i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Modelling and control of hydrogen and energy flows

in a network of green hydrogen refuelling stations powered
by mixed renewable energy systems

Hanane Dagdougui a,b, Ahmed Ouammi a,c,*, Roberto Sacile a

a
Department of Communication, Computer and System Sciences (DIST), Faculty of Engineering, University of Genoa, Genoa, Italy
b
MINES ParisTech, Sophia Antipolis Cedex, France
c
Unité des Technologies et Economie des Energies Renouvelables, CNRST B.P 8027 NU Rabat, Morocco

article info abstract

Article history: The planning of a hydrogen infrastructure with production facilities, distribution chains,
Received 7 June 2011 and refuelling stations is a hard task. Difficulties may rise essentially in the choice of the
Accepted 22 July 2011 optimal configurations. An innovative design of hydrogen network has been proposed in
Available online 10 September 2011 this paper. It consists of a network of green hydrogen refuelling stations (GHRSs) and
several production nodes. The proposed model has been formulated as a mathematical
Keywords: programming, where the main decisions are the selection of GHRSs that are powered by
Green hydrogen refuelling station the production nodes based on distance and population density criteria, as well the energy
Renewable energy sources and hydrogen flows exchanged among the system components from the production nodes
Sustainability to the demand points. The approaches and methodologies developed can be taken as
Mathematical programming a support to decision makers, stakeholders and local authorities in the implementation of
Optimization new hydrogen infrastructures. Optimal configurations have been reported taking into
account the presence of an additional hydrogen industrial market demand and a connec-
tion with the electrical network. The main challenge that has been treated within the paper
is the technical feasibility of the hydrogen supply chain, that is mainly driven by uncertain,
but clean solar and wind energy resources. Using a Northern Italian case study, the clean
hydrogen produced can be technically considered feasible to supply a network of hydrogen
refuelling stations. Results show that the demands are satisfied for each time period and
for the market penetration scenarios adopted.
Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

1. Introduction switched gradually to clean, affordable and reliable energy

The global awareness concerning greenhouse gas (GHG) emis-
sions, air pollution, fossil fuel depletion and others en-
ergy security issues [1,2] have led many governments and
researchers around the world to develop secure and environ-
mental friendly fuel. The current fossil fuel systems must be
systems, thus to reach the global drivers for a sustainable vision
of our future energy market. Among many alternative energy
sources, hydrogen can be considered as an attractive solution to
succeed the current carbon-based energy system.
The main benefits of hydrogen are even substantially
considered by the fact that hydrogen can be manufactured

