Experimental evaluation and modelling of a PEM fuel cell in the context of a hydrogen based trans seasonal energy storage system

(1)

Experimental evaluation and modelling of a PEM

fuel cell in the context of a hydrogen based trans

seasonal energy storage system

Jonas Houf

Student number: 01504634

Supervisors: prof. dr. ir. Michel De Paepe, prof. dr. ir. Jan Desmet

Counsellors: ir Katarina Simić, ir. Wim Beyne, ir. Koen Bruynoghe

(Callens & Bruynoghe)

Master's dissertation submitted in order to obtain the academic degree of Master of Science in Electromechanical Engineering

(2)

(3)

Experimental evaluation and modelling of a PEM

fuel cell in the context of a hydrogen based trans

seasonal energy storage system

Jonas Houf

Student number: 01504634

Supervisors: prof. dr. ir. Michel De Paepe, prof. dr. ir. Jan Desmet

Counsellors: ir Katarina Simić, ir. Wim Beyne, ir. Koen Bruynoghe

(Callens & Bruynoghe)

Master's dissertation submitted in order to obtain the academic degree of Master of Science in Electromechanical Engineering

(4)

I am honored and proud to present you my master dissertation, a work in which all my ac-quired knowledge and skills of the past five years of engineering studies come together. The result is a combination of dedicated work and extensive support from a group of people to whom I want to express my sincere gratitude.

First of all, I would like to thank my promotor for his belief and assistance in my thesis. I vividly remember the constructive discussions we had on the future of a sustainable Belgian society in may 2019 leading to the creation of the thesis’ subject. Thank you for your vote of confidence.

Secondly, I would like to express my appreciation to my counsellors, Katarina Simic and Wim Beyne for their great support, availability and vision. I could not have wished for any better encouraging team. Besides the thesis meetings, I enjoyed the coffee breaks and talks on future plans. In addition, I would like to thank the entire STFS staff and technician Frederik Martens in particular. Building an experimental setup from scratch is a challenge. I appreciated all the technical help from you throughout this endeavour. Also, thank you to Jens Peeters for making the intense testing days at the lab more enjoyable.

I am grateful to the cooperating companies in this thesis: Callens & Bruynoghe (fuel cell) and Modelon (modelling software), for letting me participate in their project and help with the modelling.

And lastly, I would like to thank my friends and family for the unconditional support through-out my studies. Studying electromechanical engineering is demanding in many aspects. It also gives the opportunity to meet many inspiring and great people. The combination of hard work and carefree, half-seas over moments has created phenomenal memories and made it possible to keep on going.

Jonas Houf

(5)

v

The author gives permission to make this master dissertation available for consultation and to copy parts of this master dissertation for personal use. In all cases of other use, the copyright terms have to be respected, in particular with regard to the obligation to state explicitly the source when quoting results from this master dissertation.

(6)

Jonas Houf

Master’s dissertation submitted in order to obtain the academic degree of Master of Science in Electromechanical Engineering

Promotors: prof. dr. ir. Michel De Paepe, prof. dr. ir. Jan Desmet

Supervisors: ir. Katarina Simi´c, ir. Wim Beyne, ir. Koen Bruynoghe (C&B) Department of Electrical Energy, Metals, Mechanical Construction & Systems Sustainable Thermo-Fluid Energy Systems (STFES)

Chair: Prof. dr. ir. Michel De Paepe Faculty of Engineering and Architecture Ghent University

Academic year 2019–2020 Summary

With the purpose of making a residential building grid independent and able to cope with intermittent energy sources, hydrogen is used to store energy on the long term. Fuel cells are the the current most efficient means of converting stored hydrogen back to useful energy. In that context, a commercial proton exchange membrane (PEM) fuel cell is experimentally observed and thereafter modelled in the simulation environment Dymola.

The thesis starts with a small essay to place fuel cells and hydrogen in the broader context of the renewable energy transition. Chapter 2 serves as a base, providing general knowledge on fuel cells and hydrogen. The PEM fuel cell, on which this thesis focuses, is explained in more detail. The theoretical performance is explained as well as some case studies.

In chapter 3, the test setup is extensively described. More details on the used components and in particular the PEM fuel cell can be found in this chapter. The test results and obser-vations including measured temperatures, output performance and efficiencies are presented and examined in chapter 4. By extension a case example of a hydrogen based energy storage system is elaborated.

Thereafter, a model made in Dymola, using modelling libraries of Modelon, is described in chapter 5. The modifications on the component model templates and the whole build up is explained. In chapter 6, the simulation results are discussed and in the end, the use of Dymola for the modelling of commercial fuel cells with an internally adjusted power output is assessed. A final conclusion and reflection of the entire work is made in chapter 7.

Keywords

Hydrogen, fuel cells, Modelica, Zero-emission, energy storage system

(7)

Experimental evaluation and modelling of a PEM

fuel cell in the context of a hydrogen based trans

seasonal storage system

Jonas Houf

Supervisor(s): prof. dr. ir. Michel De Paepe, prof. dr. ir. Jan Desmet Counsellor(s): ir. Katarina Simi´c, ir. Wim Beyne, ir. Koen Bruynoghe (C&B)

Abstract—With the purpose of making a residential building grid inde-pendent and able to cope with intermittent energy sources, hydrogen is used to store energy on the long term. Fuel cells are the the current most efficient means of converting stored hydrogen back to useful energy. In that context, a commercially available proton exchange membrane (PEM) fuel cell is ex-perimentally tested. A test setup, containing the fuel cell, a buffer battery and a set of discrete loads was built. The fuel cell’s performance and effi-ciency are determined for a wide range of resistive loads. In addition, the temperature on the exhaust of the fuel cell is measured to get an idea of the waste heat energy. The use of the particular fuel cell in a hydrogen based energy storage system is eventually estimated with an representative exam-ple case. The fuel cell is subsequentially modelled in Dymola, using the Modelon Fuel Cell and Base Libraries. An attempt is made to mimic the experimentally obtained performance characteristics based on the experi-mentally determined input values and confined knowledge of the design of the given fuel cell device. In the end the use of Dymola for the modelling of commercial fuel cells with an internally adjusted power output is assessed. Keywords— Hydrogen, fuel cells, Modelica, Zero-emission, energy stor-age system

I. INTRODUCTION

A. General scope

As the world’s renewable energy transition unfolds, the potential of hydrogen in playing a major role in both small as large scale applications has grown. Hydrogen (H2) gas can serve as an

en-ergy carrier and can be produced by means of renewable enen-ergy sources. Many hydrogen based initiatives around the world have been introduced from automotive applications to large scale and long term energy storage systems. Using hydrogen as fuel to generate both electricity and heat in residential buildings can lower the related 12,1% share in European greenhouse gas emis-sions [1]. The application that will be evaluated in this work is a hydrogen based energy storage system.

