coupled to a photovoltaic module. Meteorological average daily irradiance and temperature data for a typical summer
and winter day in Auckland, New Zealand, is used in a 15-hour simulation duration to show characteristic responses of the
system. In addition to the steady state output, boundary conditions are applied to the referenced mathematical models to
include the transient performance, governed by the photovoltaic module internal parameters. The simulation results show that
the steady state output of the proposed implementation matches the general response of similar systems reported in previous
published papers. Overall electrolyser efficiency of 68.48% for summer and 63.32% for winter is in accordance with results
from the referenced studies.
Keywords- Alkaline electrolyser; Simscape; hydrogen; PV module; irradiance.

1. Introduction Using DC current, a water molecule can be broken into

Energy storage systems are considered the key factor to hydrogen and oxygen. This process occurs inside an
improve reliability and stability of renewable-energy- electrolyser which consists of two electrodes and an
based electricity [1]. For intermittent renewable energy electrolyte. Depending on characteristics of the electrolyte,
sources, such as solar and wind, energy storage not only three different types of electrolyser exist: alkaline, proton
provides a way to match generation and load profiles but exchange membrane (PEM) and solid oxide (SOE) [2].
also serves as an auxiliary source to deliver power when Due to its high maturity and economical costs, alkaline
renewable sources are unavailable. electrolysis is the most commonly used technology to
Due to its sustainable production, hydrogen energy produce hydrogen using water [9]. Even though the
storage can be considered as the most promising energy technology is already advanced, many challenges still exist,
storage technology [2]–[5]. Like electricity, hydrogen can primarily in efficiencies, overall system design and range
be produced from renewable or non-renewable sources, of operation. Simulation is essentially important to further
and can have a wide range of environmental impact. analyse these issues.
Among those classified as renewable, water electrolysis is In the last three decades, several mathematical models
considered the best mechanism for hydrogen production have been proposed describing the operation of the
due to its high sustainability with low or no harmful electrolyser [10]. Models developed empirically from
emissions [6]–[8]. demonstration projects have received most attention due to
their accuracy to represent operation characteristics of real
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

devices. Specifically, the model in [11] has been used in (1)