* Corresponding author. Unité des Technologies et Economie des Energies Renouvelables, CNRST B.P 8027 NU Rabat, Morocco.
E-mail addresses: hanane.dagdougui@unige.it (H. Dagdougui), ouammi@cnrst.ma (A. Ouammi), roberto.sacile@unige.it (R. Sacile).
0360-3199/$ e see front matter Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2011.07.096
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
from a number of primary energy sources, such as natural gas,
nuclear, coal, biomass, wind and solar energy. Such diversity
in production, obviously contributes significantly in diversi-
fying the energy supply system and in ensuring the security of
fuel supply. For transport applications, there will be an
increasing requirement to use clean and low or zero emission
fuels such as hydrogen [3]. In addition, hydrogen is the most
abundant element on the earth, it is clean and has the highest
specific energy content of all conventional fuels [4]. Hydrogen
can contribute to a diversification of automotive fuel sources
and supplies and can offer long-term solution being solely
produced from renewable energies. The development of
a hydrogen infrastructure for producing and delivering
hydrogen appears as a key factor to achieve the hydrogen
economy transition and its development. In fact, the model-
ling of hydrogen infrastructure is still a complex task, the
main complexities rise from the significant uncertainties in
demand, supply, economic and environmental impacts, and
in the diversity of technologies available for production,
storage and transportation. The key question is from which
sources hydrogen can be produced in a sustainable manner
[5]. The extent to which the hydrogen benefits will occur has
a great dependency on the technologies involved. Many
authors have agreed that renewable energy sources (RES),
such as wind and solar are central for better transition to
a long-term hydrogen economy.
In order to reach that goal, it is advantageous to use
renewable energy for hydrogen generation. In fact, the
resources for the operation of renewable energy systems are
inexhaustible and practically free making. In addition,
sustainable hydrogen production from electrolysis yields
several advantages from a system point of view [6]. For
instance, wind-powered water electrolysis ranks high in
terms of technical and economical feasibility, having a great
potential to become the first competitive technology to
produce large amounts of renewable hydrogen in the future
[7e11].
From the end users perspective, the use of hydrogen in
fuel cell applications offers a number of advantages over
existing fuels and other emerging competitors, especially in
the transportation sector [12]. The fuel cell vehicles can be
a long-term solution to the persisted environmental prob-
lems associated with transportation. The fuel cell vehicles
would be less complex, have better fuel economy, lower GHG
emissions, greater oil import reductions and would lead to
a sustainable transportation system once renewable energy is
used to produce hydrogen [13]. According to Doll et al. [14],
the introduction of hydrogen coupled with the fuel cell
vehicles could reduce significantly the emissions of CO2, NOx
and SOx.
The transition to a sustainable hydrogen economy faces
paramount economic and technological barriers that must
be overcome in order to ensure a successful transition. It is
essential to study and analyse the interactions between
different hydrogen infrastructure components in advance
in order to set and build variety of options for the incor-
poration of this new economy. This will facilitate the
management of hydrogen supply chain, and help decision
makers to define adequate roadmaps for the hydrogen
developement.
5361
Many authors have detailed approaches and models for the
development of the future hydrogen infrastructures. The
approaches range from the examination of the supply chain
as a whole [15e17] to the focus on a node of the infrastructure
such as, production, storage or transportation [7,21e24]. In
[18], Kuby et al. have developed a model able to locate the
hydrogen stations that refuel maximum volume of vehicle
fuel, this latter is measured both using the number of trips and
vehicle miles travelled. Bersani et al. [19] have investigated the
planning of a network of service stations of a given company
within a competitive framework. They proposed a decision
support system that can be considered to determine the
optimal placement of service stations within a hydrogen
economy. Nicholas et al. [20] have provided an analytical
framework for locating hydrogen fuel stations assuming that
the existing petrol infrastructure is strongly related to the
needed hydrogen infrastructure of the future. In another
study, Parker et al. [21] have assessed the economic and
infrastructure requirements of the production of hydrogen
from agricultural wastes, they concluded that the delivery
price of bio-hydrogen is similar to the hydrogen produced
from the natural gas. Joffe et al. [22] have developed a tech-
nical modelling of a hydrogen infrastructure. They investi-
gated the operation of the system so to provide initial facility
for refuelling hydrogen fuel cell buses in London city. Greiner
et al. [9] have presented a simulation study of combined wind-
H2 plant on a small Norwegian island. They include chrono-
logical simulations and economic calculations enabling the
optimization of the components size. Their simulations
include a grid-connected system and an isolated system with
backup power generator. Dagdougui et al. [23] have introduced
a dynamic decision model for the real time control of hybrid
renewable energy production systems, which can be particu-
larly suitable for autonomous systems.
General interest in a wind-hydrogen system has increased
partly because the price of wind power has become
competitive with traditional power generating sources in
certain areas [24]. Wind-powered water electrolysis ranks
high in terms of technical and economical feasibility, having
a great potential to become the first competitive technology
to produce large amounts of renewable hydrogen in the
future [7]. Worldwide installations of wind turbine power
have reached a value of 194.5 GW [25]. Studies carried out by
Honnery and Moriarty [26] have evaluated the global poten-
tial of a coupled wind/hydrogen system, thus in order to
estimate the future hydrogen production.
A challenging task that is worth to be deeply studied
regards the feasibility to feed hydrogen demand points by an
uncertain renewable supply, such as the case of hydrogen
production from intermittent RES. The key question that
needs to be addressed is the ability of the renewable energy
system to meet the hydrogen fuel requirements (in amount
and time).
In this paper, an attempt has been made to plan an inno-
vative design of a hydrogen infrastructure. It consists of
a network of GHRSs and several production nodes. The
proposed model is formulated as a mathematical program-
ming, where the main decisions are the selection of GHRSs
that will be powered by each point of production based on
distance and population density criteria, as well the energy
5362 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
and hydrogen flows exchanged among the system compo-
nents from the production nodes to the demand points. The
approaches and methodologies developed can be taken as
a support to decision makers, stakeholders and local authori-
ties in the implementation of future hydrogen infrastructures.
2. The methodological approach
2.1. Model structure and components
Wind/solar energy production systems are designated to
supply electric and hydrogen needs to a network of GHRSs.
From the demand viewpoint, the system to be modelled
consists of a network of GHRSs each one having a local
hydrogen storage tank, and distributing hydrogen to