B. Hydrogen based energy storage system

With the aim to handle the seasonal intermittency of solar gen-erated power, a hydrogen based energy storage system was con-ceptualized by Callens & Bruynoghe (C&B) [2]. Excess so-lar energy during summer months is used to produce hydrogen out of water through electrolysis. The H2gas is subsequentally

compressed to store in an efficient way in a pressurized storage vessel. In the winter, when the availability of solar energy is limited, the H2gas is converted back into electricity by means

of a PEM fuel cell. The total efficiency of the system equals the product of its individual components. The components and their corresponding estimated efficiencies are: electrolyser (ηelsr =

60%, [3]), compressor (ηcompr= 80%, [4]), storage tank (ηtank

= 100%) and the PEM fuel cell (ηF C = 45%, [5]) resulting in

a theoretical system efficiency of ηsystem= 21,6% compared to

a battery storage system for which the total efficiency can be higher.

There are two main ways to convert hydrogen back to useful energy: a hydrogen engine and a fuel cell. Fuel cells clearly have higher efficiencies (both thermal and electrical), in partic-ular the Alkaline Fuel Cell [6]. Therefore they are more suitable for CHP and hydrogen based energy storage systems.

Fig. 1. Concept Callens & Bruynoghe (CB) [7]

C. Fuel cells

The fuel cell (FC) is one of the key pieces to enable a hydrogen economy and hydrogen based energy storage systems. Many types of fuel cells exist, including Alkaline, Solide Oxide (SO) and Proton Exchange Membrane (PEM) fuel cells among others. The fuel cell used in the the experimental evaluation is a PEM FC and uses H2and O2to generate power, heat and H2O. Figure

2 illustrates the working principle of a PEMFC. H2is supplied

on one side of the membrane and oxidizes forming H+_protons,

which migrate through the membrane and electrons bypassing the membrane via an external circuit. At the cathode side, am-bient air is supplied containing O2that reacts with the electrons

and migrated H+_{protons forming water vapour. The global}

re-action that happens in the fuel cell is given in equation 1, [8]. H2+

1

(8)

Fig. 2. Working principle PEM FC

II. EXPERIMENTAL EVALUATION

A. Test methods

A fuel cell’s performance can be represented by its voltage ver-sus current density [A/m2_{] curve. Different internal effects,}

each dominating in a specific region, result in voltage losses that lower the theoretical potential of 1,23V for one single cell [9]. An entire fuel cell stack consists of many single cells, placed in series. Figure 3 shows, in addition to the voltage loss due to non-standard pressure and temperature conditions, three main losses: activation (due to initial overpotential, low current den-sity values), ohmic (due to conductor resistance, medium cur-rent density values) and concentration losses (due to rapid gas consumption near the electrode, high current density values). The lowest curve depicts the real voltage evolution in function of the current density. The theoretical output voltage of a fuel cell can be described as in equation 2, [9-10].

E = E0−R_n· T · F·ln " 1 pH2· p 1/2 O2 #

−∆Vact−∆VOhm−∆VConc

(2) The commercially available fuel cell, used in the test, is de-signed to keep its output voltage constant so that by varying the load, a wide range of current values can be attained. Therefore, the current is related to the applied load through Ohm’s law in the stable region.

B. Description of the test setup

The test setup comprises three major components: the fuel cell, load and battery. They are connected by two main circuits: a chemical hydrogen circuit and an electric circuit. An overview can be seen on figure 4 where the chemical circuit is colored in green and the electrical circuit in blue. The measured pa-rameters are displayed in the boxes pointing towards where the measurement is taken. The pressure, temperature and mass flow of the incoming hydrogen is measured. The fuel cell is fed by 99,9% pure H2from a pressurized vessel that passes through a

pressure reducer and O2from the ambient air. The purged out

H2, formed H2O and cooling air are blended together before

the exhaust. There is no (unused) H2 recirculation back to the

anode hydrogen intake. The hydrogen properties and the cur-rent through the load are logged in a Labview program while the voltage and current output of the fuel cell are measured by a separate program of the manufacturer, called System Monitor. The fuel cell is a FCM-802 of Intelligent Energy with properties given in table I [11]. It requires a 48V battery to overcome the startup energy barrier and initiate the chemical reaction. Four car batteries of 12V are connected in series form this 48V en-ergy buffer in the setup. An optimization cycle is performed by the fuel cell to optimize its efficiency. During this roughly 12 second cycle, which occurs approximately 15 times per hour ac-cording to the product manual, the output current drops to zero for a couple of seconds after which it steadily resumes opera-tion. To cover this power suspension, the battery takes over the power supply to the load. The load consists of 13 parallel con-nected 4,7 Ω resistances, a 22 Ω and 47 Ω resistance that all can be switched on and off separately to form a specific equivalent load. The main testing range of the load was between 4,7Ω and 0,67 Ω (7 standard parallel resistances). For the V-I curve, the range was extended to 47 Ω. Although the transition phenom-ena die out very fast, the loads are applied for a duration of 10 minutes to exclude any accidental variation in output.

TABLE I PROPERTIESFCM-802 (IE)

Property Value

Type IE FCM-802

Rated net power 2,5 kW @ 48V

Rated current 52A @ 48V

Fuel >99,9% pure gaseous H2

Fuel consumption Less than 70g per kWh

(9)

Fig. 4. Overview of the test setup

C. Results and discussion C.1 V-I curve

Three series of increasing resistances were applied to the sys-tem represented in the graphs on figure 5 (close up) and 6 (full range). The current was chosen on the x-axis instead of the cur-rent density because it is more meaningful in analysing the ac-tual performance. It is clear that the output voltage is kept con-stant for the larger part of the applied range of resistances. How-ever, a linear decline can be observed as a result of the ohmic losses until the concentration losses take over and bend off the curve to a steeper decline. A steeper decline in output voltage can be noticed for high current values, indicating the concentra-tion losses. No higher resistances were available to reduce the current towards zero and observe the third loss phenomenon.

Fig. 5. Close up range experimental V-I plot

Fig. 6. Full range experimental V-I plot

Although the output voltage linearly decreases with increasing current, it does not lessens as much as expected from figure 3. An internal power electronic system keeps the voltage nearly constant throughout a large part of the tested range to make it a

commercially viable device and useful for application where a steady power source is required. Because no higher loads could be tested, the current could not be pushed to zero in order to complete the characteristic.

The ranges of the most important measurements are given in table II, below.

TABLE II

OVERVIEW OF THE CONCEPT EVALUATION BASED ON EXAMPLE DATA

Parameter Min Max Unit

Current 7,7 58,0 A

Power 400 2770 W

Mass flow H2 0,94· 10−5 9,89 · 10−5 kg/s

C.2 Temperature

The temperature of the supplied H2 clearly elevated with

in-creasing load or power demand. More power requires more hy-drogen supply, leading to an increased expansion rate. As a con-sequence of the Joule-Thomson effect, the temperature in the H2

supply line raised significantly but did not exceed 40,5◦_C

dur-ing testdur-ing. On figure 7, the temperatures, corresponddur-ing to the different loads during testing are presented. The temperature of

Fig. 7. Hydrogen temperature for different loads

the exhaust fluctuated approximately around 25◦_{C. Three small}

fans, taking energy from the fuel cell, blow off the exhaust into the atmosphere. On the basis of the limited exhaust temperature and small pressure drop on the exhaust flow, no heat exchanger can be justified to recover byproduct, heat, of the fuel cell. C.3 Efficiency

In the light of a hydrogen based energy storage system, where the stored amount of hydrogen is finite, it is of essential impor-tance to know the efficiency of the fuel cell for different load regimes in order to know the most efficient operation. As indi-cated on figure 4, the pressure, temperature and volume flow are logged on the hydrogen supply line. Based on the state of the gas and the ideal gas law, the molar flow ˙N can be computed and eventually the efficiency, expressed in equation 3. ∆H is the higher heating value (HHV) of H2.