many studies showing not only accuracy but high
adaptability to describe various real electrolysers. where: T is the temperature of the cell, A is the area of the
Among existing simulation platforms, MATLAB offers electrodes, r1 and r2 are the ohmic overvoltage parameters,
several benefits to model any type of systems with a and t1, t2, t3, and s are the activation overvoltage
growing community support [12]. Within the parameters.
MATLAB/Simulink environment, Simscape allows the is the reversible voltage of the cell and is
exploration of mathematical models in their own physical temperature and pressure dependant with a value of 1.229
domain simplifying the understanding of their real V at standard conditions (1 bar, 25 oC). For low
behaviour and expanding the possibilities of temperature ranges of up to 100 0C, voltage variation is
interconnecting fundamental elements [13]–[15]. In the small and can be considered constant [11].
case of alkaline electrolysers, an electrical equivalent Production rate of hydrogen can be related to the input
circuit has been proposed in [16], but the mathematical current using Faraday law. An empirical expression for the
model interaction with the MATLAB/Simulink/Simscape hydrogen production efficiency is called Faraday
circuit is not clearly defined. efficiency [11]. This relationship is summarized in Eq. (2).
This work will focus on the construction of a Simscape
power system model of an alkaline electrolyser based on (2)
[11] and its evaluation by coupling a photovoltaic module where: f1 and f2 are parameters related to Faraday
for an average solar irradiance profile in Auckland, New efficiency, z is the number of electrons transferred in the
Zealand. Initially, Simulink simscape models are reaction (2 electrons for water electrolysis), F is Faraday
constructed considering not only mathematical equations constant and is the molar flow rate per second.
but also real data from several studies. Then, both f1 and f2 are usually considered constant for the overall
elements are characterized to assess the validity of their model. However, they vary with temperature [11]. The
response against expected outcome. Finally, a 15-hour assumption of linear relation with temperature in [11]
simulation is run, using meteorological data from NIWA depends on the temperature range of operation. In this
[17], to observe the electrical interaction of the elements. research, a linear expression only applies for f1 whereas for
f2 a limit value must be set to guarantee a range value
2. Model Presentation and Simulink Block between 0 and 1. Using the experimental data shown in
Construction Table 1, a linear empirical expression is obtained for f1 as
2.1. Alkaline Electrolyser shown in Eq. (3) and a quadratic for f2 as shown in Eq. (4).
Note that an extra point was added to f2 at 0 oC to
Water can only be split into hydrogen and oxygen guarantee a limit value of 1.
when a minimum electrical voltage is applied. Ideally the
cell voltage is equal to this minimum voltage, called (3)
reversible voltage, but due to irreversibility, the cell (4)
voltage is always higher [9].
The extra voltage present in the cell is called A unit conversion is needed for the flow rate in Eq.
overvoltage and is composed of ohmic, activation and (2) to obtain a rate in kilograms rather than moles. This
diffusion overvoltages. Ohmic overvoltage is the will facilitate comparisons in the result section. Using the
overvoltage due to overall resistance of all components of volume of an ideal gas at standard conditions , the
the electrical circuit, while the activation overvoltage is the volumetric flow rate and the mass flow rate is given by Eq.
overvoltage related to extra energy required to start the (5).
oxygen and hydrogen formation process in the electrodes (5)
[18]. Diffusion overvoltage only takes precedence at high where: = 0:0224136 m3/mol and c = 0.08988 kg/m3.
current densities when the reaction changes from
electronic transfer to matter transfer [19]. Table 1. Electrolyser experimental data [11]
An empirical model involving ohmic and activation Temperature [oC] f1 f2
overvoltages proposed in [11] is expressed by Eq. (1). The
0 1.00
model has six different parameters to characterize the
logarithmic response of an electrolyser cell voltage in 40 150 0.99
terms of current and temperature. 80 250 0.96

553
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

2.2. Boundary Conditions for the Alkaline Electrolyser added to the model to produce hydrogen only if the cell
voltage is greater than . The final set of boundary
Due to the presence of a logarithmic function in Eq. (1), conditions can be expressed in equations (6) and (7).
a careful consideration is needed to avoid negative values
that can produce unrealistic results. This is obvious at low Using Matlab/Simulink math blocks, equations (1)–(7)
levels of current where the model parameters cannot are constructed for each input and output. The electrolyser
follow the electrolyser response properly. components in detail are shown in Figs. 1 and 2.
Similarly, when input current is zero in Eq. (1), the
electrolyser cell is equal to the reversible voltage and can Simscape electrical interaction is added by utilizing
potentially act as a voltage source feeding current to other voltage/current controlled sources blocks, voltage/current
elements of the electrical circuit. To avoid this situation, measurement blocks, and electrical connection ports from
the mathematical element is disconnected from the circuit the power systems specialized technology toolbox.
to block any current flow and to force the voltage to follow For the electrolyser diagram shown in Fig. 3, the
the external circuit. boundary condition of zero current was implemented with
Finally, production rate given by Eq. (2) only depends a diode between the mathematical block subsystem and the
on the input current regardless of the voltage input. This is power systems blocks. Finally, a memory block was added
not accurate since water can only be divided into hydrogen to the electrolyser model to break the algebraic loop and to
and oxygen when the cell voltage is enough to provide the facilitate the convergence of the solution when other
required activation energy. A special consideration is elements are connected.

(6)

(7)

Fig. 2. Hydrogen production block diagram.

Fig. 1. Electrolyser block diagram.

Fig. 3. Electrolyser simscape power systems integration.