costumers (fuel cell vehicle users). From the supply viewpoint,
hydrogen production systems consist of many mixed energy
production plants. Each one includes a large scale hybrid
wind/solar system, electrolyser unit, fuel cell, and a main
hydrogen storage tank. Moreover, each production plant is
connected to the electrical network, giving the system the
possibility to sell excess of energy generated by the production
Fig. 1 e System description.
nodes. Fig. 1 displays the proposed hydrogen infrastructure
that forms the basis of the hydrogen global system.
2.2. Planning horizon: plants and technologies
The implementation of the planning phase is important for
the formulation of the optimization model. It will provide the
knowledge about the possibility of using the renewable
resources for hydrogen production. Furthermore, due to the
absence of a large hydrogen market, the development of the
hydrogen infrastructure will start by the introduction of small
hydrogen decentralized systems. Hence, the operation of
these systems as a whole will depend in addition to the
available renewable resources, to design the system that will
generate and fulfil specific requirements. In fact, the system
must be accurately adapted in order to fit the mass of
hydrogen for the GHRSs. Nevertheless, economically
speaking, the configuration of the best design will strongly
depend on the scenario taken into account. In addition, it will
depend on the renewable resources condition as well as the
cost of the different components and the final selling prices of
the energy generated by each type of renewable source [27].
The main idea of using mixed renewable energy systems for
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
hydrogen production is based on generating electrical energy
by the wind turbines and the photovoltaic modules, and then
using that energy for hydrogen generation using an electro-
lyser system. Owing to the low energy density by volume of
the hydrogen, the effective use of the logistics chain to
transport hydrogen to the refuelling stations requires the
reduction of the hydrogen volume, thus by means of lique-
faction or compression. In this paper, liquefaction or
compression of the hydrogen substance are not considered,
neither the distributing chain of hydrogen to the refuelling
stations.
2.2.1. Mixed renewable energy system
The selection of the suitable renewable energy technologies
has been made on the basis of several considerations: mete-
orological conditions, renewable energy potential, load factor
and amount of hydrogen required by the refuelling stations.
The problem needs much more attention, especially because
of the use of the intermittent RES, which is mainly the case of
solar and wind. Due to these issues, the mixed renewable
energy system must be well designed in order to better exploit
the potential of RES, hence fulfil specific requirements.
2.2.2. Electrolyser unit
The hydrogen generation process becomes more beneficial
if used in conjunction with electricity generated by the RES.
Different kinds of water electrolysers exist, with various level
of technological advancements. Water electrolysers can be
divided into two categories, alkaline and proton exchange
membrane (PEM) electrolysers. PEM electrolysis is a viable
alternative for generating hydrogen from RES. Despite its
higher cost compared, the PEM benefits of high efficiency
factor, higher life cycle (approximately ten years) [28], adapt-
ability with renewable energy systems [29]. In addition, a PEM
electrolyser can deliver hydrogen at high pressure, which will
in turn be attractive for the application where hydrogen needs
to be stored [30]. In this paper, a PEM electrolyser has been
adopted, this choice is justified by two main reasons namely,
the exploitation of wind/solar potentials and the need of high
pressure hydrogen to refuelling stations.
2.2.3. Fuel cell system
Among all kinds of fuel cell available, fuel cell Proton
exchange membrane PEM will be the most suitable for the
system configuration, this choice is dictated by the PEM elec-
trolysis. The operation of PEM fuel cell needs only hydrogen,
oxygen from the air and water to operate and do not require
corrosive fluids like some fuel cells. The driven goal of using
a fuel cell system at the centralized plant is to ensure another
option for the provision of electricity (in case of low power
generation from mixed renewable energy system).
2.2.4. Hydrogen storage system
A hydrogen storage system is available in each hydrogen
production plant. Since the plant is supposed to provide
hydrogen for a network of hydrogen refuelling stations, there
is a need to assume the availability of a large scale hydrogen
storage system. The latter assumes the aggregation of both
a compressor and a storage system. The hydrogen storage
device has to be able to meet the requirements of the
5363
hydrogen demand at the specific hydrogen refuelling station
in each time period (taking into consideration the delivery
schedule). Because lack of data in particular those related to
large scale hydrogen storage. In this paper, the hydrogen
storage systems have been sized by analysing the real amount
of gasoline delivered by several service stations which is
transformed in equivalent amount of hydrogen.
2.2.5. Hydrogen demand
Due to the absence to date of a widespread hydrogen market,
the computation of the hydrogen demand cannot be esti-
mated by accuracy. The operation of these markets will
depend initially on the hydrogen demand infrastructure
available, in particular, the number of hydrogen fuel cell
vehicles and hydrogen refuelling infrastructure. Owing to
many uncertainties, the planning of scenario can be consid-
ered as the only systematic tool that helps designing the
hydrogen supply chain. The configuration of the optimal
design will strongly depend on the scenario considered. In this
study, the estimation of hydrogen demand of the hydrogen
refuelling stations is based on the current supply of the petrol
products to the conventional petrol stations. The information
related to the petrol stations may be helpful in determining
the capacity and the consumption. Then, according to these
estimations, multiple scenarios are explored including
varying hydrogen fuel cell penetrations.
2.3. Modelling the mixed renewable energy system
As previously presented, the electrolysis hydrogen production
plant is mainly driven by power generated from the renewable
energy system. The overall hydrogen production plant is
linked to the network of GHRSs. It is composed by the
following subsystems: photovoltaic modules, wind turbines,
electrolyser unit, fuel cell unit, main hydrogen storage system.
Among these subsystems and from/to the external electrical
network, the power and hydrogen flows can be exchanged as
shown in Fig. 2. The energy produced by the renewable energy
system can be sent to the electrolyser to produce hydrogen
that will be stored in main tank, and/or used to feed electrical
energy of GHRSs. The quantity of hydrogen can be used for
different purposes: to satisfy the hydrogen fuel demand by the
refuelling stations, or/and used by the fuel cell to ensure the
electricity demand of the refuelling stations in the case of
a deficit of energy production, or/and delivered to the indus-
trial markets.
2.3.1. The wind turbine subsystem model
The energy produced by a wind turbine in a point of produc-
tion i is given by:
Zvp
T rA  
Etw;i ¼ cp hgb hg f vti vt3
i dv t ¼ 1; .; T  1; i ¼ 1; .; I
1000 2
vi
(1)
where f ðvti Þ
is the probability of occurrence of wind speed
nti [m/s] in a point of production i, A [m2] is the blade sweep
area, r is the air density [kg/m3], cp [-] the power coefficient,
hgb [-] the gearbox performance and hg [-] the generator
performance.
5364 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1