η = Wel WH2

= _˙V · I

(10)

a PEM fuel cell. A higher efficiency would be possible if a H2

recirculation is installed but it was decided by the manufacturer that this adaptation was less economically interesting due to the extra cost of components. The heat generated by the fuel cell is also left unused. A heat exchanger cannot be installed nor can the heated exhaust air be used to heat up spaces as it contains fractions of (purged out) hydrogen. Therefore the FC needs to be installed in well ventilated rooms. Utilizing the generated heat, by, for example, cooling the FC with water could elevate the efficiency.

Fig. 8. Fuel cell efficiency [%]

D. Concept evaluation

In order to evaluate the concept of a hydrogen based energy storage system using the FCM-802, example data from a case of C&B was used. The accumulated shortage of energy that needs be generated through the fuel cell to meet the demand, is 2331,0 kWh or 8391,6 MJ. Hydrogen has a power density of 142 MJ/kg ([12]) and, taking the (highest experimentally recorded) fuel cell efficiency into account, 25820,3 MJ of energy needs to be contained in the stored hydrogen. The total amount of H2 gas needed, in this case, is thus 181,8 kg or 2132,9 Nm3.

The excess solar and wind energy during the summer months (March-October) is used to produce hydrogen. Due to a low system efficiency (η ≈ 20%), the hydrogen based energy stor-age system falls short of 1138,0 Nm3_H

2gas to meet the energy

demand during winter. In table III, an overview of the case re-sults is given. The very low, pure electrical, efficiency of the fuel cell has by far the largest impact on the overall system effi-ciency, followed by the electrolyser efficiency which is 55% for the one that would be used in the concept.

TABLE III

OVERVIEW OF THE CONCEPT EVALUATION BASED ON EXAMPLE DATA

Parameter Value Unit

Total energy demand 13 000 kWh

Supply excess during summer 6069,0 kWh Supply shortage during winter 2331,0 kWh

Amount of producable H2 994,9 Nm3

Amount of H2needed 2132,9 Nm3

H2balance -1138,0 Nm3

[13]. Despite their higher efficiency, they are not suitable for trans seasonal energy storage as a results of their limited capac-ity (between 9,5 kWh and 18,5kWh for commercial household battery packs). They could only be useful to manage day and night intermittency.

Fig. 9. System efficiency overview

III. MODELLING

The fuel cell was modelled in Dymola, using the object oriented programming language Modelica. Two libraries were provided by Modelon, a leading Swedish company in system modelling and simulation software [14]. The libraries were Modelon base (V3.5) and Fuel cell (V1.11) library and contained templates of key components in the system that were modified to the experi-mental set up.

A. Model explanation A.1 General structure

As explained in section II-B, the fuel cell and battery form a hy-brid system together with the variable load. The battery provides start up power for the fuel cell, covers suspension of power gen-eration during the optimization cycle and is compensated by the fuel cell for any loss of charge. However, this power exchange is very limited and is, for the sake of simplicity, neglected in the model. The fuel cell is, therefore modelled separately. The power demand will be determined externally by an energy man-agement system based on the demand of the building and will be inserted as a test function in the model. In figure 10, an overview of the model is presented.

Fig. 10. Overview model

A.2 Components

The most important adaptations to the existing PEMFC base model, included in the Modelon libraries will be explained be-low.

(11)

Membrane parameters

Two membrane templates of the Modelon Fuel Cell Library, the Simplified membrane, which is only suited for low current simulations and the, more accurate, ElectroChemical membrane were used. During the experimental evaluation, one could ob-serve that the effect of the ohmic voltage losses was heavily re-duced. Instead of the fairly declining voltage that was expected from figure 3, a roughly constant output voltage was generated, as can be seen on figure 6. This rate of decline is dictated in the model by the parameter ASR (Area Specific Resistance). Re-ducing the value of the ASR could flatten out the voltage-current characteristic.

Fuel cell

A substack template model formed the base of the fuel cell. In this submodel, the membrane type and internal flow loss models among others are defined. The substack model is subsequently integrated in the Coolstack model, which includes a cooling pipe, a positive and negative electric output pin, a replaceable substack model and feed and drain connectors for the anode gas, cathode gas and cooling gas. The Coolstack is by default cooled by water. The Distributed channel for two phase media compo-nent in the Coolstack model represents the cooling mechanism. In the FCM-802 fuel cell, however, the cooling medium is ambi-ent air, which only occurs in one phase (no condensation). The cooling air comes from the outside. The class of the cooling pipe was therefore altered to a Distributed channel, which is a predefined cooling pipe class that handles single phase fluids in-stead of two phase fluids.

Gas Sources

Two gas sources are present in the system: 99,9% pure H2kept

in a pressurized vessel, reduced to around 1,6 bar before arriv-ing at the anode side and ambient air used for both coolarriv-ing and formation of H2O at the cathode side. These two gas inputs

can be represented by a GasPressureBoundary where their re-spective medium type, pressure, temperature and mass fraction are defined. Flow resistances between the pressure source and the inlet of the fuel cell were installed which induce a pressure difference and by consequence a mass flow. After tuning the model parameters of the flow resistances, the simulated mass flows were in the order of the mass flows in the experiment. At the exhaust, corresponding medium sinks are placed.

Load

The load is modelled in the form of an electrical current that is dictated to the fuel cell’s output. By using the component Ram-pCurrent, a linearly increasing current can be induced from 5A to 60, over a period of 600 seconds, to reach the full range of currents observed in the experiment. The FC’s output voltage will evolve according to the applied current.

Input values

The parameters and input values in the model were partially based on the experimental observations. Since there was lim-ited knowledge on the design parameters of the FCM-802, a lot of estimations needed to be made and default values suggested by Modelon were used in order to complete the required model

input.

B. Model evaluation

On figure 11, the output voltage of Simplified membrane for dif-ferent ASR values is presented. The lower (red) curve represents the voltage output for an ASR value of 0,183e-4 Ω · m2_{, the}

co-inciding green and blue curve are the voltage output for ASR value 0,08 Ω · m2_{and 0,08e-10 Ω · m}2_{, respectively. The curves}

are initially coincident but diverge in the ohmic region as a result of the distinct ASR that is used and dictates the ohmic voltage losses. Despite decreasing the ASR value, the curve does not get flatter beyond a certain value. This artificial correction of the curve does not fully have the desired effect.