554
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

present even with no input energy. To avoid this situation

2.3. PV Module a boundary condition is applied to the diode current in
The semiconductor operation of a photovoltaic cell can terms of the photo generated current as shown in Eq. (13).
be represented with the equivalent circuit shown in Fig. 4.
The PV parameters are defined as follows: is photo- (13)
generated current, is diode current, is current through
parallel resistor, is diode voltage, is series resistance, The equivalent circuit represented by equations (8) or
and is parallel resistance. (9) can be easily implemented in
Based on Kirchhoff’s current law (KCL), the output MATLAB/Simulink/Simscape with the power system
current is represented by Eq. (8). specialized technology toolbox as shown in Figs. 5 and 6;
(8) The boundary condition for diode current is implemented
with a simple Simulink switch. Simulink block details for
where: and , equations (10)–(12) are not presented here since they have
been presented in [20].
or

(9)
where: is the reverse saturation current of the diode, T is
the cell temperature, is an ideality factor representing the
type of technology, is the number of cells connected in
series, k is Boltzmann constant and q is the electron charge.

Fig. 5. PV diode current simscape power systems

integration.

Fig. 4. PV module equivalent circuit.

The photo-generated current and the reverse saturation

current are related to the physical parameters of a real PV
module in [20] as expressed in equations (10)–(12).
(10)

(11)
Fig. 6. PV module simscape power systems diagram.
(12)
3. Characterization of the Elements
where: is short circuit current, is open circuit
voltage, is irradiance input, is irradiance at STC = To observe the response of each model to several
1000 W/m , 2 o
is temperature at STC = 25 C, and is temperatures, a variable input was introduced as a current
source for the electrolyser and a variable resistor for the
the material band gap energy, 1.12 eV for Si.
PV module. Ambient temperature values were selected
2.4. Boundary Condition for the PV Module according to average mean monthly minimum and
maximums for Auckland: January/February average max.
The output current of the equivalent circuit represented is 27 oC and July/August average min is 5 oC [21].
by Eq. (8) can generate negative values of current even if The parameters describing each model shown in Table
the photo generated current is zero. However, this leads 2 were tuned for the following equipment in the mentioned
to an inaccurate result where the diode current is still

555
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

studies: ELECTROCELL Micro Flow Cell [9] for the graphs match their mathematical representation. The
electrolyser and PWX 500 49W [20] for the PV module. impact of increasing the temperature in terms of voltage is
positive for the electrolyser due to less cell overvoltage
The I-V curves for the electrolyser and PV module are
and negative for the PV module due to less power
shown in Figs. 7 and 8 respectively. For both models the
delivered.

Table 2. List of parameters for both models

Electrolyser Value Units PV Module Value Units
A 0.01 m2 Isc 3.11 A
r1 3.538550x10-4 Ω m2 Voc 21.8 V
r2 -3.02150x10-6 Ω m2/oC Ns 36 None
s 0.22396 V n 1.3 None
t1 5.13093 m /A2
Rp 310.0248 Ω
t2 -2.40447 x102 o
C m2/A Rs 0.45 Ω
3 o 2 2
t3 3.410251 x10 C m /A

4. System Simulation

To evaluate how the two electrical models interact with

each other, the electrical ports of the two blocks were
directly connected without implementing any
voltage/current control or maximum power tracking, as
shown in Fig. 9. Voltage and current are directly measured
from the circuit while hydrogen production is obtained by
adding an integrator to the hydrogen mass flow rate output
of the electrolyser. The daily fluctuation of irradiance and
temperature data are inputted from external csv files.
The months with maximum and minimum daily
irradiation are evaluated in conjunction with average data
for an entire year. Based on Auckland irradiance data from
Fig. 7. Electrolyser I-V curve.
NIWA’s SolarView [17], the irradiation values are 5.13
kWh/m2 in January and 2.55 kWh/m2 in June. The
simulation is only performed during daylight hours
considering January as the reference. The start time was
selected to be close to dawn at 5 am and the end time was
selected just after sunset at 8 pm, for a total of 15 hours of
simulation time. The step size was arbitrary selected to be
60 seconds resulting in 901 data points.
Hourly average irradiance is given per hour for a total
of only 15 points. To obtain the missing points
representing 60 second intervals within each hour, an
algorithm was created to generate a base sinusoidal curve
of irradiance and then fit it to the given hourly average
data. Working by hour intervals, the algorithm first
calculates the average value of the base model, compares it
to the average data and depending on the result, values are
Fig. 8. PV module I-V curves. reduced or increased until the difference is below 0.01. A
visualisation of the January data analysis is shown in Fig.