Fig. 2 e Energy and hydrogen flows exchanged among system components.

where vti corresponds to the wind speed at the wind

turbine hub height. It is assumed that the wind speed can be
predicted by some reliable meteorological models. In general,
the wind speed measurements are given at a height different
than the hub height of the wind turbine. So, the following
equation is used to evaluate the wind speed at the desired
height:
lnðHhub =z0 Þ
vti ¼ vtdata;i   t ¼ 1; .; T  1; i ¼ 1; .; I
ln Hdata;i =z0
where Hdata;i [m] is the height of the measurement, Hhub [m] is
the hub height and z0 is the surface roughness length andvtdata;i
is the wind speed at the height of the measurements.
2.3.2. The PV module subsystem model
Hybrid energy systems are often taken into account as a viable
approach to face the RES intermittent character. The use of
different RES (such as wind and solar) can enhance the
effectiveness to face the load energy demands. The electrical
energy generated from a photovoltaic module can be calcu-
lated using the following formula:
Etpv;i ¼ Spv hPV pf hpc Gti t ¼ 1; .; T  1; i ¼ 1; .; I
where Spv [m2] is the solar cell array area, hPV [-] is the
module reference efficiency, pf [-] is the packing factor, hpc [-]
is the power conditioning efficiency and Gti [kWh/m2] is the
forecasted hourly irradiation, that is predicted by some reli-
able meteorological model.
3. Optimization problem
The decision variables of the optimization problem are the
following.
Eeti , Enti and Esti [kWh] are the electrical energy components
in time period [t,t þ 1) respectively given by the mixed
renewable energy system i to electrolyser unit, electrical
network and to feed the electrical demand of the GHRSs.
(2) Efcti ,Etti and Esf ti;j [kWh] are the electrical energy compo-
nents in time period [t,t þ 1) respectively given by the fuel cell,
consumed by the main hydrogen storage tank in point
of production i and consumed by the GHRS j and its local
hydrogen tank.
t
Qhti , Qsf i;j , Qlti;j , Qmti [kg] represent in time interval [t,t þ 1)
respectively the amount of hydrogen produced by the elec-
trolyser unit at production node i, the amount of hydrogen
sent from the ith production node to the jth GHRS, the amount
of hydrogen delivered by the jth GHRS and the amount of
hydrogen sent to the industrial market from the ith produc-
tion node.
ri;j is a binary variable, equals to 1 if the ith production
point will be connected to the jth GHRS, 0 otherwise.
(3) mj is a binary variable, equals to 1 if the jth refuelling station
is selected, 0 otherwise.
The state variables of the optimization problem are:
Mti ,Msf ti;j are respectively the amount of hydrogen stored
in time period [t,t þ 1) in the main tank of the ith
production point and the amount of hydrogen stored in the
local tank of the jth GHRS at instant t.

Table 1 e Characteristics of a network of GHRSs.

GHRS Code Longitude Latitude Distance km Density of population Inha/km2

Refuelling station 1 PV 01363 44.4286 8.7587 52 7708

Refuelling station 2 PV 01398 44.2682 8.4344 22 1750
Refuelling station 3 PV 11348 44.3419 8.5438 33 1454
Refuelling station 4 PV 01334 44.1688 8.3413 27 2160
Refuelling station 5 PV 01377 44.0884 8.2072 45 1553
Refuelling station 6 PV 01371 44.3211 8.4654 21 6168
Refuelling station 7 PV 01351 44.1294 8.259 45 2277
Production point e 44.3984 8.2193 0 e
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
Fig. 3 e Wind speed and solar radiation data.
The following parameters are used in the model
- Nw: number of wind turbines;
- Npv: number of photovoltaic modules;
- HHVH2 : hydrogen higher heating value [kWh/kg]
- LHVH2 : hydrogen lower heating value [kWh/kg];
- LHVff : fossil fuel lower heating value [MJ/kg]
- h1 ; h2 : parameters of the electrolyser plant: the first one
represents the efficiency of the electrolysis system, while
the second one is an additional efficiency coefficient
included to take into account the (energy) losses in the
electrolyser.
- hff : efficiency of the fossil fuelled engine
- hH2 : efficiency of the H2 engine/fuel cell [-]
- hFC : fuel cell efficiency [-]
- Qff ti;j : demand of fossil fuel in time interval [t,t þ 1) at the jth
refuelling station [kg]
- ~ t : hydrogen market demand in time interval [t,t þ 1)
Qm
i
requested from the ith production point [kg]
- ~ t : hydrogen demand in time interval [t,t þ 1) of the jth
Ql
i;j
GHRS supplied by ith production point [kg]
- ~ t : electrical demand in time interval [t,t þ 1) of the jth
Ee
i;j
GHRS connected to the ith production point [kWh]
- Di;j : distance from the ith production point to the jth
GHRS [m]
Fig. 4 e Manufacturer power curve of the considered wind
turbine.
5365
Fig. 5 e Hydrogen demand of the network of GHRSs.
- Nj : population density around a circle of unit diameter of
distance of jth GHRS [inhabitant/km2]
- di : binary variable that is implemented to take into account
the exchange in power between the hydrogen production
plant and the electric network. It is equal to 1 when the
market penetration of hydrogen is less enough to allow
power exchange with the electrical network.
3.1. Objective function
The objective function to be minimized is the sum of seven
terms: two first terms reflect the satisfaction of different
hydrogen demands, namely the hydrogen fuel demands of the
GHRSs and the industrial market demand. Third term is
related to ensure the electrical demands of GHRSs. The fourth
tem is related to the energy sold to the electrical network to be
maximized/minimized according to the market penetration, it
means, for low hydrogen demand, the energy sold to the
electrical network needs to be maximized, otherwise it will be
minimized, this assuming that the renewable energy systems
are well sized. Fifth term to be maximized is related to the
amount of hydrogen stored in the main hydrogen tank. The
two last terms are added to select the suitable GHRSs to be
powered by the appropriate point of production.
Fig. 6 e Electrical energy demand of the network of GHRSs.
5366 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1