Fig. 11. Voltage simplified membrane for different ASR

The simulated polarization curve, shown in figure 12 deviates from the observed curve during the experiment. The more ac-curate ElectroChemical membrane was used for this simula-tion. Despite that the shape of the characteristics does resemble the theoretical characteristic, it fails to match the experimental curve. The output voltage values do not fully correspond as well as the course of the curve. In the experiments, output voltage values between 51 and 52 V were measured while the simulated stack voltage begins around 47V and quickly drops to lower val-ues, ending around 44V.

Fig. 12. Polarization curve Coolstack with ElectroChemical membrane

The simulated output power, given on figure 13 of the fuel cell is more conform to experiment. In the experiment, power out-puts between 400 and 2770 W were recorded. Although the sim-ulated power output is approximately the same as in the exper-iment, the experimental power stagnated for high range loads. In the simulation, the curve keeps linearly growing and does not bend off.

(12)

Fig. 13. Simulation power output for the experimental load

IV. CONCLUSIONS

A hydrogen based trans seasonal storage system, conceptual-ized by the company Callens & Bruynoghe was evaluated. The PEM fuel cell, the key element in the system, was experimen-tally evaluated and subsequently modelled in Dymola.

A new experimental setup was built to map the performance of an commercial fuel cell. The V-I curve could be experimentally constructed to a certain extent due to a limitation testing loads. Throughout a large part of the testing range, the output voltage remained roughly constant as a result of internal power electron-ics. The generated power ranged between 400 and 2770W. Temperatures did not exceed critical values. The maximum recorded H2supply temperature was 40,5◦C while the exhaust

temperature fluctuated closely around 25◦_{C. No waste heat, a}

byproduct of the chemical reaction, can be recovered in this par-ticular fuel cell.

Efficiencies were ranged between 27,5 to 32,5%, which is lower than expected from the product manual (36,2%). The efficiency could be elevated if H2recirculation (unused H2goes back to the

inlet) was used or if the generated heat was utilized for thermal purposes through a different fuel cell stack cooling system. The efficiency of the entire hydrogen based energy storage system was confined to around 20% due to the many subcomponents. The electrolyser and the fuel cell had the biggest effect on the global efficiency. Battery energy storage systems perform better in terms of efficiency but do not have the capacity to store large trans seasonal amounts of energy.

The fuel cell was modelled in Dymola, based on the libraries of Modelon. Various adaptations were applied starting from the provided templates in the Modelon libraries. Input parameters were based on the experimental observations but needed to be, for the most part, estimated. The strategy to imitate the effect the internal power electronics did not have the desired effect of entirely flattening out the curve. Although the voltage-current density characteristic resembled the theoretical curve, it failed to represent the curve observed in the experiment. The power

libraries of Modelon form a great tool to design fuel cells and fuel cell system, it is less suited for modelling the behaviour of existing, commercial fuel cells with an adjusted output. Their behaviour clearly deviates from the theoretical output evolution, described by the electrochemical equations, as a result of inter-nal power electronics and interinter-nal optimization program.

ACKNOWLEDGMENTS

I would like to pay my special regards to the company C&B for providing a PEMFC to perform the tests. Also, I wish to express my gratitude towards Modelon for giving me access to their de-tailed and valuable libraries that allowed me to model the fuel cell.

In addition, I would like to thank my supervisor prof. dr. ir. Michel De Paepe and my counsellors ir. Katarina Simi´c and ir. Wim Beyne for their support and advice during my master dissertation.

REFERENCES

[1] European Commission, EU energy in figures, European Union, 2019. [2] Callens & Bruynoghe, C&B website , https://www.callensbruynoghe.eu/. [3] J. D. Holladay et al., An overview of hydrogen production technologie,

2008.

[4] S. Verhelst, Displacement pumps and compressors, Ghent University, 2019. [5] Z. Salameh, Renewable energy system design, Academic Press, 2014. [6] M. De Paepe et al., Micro-CHP systems for residential applications, In:

Energy conversion and management, 2006.

[7] Callens & Bruynoghe, Personal Energy: concept, C&B, 2019.

[8] Sharaf, Omar Z and Orhan, Mehmet F, An overview of fuel cell technology: Fundamentals and applications, Renewable and sustainable energy reviews, 2014.

[9] F. Barbir, PEM Fuel Cells, second edition, Academic Press, 2013. [10] O’hayre, Ryan and Cha, Suk-Won and Colella, Whitney and Prinz, Fritz

B, Fuel cell fundamentals, John Wiley & Sons, 2016.

[11] Intelligent Energy, The Intelligent Energy products, https://www.intelligent-energy.com/our-products/product-information/. [12] S. Verhelst et al., Hydrogen-fueled internal combustion engines, In:

Progress in Energy and Combustion Science, 2009.

[13] K. Li and K. J. Tseng, Energy efficiency of lithium-ion battery used as energy storage devices in micro-grid, IEEE, 2015.

(13)

List Of Abbreviations

AFC Alkaline Fuel Cell ASR Area Specific Resistance

FC Fuel Cell

FCV Fuel Cell Vehicle

ICE Internal Combustion Engines HHV Higher Heating Value

MCFC Molten Carbonate Fuel Cell PAFC Phosphoric Acid Fuel Cell

PEM Proton Exchange Membrane

PV Photo-Voltaic

SOFC Solid Oxide Fuel Cell

(20)

α Transfer coefficient A/cm2

ac Catalyst specific area m2/kg

C Concentration

∆ Change = final minus initial

F Faraday’s constant s_mol·A

η Efficiency

γ Pressure coefficient

G Gibbs free energy J

H Total enthalphy J

h Specific enthalpy J/kg

I Current A

i0 Exchange current A

j Current density A/m2

j0 Exchange current density A/m2

Lc The loading of the catalyst g/m2

˙

m Mass flow rate kg/s

˙

N Molar flow rate mol/s

n Number of electrons per molecule

p Pressure bar

R Universal gas constant J/ mol· K

ρ Density kg/m3 σ Conductivity S/m S Enthaly J/K s Specific entropy J/K· kg T Temperature °C/K ˙

V Volume flow rate m3_/s

V Voltage V

W Power W

(21)

Chapter 1

Introduction to the subject

1.1 The importance of energy

What is energy? If we look solely at the physical unit in which it is expressed: Joule, we can state that it is the ability to do work. Joule equals newton times meter, so it is defined as exerting a force over a distance. Lifting a bucket of water or climbing a hill are examples of physical work. Watt, Joule per second, is the rate at which energy is consumed. Although energy can be quantified, predicted or converted, it still remains hard to delineate the fundamental nature of energy itself. Even nobel price laureate Richard Feynman admitted that ”It is important to realize that in physics today, we have no knowledge of what energy is” [1]. In the 18th _{century it was believed that heat was a phenomenon caused by an invisible}

weightless fluid called caloric that flows through pores from a hotter object to a colder one. The theory could not be applied in every situation and showed signs of misunderstanding of the heat phenomenon. The first scientist that paved the way to a modern definition of energy was Sir Benjamin Thompson. He challenged the idea of the caloric theory and said that caloric could not be a real substance and that heat must instead be a form of motion [2]. After Thompson’s introduction of ’The mechanical theory of heat’, French physicist Sadi Carnot further developed the theory which stated the equivalence of heat and mechanical work [3]. Even though it is difficult to understand the very fundamental nature of energy, it can be said that it is perhaps the most essential thing for human survival together next to water. When NASA looks for existence of any living species on another planet, the first thing they verify is the presence of water. Water helps to give DNA its shape and consequently it helps to give life its shape [4]. In order to get clean water, however, energy is needed. Energy and water are in fact closely related and will always form the base of our society. The global energy use has been growing rapidly in the past decades and will only keep growing in the future as it sustains our economies and human development. Richard Smalley, a noble price laureate, stated that energy, followed by water, are the top priority problems for humanity in the next 50 years [5].