556
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

10. The error between the total cumulative values of the guarantying a correct input for the simulation.
initial data and the fitted curve was only 0.04%,

Fig. 9. System interconnection.

Fig. 10. Input irradiance data – January. Fig. 11. Input Temperature data – January.

Temperature data from NIWA in [21] gives hourly 5. Results and Discussion
average values for a total of 15 points. To obtain the
Since the voltage variation of the electrolyser cell for
missing points representing 60 second intervals within
all given temperatures is much lower than the knee of the
each hour, the algorithm generates a series of steps equal
I-V curve of the PV module (Vrev + 1.5 V << Voc), the
to the average temperature of the specific hour. A
module acts as a current source proportional to the
visualisation of the data analysis for January is shown Fig.
irradiance. This proportional relationship is shown in Fig.
11. For average hourly temperature in a year, the mean
12 with both irradiance and current curves superimposed
value is calculated between the available data, which are
for all evaluated cases. The compared studies also show
summer maximum and winter minimum.
the same responses where current follows the irradiance
To assess the validity of the patterns.
MATLAB/Simulink/simscape power systems model, four
The voltage curve in Fig. 13 shows a different pattern
previous studies with similar simulation architecture were
where two zones can be clearly identified. Steady state
selected. Study A [22] ran a simulation using irradiance
zone A is where the electrolyser is fully active and
data for Miami USA, study B [23] used real devices for
transient zone B is a startup/shutdown region where
irradiance in Cairo Egypt, study C [24] used a real
voltage fluctuation is predominant.
experimental setup in Algeria and study D [25] ran a
simulation using irradiance data for Pamplona Spain. The shape of the output curves agrees with all four
Unfortunately, due to the size difference of each system, compared studies for the steady state zone A. However,
the comparison mostly just involves the general trend of the zone B characteristics cannot be compared since all
the output, following the shape of the curves for each referenced papers ignored the transient response.
variable analysed.

557
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

Fig. 12. Input Irradiance and output current.

Fig. 13. Voltage curves.

Fig. 14. Zone B region: voltage and current (left), voltage and irradiance (right).
To further assess the response of the transient zone B, a Wh and for the winter day 19.668 Wh. Based on Eq. (5),
closer examination is performed by analysing the specific the amount of hydrogen produced by the single
times of occurrence and the level of variation. By visual electrolyser cell was found to be 0.5856 grams in summer
inspection, for January data, zone B finishes around 6:30 and 0.29 grams in winter which represents 19.3248 Wh
am and starts again at 6:30 pm. For June data, zone B and 9.57 Wh, respectively. The electrolyser efficiency is
finishes at 8 am and starts again at 5 pm. After a careful then calculated to be 68.48% and 63.32% for summer and
inspection of the data, with voltage superimposed with winter, respectively. These values fall in the range of
both irradiance and current shown in Fig. 14, the variation reported efficiency values in [2] and [18]. The energy and
in voltage occurs in the PV module due to a low-level volume were calculated based on the hydrogen energy
input of irradiance that is not high enough to generate content of 120 MJ/kg or 33 kWh/kg and the volumetric
current to the external circuit, dissipating through the PV density of 0.089 kg/m3 [26], [27].
module parallel resistance. The resulting voltage is then
To allow a further comparison with the reference
directly proportional to the irradiance and the internal
studies, instantaneous efficiency and flow rate was plotted
parallel resistance of the PV module.
in Fig. 15 for January, June and year average irradiance
The amount of electrical energy inputted to the system values. As expected from Eq. (2), the flow rate follows
can be found by calculating the area under the power curve. current. This general outcome matches the curves in the
For the summer day, the total input energy was 26.1625 four reference studies.