refuelling stations (Esti ), and a part sent directly to the elec-

trical network (Enti ).

Ettot;i ¼ Eeti þ Enti þ Esti t ¼ 1; .:; T  1; i ¼ 1; .:; I (6)

t
Esti þ Efci  Eesti  Etti ¼ 0 t ¼ 1; .:; T  1; i ¼ 1; .:; I (7)

The electrical energy Eesti [kWh] consumed in the time

interval [t,t þ 1) by the network of GHRSs will be equal to the
sum of electrical energy consumed by each one, and it is given
by:

P
J
t
Eesti ¼ Esf i;j t ¼ 1; .:; T  1; i ¼ 1; .:; I; j ¼ 1; .:; J (8)
j¼1
Fig. 7 e Energy produced by the mixed energy system.
t t
Esf i;j ¼ Ei;j þ fMsf i;j t ¼ 1; .:; T  1; i ¼ 1; .:; I; j ¼ 1; .:; J (9)
t
where Esf i;j [kWh] is the electrical energy consumed in time
J 
T1 X
X 2 XT1 X
I  2 interval by the jth GHRS that is composed by a constant term
Z¼
t ~t
Qli;j  Ql þ ~ t
a Qmti  Qm t
Ei;j and a variable one fMsf i;j , with f the unitary demand of
i;j i
t¼1 j¼1 t¼1 i¼1
electric energy for a unit of hydrogen stocked reservoir, and
T1 X
X I 
J X 2 X
T1 X
I X
I t
þg
t ~ t þz
Ei;j þ fMsf i;j  Ee di Enti  l MTi Msf i;j the amount of hydrogen available in the jth local tank of
i;j (4)
t¼1 j¼1 i¼1 t¼1 i¼1 i¼1 the GHRS in time interval [t,t þ 1).
J X
X I X
J The electrical energy Etti [kWh] consumed in time interval
þb ri;j D2i;j þ c mj N j [t,t þ 1) by the main hydrogen storage reservoir in the point of
j¼1 i¼1 j¼1
production i is given by:
a,g,z,l,b,c are weight factors.
Etti ¼ fMti t ¼ 1; .:; T  1; i ¼ 1; .:; I (10)

where Mti is the amount of hydrogen available in the main

storage reservoir in time interval [t,t þ 1).
3.2. Constraints
The electrical energy Efcti [kWh] delivered by the fuel cell in
time interval [t,t þ 1) is calculated by:
3.2.1. Flow conservation
The overall energy produced Ettot;i [kWh] by each mixed t t
Efci ¼ LHVH2 $Qfci $hfc t ¼ 1; .:; T  1; i ¼ 1; .:; I (11)
renewable energy system in time interval [t,t þ 1) is given by:
t
The amount of hydrogen [kg] delivered in time interval
Qhi
Ettot;i ¼ Nw $Etw;i þ Npv $Etpv;i t ¼ 1; .; T  1; i ¼ 1; .:; I (5)
[t,t þ 1) by the electrolyser plant is equal to:
The total energy Ettot;i [kWh] can be used for three different
purposes in the same time interval: direct for the hydrogen t h1 h2 Eeti
Qhi ¼ t ¼ 1; .:; T  1; i ¼ 1; .:; I (12)
production (Eeti ), satisfy the electricity demand of the HHVH2

Fig. 8 e Electrical energy sent to the electrolyser plant.

 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1 5367

tþ1 t t t
Msf i;j ¼ Msf i;j þ Qsf i;j  Qli;j t ¼ 1; .:; T1; i ¼ 1; .:; I; j ¼ 1; .:; J
(16)

M1i ¼ Mi;1 i ¼ 1; .:; I (17)

1
Msf i;j ¼ Msf i;j;1 i ¼ 1; .:; I; j ¼ 1; .:; J (18)

where Mi;1 , Msf i;j;1 [kg] are respectively the storage system
level at the initial time for each main hydrogen storage tank
and each local one.