(22)

1.2 The energy transition

Throughout the history, the main source of energy has evolved from wood, during the major part of the 1800s, to coal until about 1950 and eventually oil, which is the most important source of energy today [6]. The graphic on figure 1.1 shows the dominant energy source in the US from 1800 until 2013 indicating the growing importance of natural gas and renewable sources in recent decades. In 2018, natural gas demand grew 4,7%, while oil demand rose 1,3% and coal 0,7%, respectively [7]. The share of renewables is also steadily increasing with an estimated growth of 4% in 2018 and will continue to do so in the future. In that year, renewables accounted for over 25% of the global power output and 45% of the world’s electricity generation growth. CO2 emission grew with 1,7% to a total of 33,1 Gt, mainly in

China, India and the US. France, Germany, Japan, Mexico and the UK reported declining numbers of CO2 emission in 2018.

Figure 1.1: Dominant Energy Source in the USA from 1800 until 2013 [6]

In Belgium, there are limited ways to produce vast amounts of renewable energy. The sun does not shine as intense as in desert areas like Morocco or California nor is there place to build massive solar installations. The oceanic climate does bring a lot of wind over the land but again, in a densely populated country as Belgium, it is not evident to build onshore wind turbine parks. In a geographical way, Belgium is not as favoured as France, where mountainous areas can host hydroelectric dams or Iceland, where geothermal energy can be harvested in large parts of the country. Major renewable energy sources in Belgium in 2018 were: hydro (1424 MW), wind (3157 MW), solar (4026 MW) and bio-energy (908 MW) [8]. If the possibilities to produce renewable energy are limited, the focus should be on lifting the efficient use of it. Buildings are an enormous source of untapped efficiency potential. In 2018, residential buildings accounted for almost 8% of the global power demand and 6,6% of the global CO2 emissions [9]. Passive radiative cooling is, for example, a technique that

(23)

Chapter 1. Introduction to the subject 3

has a huge potential reducing energy use in buildings. Air cooling systems are a major source of greenhouse gases and power consumers as well. An efficiency gain of 10 to 20% can already bring a significant difference in their associated power demand. Despite the precarious Belgian weather, PV panels will be an important renewable source in Belgium. According to a study of the Flemish Institute of Technological Research (VITO), there is still a large untapped potential on the roofs of Flemish houses [10]. Their investigation stated that a possible 62 GWh of electricity from solar energy could be generated each year while all Flemish households combined consume only 11 GWh. As the technology behind solar energy becomes more mature, the prices to generate solar energy drops exponentially and will continue to do so in the coming years [11]. On figure 1.2, the evolution of the global weighted average total system costs breakdown of utility-scale solar PV systems can be seen. It is expected that the fixed cost (in US$/kW) of PV systems will drop below 1000 US$/kW.

Figure 1.2: Global weighted average total system Costs breakdown of utility-scale solar pv systems, 2009-2025 [11]

1.3 Hydrogen, the missing link in the energy transition?

In 1874, Jules Verne published his book ’The Mysterious Island’ in which he already men-tioned the idea of using hydrogen as a source of power in the future. He quotes ’Yes, my friends, I believe that water will one day be used as fuel, that the hydrogen and the oxygen which constitute it, used alone or simultaneously, will provide an inexhaustible source of heat and light and an intensity that coal can not have’ [12]. About a century later, in the late 1960s, the idea of so-called ’hydrogen economy’ was again brought up by scientists at a dinner during a scientific meeting in Stockholm [13]. They discussed about how it could be cheaper

(24)

to transmit energy in the form of hydrogen through pipelines rather than sending it in copper wires. Under the impulse of T. Nejat Veziroglu, a professor at Miami university, academic meetings on hydrogen were organized and the International Journal of Hydrogen was estab-lished in 1976, with Veziroglu as editor in chief. The journal serves as a central platform to exchange ideas, research outcomes and recent technological developments ever since.

But what exactly is a ’hydrogen economy’ ? It is the idea to use hydrogen as a low car-bon energy carrier to replace, for example, gasoline as transport fuel or natural gas in various applications. It can also serve as a way to store renewable energy from intermittent sources, such as wind and solar. Despite being the most abundant element in the universe, hydro-gen is rarely found on earth in its molecular form. It is mostly found in the form of water (H2O) or natural gas (CH4). Before becoming a viable fuel, hydrogen has to overcome three

important hurdles: safety, infrastructure and cost. Hydrogen has been used in industry for over 40 years in large amounts. Over that period of time, well developed production, trans-portation and utilization methods ensure safe use of hydrogen. One big advantage of (high flammable) hydrogen is that it is many times lighter than air, in contrast to gasoline, and in case of fire, it ascends rapidly into the air. It does not produce burned ash, which is a source for radiant heat. Stored hydrogen can be purged out by safety valves to avoid expansion of the fire. Hydrogen refueling stations have yet been exploded causing anxiety for potential future buyers of fuel cell vehicles (FCV) [14]. Many people might think back of ’hydrogen related’ accidents that happened in the past like the Hindenburg disaster or the unsuccessful Apollo 13 mission. Both events were, however, not caused by hydrogen at all [15]. In the first case, the aluminum-cotton framework caught fire as a result of an accidental discharge on the electrostatic coating on the framework. In the second case, the accident was provoked by a short-circuit in the fan cables in one of the oxygen tanks. Although a hydrogen-natural gas mix is already transported through pipelines, there is no pure hydrogen grid yet. Regular cheap steel pipes do not form sustainable solutions as they suffer from embrittlement after some time, because the tiny H2 molecule can infiltrate in and therefore degrade the

encap-sulating material. Transportation from the few large scale hydrogen production facilities by pressurized trucks would be too expensive. Local, on site production and storage facilities are therefore the remaining option this day. Hydrogen production can only become sustainable if it is driven by renewable energy.

As the renewable energy transition unfolds, hydrogen is regularly put forward as a substitute for fossil fuel in many countries. Nations that lead this transition are Japan, Australia, Nor-way,... Japan is the leading nation for the creation of a hydrogen-based society [16]. Japanese car manufacturing giant, Toyota, is an important player in this project. Recently it planned to develop a concept village next to the famous mount Fuji that should run entirely on

(25)

hy-Chapter 1. Introduction to the subject 5

drogen [17].