558
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

Fig. 15. Flow rate (left) and instantaneous efficiency (right).

6. Conclusions [4] B. Mandal, A. Sirkar, A. Shau, P. De, and P. Ray,
“Effects of Geometry of Electrodes and Pulsating DC
A direct coupling of an alkaline electrolyser cell and a Input on Water Splitting for Production of Hydrogen”,
PV module was implemented in International Journal of Renewable Energy Research,
MATLAB/Simulink/Simscape power systems. vol. 2, no. 1, pp. 99-102, 2012.
Meteorological irradiance data from Auckland was used to [5] Y. Allahvirdizadeh, M. Mohamadian, and M.
simulate and verify the model. Boundary conditions were HaghiFam, “Study of Energy Control Strategies for a
added to the referenced mathematical models to obtain an Standalone PV/FC/UC Microgrid in a Remote Area”,
accurate transient response of the system. Voltage, flow International Journal of Renewable Energy Research,
rate and current steady state outputs matched previous vol. 7, no. 3, pp. 1495-1508, 2017.
studies while the transient response showed a direct [6] O. Bilgin, “Evaluation of hydrogen energy
relation to the PV internal parallel resistance. Overall production of mining waste waters and pools”, 4th
electrolyser efficiency of 68.48% for summer and 63.32% International Conference on Renewable Energy
for winter agrees with results provided in the referred Research and Applications, Palermo, pp. 557-561,
published articles. November 2015.
Further work is needed to add the thermal response of [7] S. Kaya, B. Öztürk, and H. Aykaç, “Hydrogen
the model, storage alternatives and include other elements Production from Renewable Source: Biogas”,
of the Simulink/Simscape family to allow a multi physical International Conference on Renewable Energy
domain simulation. Research and Applications, Madrid, pp. 633-637,
October 2013.
References [8] K. Koiwa, R. Takahashi, and J. Tamura, “A Study of
Hydrogen Production in Stand-alone Wind Farm”,
[1] B. Zakeri and S. Syri, “Electrical energy storage 2012 International Conference on Renewable Energy
systems: a comparative life cycle cost analysis”, Research and Applications, Nagasaki, pp. 1-6,
Renewable and Sustainable Energy Views, vol. 42, November 2012.
pp. 569-596, Ferbuary 2015. [9] E. Amores, J. Rodriguez, and C. Carreras, “Influence
[2] F. Zhang, P. Zhao, M. Niu, and J. Maddy, “The of operation parameters in the modelling of alkaline
survey of key technologies in hydrogen energy water electrolysers for hydrogen production”,
storage”, International Journal of Hydrogen Energy, International Journal of Hydrogen Energy, vol. 39, pp.
vol. 41, pp. 14535-14552, September 2016. 13063-13078, 2014.
[3] S. Dahbi, I. Mazozi, M. E. Ouariachi, A. Messaoudi, [10] P. Oliviera, C. Bourasseaua, and B. Bouamamab,
and A. Aziz, “Implementation of a Multi-control “Low-temperature electrolysis system modelling: A
Architecture in a Photovoltaic/ Grid/ Electrolysis review”, Renewable and Sustainable Energy Reviews,
System for Usual Use and Clean Storage by vol 78, pp. 280-300, 2017.
Hydrogen Production”, International Journal of [11] O. Ulleberg, “Modeling of advanced alkaline
Renewable Energy Research, vol. 7, no. 4, pp. 1825- electrolyzers: a system simulation approach”,
1835, 2017. International Journal of Hydrogen Energy, vol. 28, pp.
21-33, 2003.