3.2.3. Other constraints

3.2.3.1. Selection of the network of GHRSs

Constraint (19) is introduced in order to impose that if the link
Fig. 9 e Electrical energy sold to the network.
between the ith production point and the jth GHRS is estab-
lished. The jth GHRS must be selected following the safety
viewpoint.
The amount of hydrogen Qsti consumed in time interval
[t,t þ 1) by the network of GHRSs is equal to the sum of If dri;j ¼ 1/gj ¼ 1
hydrogen that is consumed by each one, and it is given by:

P
J 1 if the linkði; jÞis established
Qsti ¼
t
Qsf i;j t ¼ 1; .:; T  1; i ¼ 1; .:; I; j ¼ 1; .:; J ri;j ¼ (19)
(13) 0 otherwise
j¼1

~ t [kg] of the jth GHRS in time Eq. (19) can be also written as:
The hydrogen demand Ql i;j
interval [t,t þ 1) is given by: ri;j  Rgj  0 (20)
t
Qff i;j $LHVff $hff where R is a big number
~t ¼
Ql t ¼ 1; .:; T  1; i ¼ 1; .:; I; j ¼ 1; .:; J
i;j
LHVH2 $hH2
(14) 3.2.3.2. Hydrogen storage systems
The main hydrogen storage tanks are limited in upper and
3.2.2. Storage system state equations lower bands:
The state equations of the main hydrogen storage tank of each
Mi;min  Mti  Mi;max t ¼ 1; .:; T  1;i ¼ 1;.:; I (21)
point of production and the local hydrogen storage tank of
each GHRS in time interval [t,t þ 1) are given by: The local hydrogen storage tanks are limited in upper and
lower bands:

t t t
Mtþ1
i ¼ Mti þ Qhi  Qsti  Qmti  Qfci t ¼ 1; .:; T  1; i ¼ 1; .:; I Msf i;j;min  Msf i;j  Msf i;j;max t ¼ 1; .:; T1; i ¼ 1; .:; I; j ¼ 1; .:; J
(15) (22)

Fig. 10 e Electrical energy consumed by GHRSs.

5368 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1

3.2.3.3. Activation of transport

The hydrogen flows sent from a supply point i to the jth GHRS
are limited in upper and lower bands:

t
ri;j Qsf j;min  Qsf i;j  ri;j Qsf j;max t ¼ 1; .:; T1;i ¼ 1; .:; I; j ¼ 1; .:; J
(23)

3.2.3.4. Electrolyser
The energy flows sent to the electrolyser unit at each
production point i are limited in upper and lower bands:

dte Eei;min  Eeti  dte Eei;max t ¼ 1; .:; T  1; i ¼ 1; .:; I (24)


0 if Eeti < Eei;min
dte ¼ (25) Fig. 12 e Hydrogen storage level of GHRS1.
1 if Eeti  Eei;mim

3.2.3.5. Fuel cell

The hydrogen flows sent to the fuel cell are limited in upper the model to one point of production and seven existing
and lower bands: refuelling stations where related data are available. In this
t section we assume that this network of GHRSs is the suitable
dtfc Qfci;min  Qfci  dtfc Qfci;max t ¼ 1; .:; T  1; i ¼ 1; .:; I (26)
one for the considered point of production. Table 1 reports the
( characteristics of the network of GHRSs.
t
0 if Qfci < Qfci;min The developed approach supposes that the energy gener-
dte ¼ t (27)
1 if Qfci  Qfci;mim ated from the mixed renewable energy systems is used to
ensure electrical energy and hydrogen fuel demands of the
3.2.3.6. Electrical network GHRSs as well as a hydrogen market demand. The last is
The energy sold to the electrical network could be maximized assumed to be constant and equals to 10 kg in time period. The
or minimized according to the market penetration (MP). hydrogen market could reflect different users that may buy
 hydrogen from the production plant.
1 if MP < MPthreshold
di ¼ (28) The energy surplus is sold to the electrical network. It is
1 otherwise
assumed in this case study that the forecasted data from
which the optimization problem must be solved are exactly
equal to the historical data recorded in the Capo Vado site, and
4. Results and discussion which consist of the daily wind speed and solar radiation,
recorded at a height of 10 m. Fig. 3 shows the daily variation of
The problem is here solved using mathematical programming the wind speed and solar radiation of Capo Vado site, it can be
techniques through a commercial optimization package seen that the wind speeds range between 4.8 and 9.5 m/s.
(Lingo, www.lindosystems.org). In particular, the optimiza- Generally, the wind speed takes an average value equal
tion problem is solved for a case study in Capo Vado site approximately to 7.54 m/s at 10 m of height. Whereas, solar
(Savona district) for a time period of one month (April 2009), radiation ranges between 0.46 and 7.3 kWh/m2/day, with an
which consists of a network of seven real refuelling stations. average value of 4.8 kWh/m2/day.
Due to the lack of data regarding the fossil fuel and electrical In this paper, ten wind turbines (ENERCON E-53) (www.
energy consumptions as well as some territorial characteris- enercon.de), with the following geometric and technical
tics (population density, ..,), in addition to the meteorological characteristics have been considered: vc ¼ 2 [m/s], vr ¼ 13 [m/
measurements of wind speed and solar radiation, we applied