TU Delft professor, Ad van Wijk, suggested the Northern Netherlands as a strong candi-date to implement a green hydrogen economy because the following set of reasons [18]. A large-scale green electricity production (especially through offshore wind) is available in the North see in addition to extra power supplied by the COBRA-cable and NorNed-cable. The large existing chemical industry can produce green ammonia and green methanol by using large quantities of green hydrogen. There is yet a lot of gas infrastructure, which can be retrofitted easily and cheaply to transport green hydrogen. An estimated 17,5 - 25 billion euros is needed in total up to the year 2030 for the development of a green hydrogen economy in the northern Netherlands. According to Ad van Wijk, hydrogen can become as cheap as €2-3/kg due to a large scale hydrogen production. The European Union also exhibited their ambition to implement hydrogen into the transition to renewable energy sources in their ’Hydrogen Roadmap’ [19]. The EU goal is to reduce emissions with 45% by 2030 (compared to 2010 emissions) and to have zero-emission by 2050.

It is, however, not always convenient to transform a nation’s power grid to hydrogen. A vast amount of pure, drinkable water is required for the production of hydrogen. Take for example Australia, a well developed country that frequently suffers from droughts due to its particular climate in large parts of the country. A case study from 2019 examined the implementation of a sustainable hydrogen supply chain model [20]. It concluded that large-scale hydrogen production would not be possible with the current hydrogen supply chain model as the available water resources are now already stretched and some Australian cities are reliant on energy-intensive desalinated water plants. Access to clean water is often taken for granted in studies on a hydrogen supply chain model. As a solution, the paper suggested two options: using expensive water from desalination plants or, the more preferred option, recycled water sourced from consistent and abundant wastewater treatment plants. Nevertheless, as it was not more economically viable, it did not give enough argumentation yet to abandon the current, mainly coal fired, power generation. The cost related to the electrolyser usage rate was the most predominant.

1.4 The role of fuel cells

The fuel cell is one of the key pieces to enable a hydrogen economy. Some controversy exists on who invented the principle of fuel cells for the first time. As reported by the United States Energy Department, German chemist Christian Friedrich Sch¨onbein first conducted research on the principle of a fuel cell, which he reported in the Philosophical Magazine in January 1839 [21]. Other sources suggest that the original concept comes from the Welsch scientist

(26)

Sir William Robert Grove who performed experiments submerging platinum plates in a sul-phuric acid solution on one side, and in a separate oxygen and hydrogen tank on the other two ends, creating an electrical current between the electrodes [22]. In 1893, F.W. Ostwald firstly described the theoretical performance of a fuel cell and in 1896, W. Jacques develops the first fuel cell with practical implementations.

There are two common ways to use hydrogen as a fuel: in internal combustion engines (ICE) or in fuel cells. In ICE’s the hydrogen is combusted with ambient air according to formula 1.1 and results in the emission of harmful NOx gases. The efficiency of ICE’s is around 25-35%

[23].

H2+ O2+ N2 → H2O + N Ox (1.1)

Fuel cells convert hydrogen to electricity and heat through equation 1.2 without producing harmful emissions. The efficiency of a fuel cell is much higher than ICE’s, with an electrical efficiency between 45-55% and a thermal efficiency between 35-40%.

2H2+ O2→ 2H2O (1.2)

In the case of vehicles, where onboard storage space is limited, it is way more interesting to opt for fuel cells as they allow a larger travel range for the same amount of hydrogen fuel, compared to ICE’s. In addition fuel cells have the advantage of not emitting NOx. In

gen-eral, it can be concluded that, despite the relative high cost, fuel cells are the best method to convert hydrogen back to useful energy to this day and are therefore a cornerstone of the hydrogen economy.

But what about powering vehicles using batteries like Tesla? This is a story with strong arguments from both sides. First of all, hydrogen has a huge specific energy of almost 40 000 Wh/kg (142 MJ/kg, [23]) while lithium-ion batteries only have 100-265 Wh/kg [24]. This implies that, if allowed by the onboard storage space, the travelling range can easily be extended without adding considerable more weight to the hydrogen based vehicle. For battery powered vehicles, this range extension comes at the expense of a lot of weight and subsequentially higher performing vehicle components like, for example, heavier breaks. The counter argument is the total efficiency of both systems to deliver energy to the electromotor of the vehicle. The production, compression and power generation from hydrogen is different from batteries and has an estimated equivalent efficiency of 33,6%. That is less than half the efficiency of the same process from batteries which has an equivalent efficiency of around 85,7%. The cost per kilometer consequently gets much higher from hydrogen vehicles.

It can be concluded that hydrogen and fuel cells have the potential to power our society in many ways and to replace oil and gas in the far future, as these sources will dry out

(27)

Chapter 1. Introduction to the subject 7

eventually. The greatest barrier is the cost for this change to be implemented. First of all, renewable energy sources need to become cheaper and become more dominant in the global energy supply. The production of hydrogen using main grid electricity will result in an economically non competitive price of hydrogen, compared to batteries in small scale applica-tions. It can therefore be only viable if off-peak or excess renewable electricity is being used for hydrogen production. Secondly adaptations to facilitate the new type of fuel need to be installed. Construction of these advanced technologies like fuel cells, PV panels,... necessarily need rare materials (lithium, platinum, cadmium,...) which also impose limitations for future transitions. Hydrogen, fuel cells and batteries will be of big importance in the future to power our society. However, there will not be a dominant power source as there was in the past. The simplicity in terms of process and required raw materials of oil and combustion engines is hard to rival but should anyhow be conquered.

(28)

1.5 Goal of the thesis

The thesis consists of two major parts: the experimental evaluation and the modelling of the fuel cell in Dymola.

In the experimental evaluation, a test setup will be built in order to observe the behaviour of a commercially available fuel cell. Both the inlet and output variables will be measured. On the inlet, the temperature, pressure and massflow of the incoming hydrogen will be logged. On the other side, the output voltage, current and power will be recorded. Based on this experimentally obtained information, the performance of the fuel cell in various operation modes will be estimated in order to get an idea of its use in a hydrogen based energy storage system. In regard to this experimental performance analysis, a concept evaluation will be conducted on an example hydrogen based energy storage system in an attempt to get more insight in the system requirements and efficiency.

Afterwards, the fuel cell will be modelled in Dymola, a simulation program that uses the object-oriented programming language Modelica. An attempt will be made to adjust the model in accordance with the experimental observation by setting input parameters and ini-tial conditions. The model will be based on libraries of Modelon that contain templates for subcomponents. The possibilities of Dymola for the modelling of this type of fuel cell will be examined and compared to reality.

(29)

Chapter 2

Introduction to fuel cells

2.1 Working principle

A fuel cell (FC) is an electrochemical device that converts the chemical energy of a fuel directly into electrical energy [25]. It, basically, consists of three main components: a fuel electrode (anode), an oxidant electrode (cathode) and an electrolyte in between. In proton exchange membrane (PEM) fuel cells, the type that will be examined in this thesis, molecular hydrogen (H2) is used as fuel and arrives at the anode side in a gaseous form . At the anode,

it undergoes an oxidizing reaction, given by reaction 2.1, and forms hydrogen ions (H+_{) and}

electrons (e−).