559
INTERNATIONAL JOURNAL OF RENEWABLE ENERGY RESEARCH
D. Martinez and R. Zamora. ,Vol. 8, No. 1, March, 2018

[12] MathWorks, “Matlab”, [Online]. Available: [20] H. Bellia, R. Youcef, and M. Fatima, “A detailed
https://www.mathworks.com/products/matlab/feature modeling of photovoltaic module using MATLAB”,
s.html. [Accessed: 20-Nov-2017]. NRIAG Journal of Astronomy and Geophysics, vol.
[13] MathWorks, “Simscape: Model and simulate 3, pp. 53-61, 2014.
multidomain physical systems”, [Online]. Available: [21] P.R. Chappel, “The Climate and Weather of
https://www.mathworks.com/products/simscape.html. Auckland, 2nd Edition” NIWA Science and
[Accessed: 20-Nov-2017]. technology series, Num. 60, pp. 23, [Online].
[14] M. R. Rashel, A. Albino, M Tlemcani, and T. C. F. Available:
Goncalves, “MATLAB Simulink modelling of https://www.niwa.co.nz/static/Auckland%20Climate
photovoltaic cells for understanding shadow effect”, WEB.pdf. [Accessed: 10-Nov-2017].
5th International Conference on Renewable Energy [22] A. Khalilnejad, A. Abbaspour, and A.I. Sarwat,
Research and Applications, Birmingham, pp.747-750, “Multi-level optimization approach for directly
November 2016. coupled photovoltaic-electrolyser system”,
[15] K. Jazayeri, S. Uysal, and M. Jazayeri, International Journal of Hydrogen Energy, vol. 41, pp.
“MATLAB/simulink based simulation of solar 11884-11894, 2016.
incidence angle and the sun's position in the sky with [23] G.E. Ahmad and E.T. El Shenawy, “Optimized
respect to observation points on the Earth”, photovoltiac system for hydrogen production”,
International Conference on Renewable Energy Renewable Energy, vol. 31, pp. 1043-1054, 2006.
Research and Applications, Madrid, pp.173-177, [24] A. Djafour, M. Matoug, H. Bouras, B. Bouchekima,
October 2013. M.S. Aida, and B. Azoui, “Photovoltaic-assisted
[16] C. Henao, K. Agbossou, M. Hammoudi, Y. Dubé, alkaline water electrolysis: Basic principles”,
and A. Cardenas, “Simulation tool based on a physics International Journal of Hydrogen Energy, vol. 36, pp.
model and an electrical analogy for an alkaline 4117-4124, 2011.
electrolyser”, Journal of Power Sources, vol. 250, pp. [25] A. Ursua, I. S. Martin, E. L. Barrios, and P. Sanchis,
58-67, 2014. “Stand-alone operation of an alkaline water
[17] The National Institute of Water and Atmospheric electrolyser fed by wind and photovoltaic systems”,
Research, “SolarView”, [Online]. Available: International Journal of Hydrogen Energy, Vol. 38,
https://solarview.niwa.co.nz/. [Accessed: 20-Nov- pp. 14952-14967, 2013.
2017]. [26] K. Lu, Materials in Energy Conversion, Harvesting,
[18] K. Zeng and D. Zhang, “Recent progress in alkaline and Storage, Hoboken, New Jersey: John Wiley &
water electrolysis for hydrogen production and Sons, Inc. 2014, pp. 387.
applications”, Progress in Energy and Combustion [27] Universal Industrial Gases, Inc., “Unit Conversion
Science, vol. 36, pp. 307-326, 2010. Data for Hydrogen”, [Online]. Available:
[19] P. Oliviera, C. Bourasseaua and B. Bouamamab, http://www.uigi.com/h2_conv.html. [Accessed: 15-
“Low-temperature electrolysis system modelling: A Nov-2017].
review”, Renewable and Sustainable Energy Reviews,
vol 78, pp. 280-300, 2017.

560