Fig. 11 e Hydrogen storage level of the main tank. Fig. 13 e Hydrogen storage level of GHRS2.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
Fig. 14 e Hydrogen storage level of GHRS3.
s], vf ¼ 25 [m/s], Pr ¼ 800 [kW], Hhub ¼ 75 [m]. The power curve
of the considered wind turbine supplied by the manufacturer
is shown in Fig. 4. For the photovoltaic module, it is assumed
that Spv ¼ 10000 [m2], hpv ¼ 0.11, hpc ¼ 0.86, and Pf ¼ 0.9.
In this paper, the hydrogen demands are assumed to be
known, and they are estimated using the real existing
demands of the petrol stations available in the province.
These real data are recorded from petrol stations of an Italian
Petrol Company. The demand of hydrogen used in this paper
is calculated according to several scenarios of market pene-
tration (5%, 15%, 25%). This hypothesis seems reasonable
since in a short term scenario, the demand of hydrogen
cannot be equal to the one of petrol products.
Fig. 5 displays the hydrogen demands of the network of the
GHRSs considered. It can be seen that the consumption
behaviours of the considered stations in April month are
different from each other, and it ranges between 54 and
3500 kg.
Fig. 6 shows the daily electrical energy needs of the GHRSs,
they range between 1.7 and 318 kWh. These demands are
related to the real consumption recorded at the petrol stations
available in the province of Savona. These demands are
almost coming from the electric network, but, since in this
study, the green aspect of the refuelling stations is considered,
this means that electricity consumption needs to be satisfied
by the energy produced by the renewable energy system.
These demands are mainly coming from lighting, pumps, and
others. The difference in electricity consumption among
refuelling stations is mainly due to the availability or not of
Fig. 15 e Hydrogen storage level of GHRS4.
5369
Fig. 16 e Hydrogen storage level of GHRS5.
special services (such car washing machines) and the demand
of hydrogen for vehicles at each time instant that is directly
linked to the pumps consumption. The electrical energy
demands used in this case study are calculated according to
the scenarios of the market penetration.
Fig. 7 displays the daily energy produced by the mixed
renewable energy system, we should notate that we do not
distinct between the energy produced by wind turbines and
photovoltaic modules, we assumed that is a clean energy
produced by a renewable energy system, since the economical
aspect is not taken into account in this study. The daily
produced energy ranges between 31 and 188 MWh. Fig. 8
shows the daily energy sent to the electrolyser plant accord-
ing to the three scenarios. Fig. 9 reports the daily electrical
energy surplus sold to the electrical network according to each
scenario. It can be seen that the most part of electrical energy
has been sold in scenarios where the market penetration
equals to 5% followed by 15%, while in the third scenario (25%)
the system did not exchange energy with the electrical
network. This is due to the low demand of hydrogen with
a high energy production in the two first scenarios and a high
hydrogen demand in the third one. Fig. 10 displays the daily
electrical energy supplied to the network of GHRSs to satisfy
their electrical energy needs including the energy consumed
by the local hydrogen storage tanks.
Fig. 11 shows the trend of the energy stored along the
month in the main hydrogen storage tank according to the
scenarios of the market penetration. It appears that the higher
amount of hydrogen stored is reached for a market
Fig. 17 e Hydrogen storage level of GHRS6.
5370 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
Fig. 18 e Hydrogen storage level of GHRS7.
penetration equals to 5% which is due to the low hydrogen
demand. Figs. 12e18 show the level of hydrogen in each local
storage system according to the three scenarios of the market
penetration. It is observed that each GHRS exhibits a different
storage tendency, which is mainly due to the difference in
hydrogen demands. In general, all demands are always
satisfied in each time period and for each scenario.
5. Conclusion
An optimization model of a network of GHRSs is presented.
The network of GHRSs is powered completely by mixed
renewable energy systems. The main decision within the
problem can be categorized into two types: first one is related
to the optimal selection of the network of GHRSs following
roads network distance and population density criteria, while
the second is related to the control and management of the
hydrogen and electric flows that are sent to the GHRSs.
The optimization model has been simplified, and then
applied on one month basis to a case study in the province of
Savona, Italy. Optimal results are reported taking into account
the presence of an additional hydrogen industrial market and
a connection with the electrical network. Results show that
the demands are satisfied for each time period and for all
market penetration scenarios. Future developments of the
present work regard the possibility of introducing stochastic
issues both in demand and resource predictions. In this case,
in order to reduce the overall complexity of the problem,