H2 ⇒ 2H++ 2e− (2.1)

The formed hydrogen ions migrate through the membrane while the electrons bypass this membrane through an external electrical circuit. On the other side of the membrane, at the cathode, the electrons and the migrated hydrogen ions are consumed together with oxygen supplied from an intake gas flow stream to form water in a gaseous or liquid state, by reaction 2.2.

1

2O2+ 2H

+_{+ 2e}−

⇒ H2O (2.2)

The reactions that occur in the fuel cell come with a heat release and an ability to perform work due to the electron flow. The overall reaction of a PEM fuel cell can be written as follows:

H2+

1

2O2 ⇒ H20 + Wel+ Qheat (2.3)

Figure 2.1 represents the working principle of a PEM fuel cell.

In a fuel cell chemical energy is directly converted into electricity whereas in combustion based technologies a stopover in thermal energy is being made.

Oxygen is supplied to the cathode by the ambient air and is at atmospheric pressure. The

(30)

Figure 2.1: Working principle of a PEM fuel cell [26]

hydrogen gas is supplied at a slightly higher pressure than the atmospheric pressure, for ex-ample around 1,6 bar. The created H20 leaves the fuel cell at the exhaust, most often in a

gaseous state. A fuel cell can be air cooled as well as water cooled, which has, due to its higher heat capacity, a more efficient heat drain ability.

2.2 Types of fuel cells

Multiple types of fuel cells can be distinguished according to the involved gases, type of electrolyte and operating temperature. The types that are most used today are: Alkaline Fuel Cells (AFC), High/Low temperature proton exchange membrane (PEM) Fuel Cells, Phosphoric Acid Fuel Cells (PAFC), Molten-Carbonate Fuel Cells and solid oxide Fuel Cells (SOFC). An overview of the most important fuel cells, with their corresponding fuel, oxidant, electrolyte and operating temperature is given in figure 2.2.

(31)

Chapter 2. Introduction to fuel cells 11

Figure 2.2: Overview of fuel cell types [27]

In the last decade, alternative fuel cell technologies have been extensively investigated. For example, microbial fuel cells use the electrogenic properties of specific bacteria in an effort to generate electricity from different types of wastewater [28].

The focus in this report will be on the PEM fuel cell as the project that will be evaluated makes use of this type. All further mentioned properties and characteristics concern the PEM fuel cell unless explicitly indicated.

2.3 PEM fuel cell

In this paragraph the composition of a proton exchange membrane or polymer electrolyte membrane fuel cell will be broken down and explained as well as the key features of this type.

A PEM fuel cell consists of three layers: the gas diffusion layer, the catalyst layer and the active layer that will be explained below [25, 29, 30].

The gas diffusion layer

Hydrogen is forced by the applied pressure through the inlet channel, passing the diffusion layer to the exit. The higher the flow rate, the better the distribution of the diffusion layer. Driven by the concentration difference between both sides of the diffusion layer, H2 is diverted

into the layer through the pores leading to the catalyst layer, extending out into the anode. This phenomenon is illustrated on figure 2.3.

(32)

Figure 2.3: Concentration difference in diffusion layer [29]

The catalyst layer

The catalyst layer sits between the membrane and gas diffusion layer, on both (anode and cathode) sides of the membrane. It consists of a carbon frame on which nano-sized particles of Platinum are distributed. This catalyst is mixed with an ionomer (ion-conducting polymer) that allows the protons to migrate through the layer. On the anode side, H2 molecules are

split into separate H+ ions, which consequently migrate through the active layer: the mem-brane. On the cathode side, the catalyst helps in the reduction of oxygen and subsequent production of water.

The active layer - membrane

A fuel cell membrane must exhibit relatively high proton conductivity, must present an ade-quate barrier to mixing of fuel and reactant gases, and must be chemically and mechanically stable in the fuel cell environment [31]. The most common ionomer used for membranes for PEM fuel cells are made of perfluorocarbonsulfonic acid (PSA). Nafion, made by Dupont, is the most well known membrane material. Hydrogen protons are transported through the active layer to the cathode side while the electrons bypass the membrane by means of an external circuit. To obtain this separate flow, it is essential to design the membrane in a way that electrons cannot pass through.

(33)

2.4 Production of hydrogen

Hydrogen is used in the PEM fuel cells as the main fuel. H2 is most often produced by

means of industrial processes. Nevertheless, natural hydrogen wells have yet been discovered. Recently, a natural hydrogen gas field, of 98% purity, in the Bourakebougou area in Mali was discovered during the drilling for a water well [32]. Although it will take some time to use these natural sources in an efficient way, major energy companies are keeping an eye on the possible natural hydrogen wells. Industrial processes to obtain a desired purity of over 99% are the most common way of production these days. An overview is given below in the chart 2.4, [33, 34].

Figure 2.4: Brief overview of H2 production methods today

95% of the produced hydrogen in the US comes from natural gas reformation in large central plants [35]. In the first endothermic process, high-temperature steam (700°C–1,000°C) reacts with methane, in the presence of a catalyst in order to form hydrogen, CO and small fractions of CO2. Secondly, again by using a catalyst, CO reacts with steam creating CO2 and more

hydrogen. Finally, CO2 and other impurities are removed from the gas stream resulting

in practically pure H2. The process can be applied to methane, ethanol, propane or even

gasoline.

2.4.1 Electrolysis

The production method of hydrogen used in the concept related to this thesis is electrolysis. Electrolysis is basically the opposite reaction of what happens in a fuel cell. The technique was already applied for commercial purposes in the 1890s [36]. By applying a DC current to two metal coated electrodes in water, one will become negatively charged (cathode) the other

(34)

one positively charged (anode). Then hydrogen, H2, at the cathode and oxygen, O2, on the

anode will be created. The rate of production depends on the current, which equals voltage times conductivity (I = V· σ). To raise the conductivity, an electrolyte is added to the water (e.g. salt). Throughout the history of the development of electrolysers, this water-salt mixture has been replaced by more efficient solid electrolytes. Commercial, low temperature electrolyser systems are now able to reach an efficiency between 56 - 73% [33]. Three main types of electrolysis exist. First of all, the alkaline based electrolyser. It is the most common, most developed and least costly electrolyser. Hydroxide (OH−) is transported, through the electrolyte, from cathode to the anode side where oxygen is formed. The hydrogen stays at the cathode side, in the water solution and is externally separated out the solution. Typical efficiencies for this type are between 50-60 %.

A second type, is the proton exchange membrane (PEM) electrolyser where hydrogen protons (H+_{) travel through the membrane and recombine at the anode side. Oxygen is formed at}

the cathode side. This technique is more expensive than the alkaline electrolyser but more efficient (55-70%).

A last type of electrolyser are the solid oxide electrolysis cells. The efficiency of this type is largely dependent on the temperature. Both thermal and electrical energy is used to split water into hydrogen and water. Expensive materials are, however, needed for this high tem-perature operation. Efficiencies for this type range from 60 to even 90 % [37].

Commercial electrolysers can produce hydrogen with a purity of 99.9995% at up to 1.05 Nm3/h [38]. By 2030-2040 it is expected that hydrogen produced with excess energy of re-newable sources can become competitive with the production by natural gas as this latter production method leads to avoidable CO2 emission and gas prices will start to rise [39].

Once hydrogen is created, it needs an appropriate way to be stored efficiently. Being 14 times lighter than air, H2 takes up a large volume of space at atmospheric pressure and

temperature. Therefore, compression of the gas is essential before storing it in a tank in an efficient way. In addition, the gas can be cooled to slow down the molecules in order to take up less space. Hydrogen can be stored with different techniques [40]. It can be compressed from 20 to 100 Mpa in a carbon fibres composite pressure vessel, liquefied in cryogenic form at -253◦C for transportation of large amounts of hydrogen and when high gravimetric per-formance is desired. For intermediate low temperatures (around 20 K) and high (>30 MPa) pressure, the gas can be cryo-compressed in order to obtain improved gravimetric perfor-mance. Metal-organic frameworks can also be a solution to store hydrogen in a more efficient, compact way. These programmable materials, consisting of metals and organic ligands, are light weight, have tunable pore sizes and large free volumes. In the past decade, a lot of

(35)

research has been done on these types of materials. Besides storing gases, MOF’s can also be applied to separate two types of gases based on their polarizability and size.

Even when it is stored, hydrogen requires special attention for the material in which it is contained. Due to the fact that hydrogen is a very tiny molecule, it finds its way through the encapsulating material. Hydrogen embrittlement can be the cause for degradation of the storing vessels and new materials are required if hydrogen is to be used on a large scale as a replacement for hydrocarbon fuels [41, 42]. Hence, coatings are often the solution to prevent this phenomenon from occurring. Metal-organic frameworks are getting more attention for their possibility to store hydrogen among others.

2.5 Thermodynamic and electrochemical principles of

hydro-gen fuel cells

Efficiency

Since a fuel cell is an electromechanical device the amount of heat that could be converted to useful work is not limited by the Carnot efficiency, in contrast to heat engines [25]. Instead the Gibbs free energy bounds the maximum conversion efficiency:

ηrev =

∆Gf

∆Hf

(2.4)

Where ∆Gf is the change in Gibbs free energy of formation during the reactions and ∆Hf is

the change in enthalpy of formation. (hf)H2 and (hf)O2 are zero by definition of enthalpy of formation. ∆Hf = (hf)H20− (hf)H2 − (hf)O2 = (hf)H20 (2.5) ∆Gf = ∆Hf − T ∆Sf (2.6) ∆Sf = (sf)H20− (sf)H2− (sf)O2 (2.7) At 298.15 K, the change in Enthalphy, Gibbs Free Energy, and Entropy of the hydrogen/oxy-gen reaction equal -286.02 kJ

mol, -237.34 kJ

mol and -0.16328 kJ

mol·K, respectively. These values are

temperature dependent and therefore result in a slightly different efficiencies. For standard conditions, at 298.15K, the ideal fuel cell efficiency is:

η = ∆G

∆H =

237.34

286.02 = 83% (2.8)

Alternatively, this ideal efficiency can also be computed as the ratio between the theoretical cell potential, given in equation 2.9 and the potential corresponding to the higher heating value

(36)

of hydrogen. This ratio leading to the same theoretical efficiency is presented in equation 2.10. Where n are the number of electrons per molecule of H2, 2, and F Faraday’s constant.

E298K,1atm = −∆G nF = 237.340 2· 96.485 Jmol−1 Asmol−1 = 1.23V (2.9) η = −∆G nF −∆H nF = 1.23 1.482 = 0.83 (2.10)

In reality, PEM fuel cells have an efficiency between 40-50% [43].

2.6 Performance evaluation

The voltage can be predicted using the Nernst equation. This equation is given in its general form in equation 2.11 [29]. E = E0−R· T n· F · ln Q aci products Q aci reactants ! (2.11)

ci represents the stoichiometric coefficient of the corresponding substance in the chemical

equation and a represents the species concentration, gas pressure, and so on. R and F are the gas and Faraday constant, respectively. E0 _{is the voltage at reference conditions. T is the}

temperature. For PEM fuel cells, the partial pressures are used to substitute the values of a in the Nernst equation: aH2 = pH2, aO2 = pO2, aH2O = 1. So applied to the corresponding reaction for PEM fuel cells, equation 2.3, the Nernst equation becomes:

E = E0₋ R· T n· F · ln " 1 pH2· p 1/2 O2 # (2.12)

In reality, a couple of phenomena affect the theoretical voltage output, calculated via the Nernst equation. Three major losses contribute to the reduction of output voltage of the fuel cell [27, 29, 30, 44].

Activation loss

To initiate the chemical reaction at both the anode as the cathode, a certain energy barrier needs to be exceeded. This can be compared to the spark in an internal combustion engine and results in a voltage loss from the moment some current flows in the fuel cell. This loss can be described by the Butler-Volmer equation (2.13) [27].

j = j0· exp αa· z · F R· T (E− Eeq) − exp αc· z · F R· T (E− Eeq) (2.13)

In this equation i represents the current density, i0the exchange current density, α the transfer

coefficient. The exchange current density can be written as presented in equation 2.14 and reflects the electrode activity at equilibrium for a particular reaction [44]. The higher the

(37)

exchange current i0, the lower the barrier for electrons to overcome, the more active the

surface electrode is and the higher the performance.

i0= iref₀ · ac· Lc· Pr Prref γ exp −_RE_{· T}r (1₋ T Tref ) (2.14)

Where irefo is the reference exchange current density, ac the catalyst specific area, Lc the

loading of the catalyst, γ the pressure coefficient, T and Tref the temperature and reference

temperature respectively.

The first term of the Butler-Volmer equation becomes dominant because of the high neg-ative overpotentials (E-Eeq << 0) at the cathode. At the anode, the second term becomes

dominant because of high positive overpotentials (E-Eeq >> 0). In the region with a high

overpotential, the Butler-Volmer equation can be simplified to the Tafel equation, presented in its general form in equation 2.15 [27].

∆Vact= a + b· log(i) (2.15)

As a result of this simplification, the activation loss at the anode and cathode side can be written as in equation 2.16 and 2.17, respectively.

∆Vact,a= Ea− Er,a= RT αaF · ln i i0,a (2.16) ∆Vact,c= Er,c− Ec= RT αcF · ln i i0,c (2.17) The total activation loss can now be written as in equation 2.18. Since the reaction at the cathode, the reduction of oxygen, needs higher overpotentials than the anode reaction, this corresponding loss, ∆Vact,c, dominates [27]. By approximation the activation loss at the

anode, ∆Vact,a, can be left out.

∆Vact= Er− ∆Vact,c− ∆Vact,a

= Er− RT αcF · ln i i0,c − RT αaF · ln i i0,a ≈ Er− RT αcF · ln i i0,c (2.18) Ohmic loss

Ohmic losses are the consequence of the fact that no conductor is perfect. Charge transport is always accompanied with losses induced by the material itself. In this case, these losses are reflected in terms of voltage loss accross the length of the conductor. The voltage losses are dependent on the cross-sectional A and the length L of the conductor, the conductivity σ of