dynamic programming could be used. Then, the model could
be detailed from the electrical and technological point of view.
Moreover, an objective function based on hydrogen costs
could also be adopted. Finally, the inclusion of the developed
dynamic decision model in a decision support system that can
allow the user to interact with the mathematical models could
be helpful.
Acknowledgements
Authors wish to thank ARPAL (Agenzia Regionale per la Pro-
tezione dell’Ambiente Ligure, http://www.arpal.org/) and the
Italian petrol company ENI that have kindly collected and
made available the important historical dataset at the basis of
this work.
references
[1] Li Y, Xue B, He X. Catalytic synthesis of ethylbenzene by
alkylation of benzene with diethyl carbonate over HZSM-5.
Catal Commun 2009;10:702.
[2] Demirbas A. Present and future transportation fuels. Energy
Sources Part A 2008;30:1473.
[3] Clarke RE, Giddey S, Ciacchi FT, Badwal SPS, Paul B,
Andrews J. Direct coupling of an electrolyser to a solar PV
system for generating hydrogen. Int J Hydrogen Energy 2009;
34:2531e42.
[4] Campen A, Mondal K, Wiltowski T. Separation of hydrogen
from syngas using a regenerative system. Int J Hydrogen
Energy 2008;33:332.
[5] Ingason HT, Ingolfsson HP, Jensson P. Optimizing site
selection for hydrogen production in Iceland. Int J Hydrogen
Energy 2008;33:3632e43.
[6] Jørgensen C, Ropenus S. Production price of hydrogen from
grid connected electrolysis in a power market with high wind
penetration. Int J Hydrogen Energy 2008;33:5335e44.
[7] Sherif S, Barbir F, Veziroglu T. Wind energy and the hydrogen
economy e review of the technology. Solar Energy 2005;18:
647e60.
[8] Bokris J, Veziroglu T. Estimates of the price of hydrogen as
a medium for wind and solar sources. Int J Hydrogen Energy
2007;32(12):1605e10.
[9] Greiner C, Korpås M, Holen AA. Norwegian case study on the
production of hydrogen from wind power. Int J Hydrogen
Energy 2008;32:1500e7.
[10] Segura I, Pérez-Navarro A, Sánchez C, Ibáñez F, Payá J,
Bernal E. Technical requirements for economical viability of
electricity generation in stabilized wind parks. Int J Hydrogen
Energy 2007;32:3811e9.
[11] Prince-Richard S, Whale1 M, Djilali N. A techno-economic
analysis of decentralized electrolytic hydrogen production
for fuel cell vehicles. Int J Hydrogen Energy 2005;30:1159e79.
[12] Hugo A, Rutter P, Pistikopoulos S, Amorelli A, Zoiac G.
Hydrogen infrastructure strategic planning using multi-
objective optimization. Int J Hydrogen Energy 2005;30:
1523e34.
[13] Thomas CE, Kuhn1 F, James BD, Lomax FD, Baum GN.
Affordable hydrogen supply pathways for fuel cell vehicles.
Int J Hydrogen Energy 1998;23:507e16.
[14] Doll C, Wietschel M. Externalities of the transport sector and
the role of hydrogen in a sustainable transport vision. Energy
Policy 2008;36:4069e78.
[15] Almansoori A, Shah N. Design and operation of a future
hydrogen supply chain: multi-period model. Int J Hydrogen
Energy; 2009:347883e97.
[16] Almansoori A, Shah N. Design and operation of future
hydrogen supply chain Snapshot Model. Chem Eng Res Des
2006;84:423e38.
[17] Kim J, Lee Y, Moon II. Optimization of a hydrogen supply
chain under demand uncertainty. Int J Hydrogen Energy
2008;33:4715e29.
[18] Kuby M, Schultzc LLR, Xie Z, Kim JG, Lim S. Optimization of
hydrogen stations in Florida using the flow-refueling
location model. Int J Hydrogen Energy 2009;34:6045e64.
[19] Bersani C, Minciardi R, Sacile R, Trasforini E. Network
planning of fuelling service stations in a near-term
competitive scenario of the hydrogen economy. Socio-Econ
Plan Sci 2009;43:55e71.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 5 3 6 0 e5 3 7 1
[20] Nicholas MA, Handy SL, Sperling D. Using geographic
information systems to evaluate siting and networks of
hydrogen stations. Transp Res Rec 2004;1880:126e34.
[21] Parker N, Fan Y, Ogden J. From waste to hydrogen: an
optimal design of energy production and distribution
network. Transp Res Part E 2009;46:534e45.
[22] Joffe D, Hart D, Bauen A. Modeling of hydrogen infrastructure
for vehicle refueling in London. J Power Sources 2004;13:
13e22.
[23] Dagdougui H, Minciardi R, Ouammi A, Robba M, Sacile R. A
dynamic decision model for the real time control of hybrid
renewable energy production systems. IEEE Syst J 2010;4(3).
[24] Martin KB, Grasman SE. An assessment of wind-hydrogen
systems for light duty vehicles. Int J Hydrogen Energy 2009;
34:6581e8.
5371
[25] Systèmes solaires le journal de l’éolien n 8 2011. Baromètre
éolien - eurobserv’er -Février 2011.
[26] Honnery D, Moriarty P. Estimating global hydrogen
production from wind. Int J Hydrogen Energy 2009;34:727e36.
[27] López R, Bernal-Agust?n JL, Mendoza F. Design and
economical analysis of hybrid PVewind systems connected
to the grid for the intermittent production of hydrogen.
Energy Policy 2009;37:3082e95.
[28] Benchrifa R, Bennouna A, Zejli D. Role de l’hydrogène dans le
stockage de l’électricité a base des Energies renouvelables.
2IWH 2007; 27e29 October 2007. Ghardaı̈a e Algeria.
[29] Kruse B, Grinna S, Buch C. Hydrogen Status og muligheter
Bellona rapport nr, vol. 6; 2002.
[30] Barbir Frano. PEM electrolysis for production of hydrogen
from RES. Solar Energy 2005;78:661e9.