Dynamic water management of an open-cathode self-humidified PEMFC system

Transcription

1 Dynamic water management of an open-cathode self-humidified PEMFC system Doctoral Candidate: Director: Maria Serra i Prat Jordi Riera i Colomer Tutor: Assensi Oliva i Llena Thermal Engineering Department Universitat Politècnica de Catalunya

2 Contents: 1. Introduction Sustainable energy, hydrogen as an energy carrier PEMFC stack and system design summary Automotive system description Stationary system description Portable system description The role of water in the PEMFC Literature review Electrochemical reaction Mathematical modeling Fuel cell performance models Fundamental water transport experiments System models Fuel cell system operation and control Objective and scope Thesis objective Approach to the thesis Methodology Fuel cell test station Fuel cell test station design and specifications Description of experiments Water transport experimental characterization Diffusion coefficient experimental design Electro-osmotic drag coefficient experimental design Heat transfer constants Effect of the 3 control mechanisms The duty cycle and period of the hydrogen purge valve Stack short circuit duty cycle and period Fuel cell system numerical model Expected results Hypothesis on the effects of the different control mechanisms: Contribution Work plan/timetable Required resources Relevant published works Bibliography i

3 Abstract: Polymer electrolyte membrane fuel cells (PEMFC) are gaining increased attention as viable energy conversion devices for a wide range of applications from automotive, stationary to portable. However, to optimize performance, these systems require active control and thus in-depth understanding of the system dynamics. Understanding the water transport mechanisms through the membrane and membrane water content is a main issue in PEMFC. The control of these phenomena is still a problem to be solved. This thesis proposal is based on the hypothesis statement that the management of membrane hydration of an open cathode self-humidified PEMFC system may be improved through fast acting control mechanisms including changes in fan velocity, hydrogen purging rates, and the frequency of the stack short circuit. Hence, this proposal describes the methodology for experimentally characterizing the water transport mechanisms through the membrane and membrane water content for the purpose of studying their dynamic influence on fuel cell system performance using a control-oriented model and experimental observation. The methodology will be applied to an open cathode, self humidified and hydrogen purged PEMFC system with the following simple control mechanisms: single fan for both cooling and air supply, a hydrogen purge valve, and periodic short circuit of the stack. The system performance can be significantly improved with greater understanding of water flux dynamics through the membrane and its dependence on the internal conditions of the fuel cell. This can be achieved by experimentally determine physical parameters of a control-oriented model. The analysis will then lead to an in depth understanding of how a system with minimal parts and complexity can be improved in terms of overall efficiency, stability and operating rage. The conclusions from this work will also show how the system reacts to changes in the environmental conditions. These conclusions will provide guidelines to develop a proper control strategy with the necessary sensors to ensure that the system operates properly under a wide range of environmental conditions. ii

4 1. Introduction Modern industrialized society is dependent upon energy for its economic growth, and standard of living. In the order of 85% of the energy is obtained from fossil fuels [DOE/EIA, 2001]. Availability of these fuels within the earth is finite and their use has harmful environmental impacts, which in turn cause direct and indirect adverse effect on the global economy [Stern, 2007]. Necessary steps need to be taken towards the development of new energy production technologies before these fuels become scarce or irreversibly change the planet. Investing in sustainable energy production will also hedge against fossil fuel uncertainty and carbon policy. Thus by fixing the cost of energy production, sustainable energy will create economic stability. Therefore, there is an impelling need for a sustainable energy approach capable of maintaining future economic standards and minimizing the environmental impact due to energy production. The cost for the development of sustainable energy technologies is substantial. Thus, the monetary and energy investment required for these technologies should be made early enough so that, by the time the supply of fossil fuels becomes limited and, therefore, very expensive, the transition has already occurred towards other sustainable energy techniques. Various predictions have been made on the time it will take for these fossil fuels to be depleted. Most of these predict fossil fuels production will reach its peak by the middle of this century or sooner. Therefore, the transition towards other means of energy production, storage and distribution must take place now. 1.1 Sustainable energy, hydrogen as an energy carrier If one looks historically at the types of fuels used by human civilization to produce energy there is a clear trend towards fuels with higher hydrogen to carbon ratios. This movement towards fuels that contain decreasing amounts of carbon is due to their higher efficiencies, lower costs, and environmental friendliness [Barbir, 2005]. So, from this it can be extrapolated that the logical choice for the fuel of the future is pure hydrogen. However hydrogen does not exist on earth in its pure form. So, it will be probably produced initially from fossil fuel and then from sustainable energy sources. Another reason why theoretically hydrogen would be an ideal energy carrier is because it is very light and has minimal resistance to flow. Hydrogen can be also used in many ways to produce thermal, mechanical or electrical energy by burning it directly, using it as a fuel in a combustion engine or transforming it directly into electricity through a fuel cell. However there are still quite a few engineering challenges that need to be met such as hydrogen production, storage, distribution, and efficient low cost energy converters. With proper investments these technological challenges can be overcome. For most sustainable energy production technologies there is usually a limitation due to when the energy can be produced. For instance, solar power only works when there is sun or wind power when there is wind. So, when the demand for energy is low and the supply is high, hydrogen can be produced, stored and then used when needed. Hydrogen is one possible energy carrier that has the potential to fill this need. 1

5 Some of the unique properties of hydrogen that make it appealing for energy storage and distribution are: When hydrogen is converted into energy the product is water. The main raw material used to generate hydrogen is water, and the product of converting hydrogen back to energy (through combustion or electrochemistry) is water. So there is a complete cycle with minimal byproducts (some NO x can be produced with combustion). Hydrogen can also be produced from biomass or fossil fuels however with undesirable byproducts. (such as CO and CO 2 ) Hydrogen can be generated or converted to electricity relatively easily and efficiently. Hydrogen can be stored in multiple forms for example: gas, liquid, metal or chemical hydrides. It can be transported through pipelines Hydrogen safety compared to other currently used energy storage media is similar. [Swain, 1992],[Thomas, 1996],[Romm, 2004] Hydrogen has a high energy density based on mass: the energy density of hydrogen is 120 MJ/kg, compared to natural gas, which is 50 MJ/kg. However there are some technical challenges and disadvantage to hydrogen as an energy carrier. Hydrogen has a low volumetric energy density, 10.7 MJ/Mm3, compared to natural gas which is 39 MJ/Mm3. This means the volume needed to store hydrogen is relatively large. Hydrogen does not exist on earth in its pure form so it always has to be produced, so the efficiency of hydrogen production is also critical. The obvious question then is, can we produce enough energy from sustainable sources? Yes, it is estimated that the world energy needs in 2030 will be 30 billion kilowatts and it could be supplied by only capture of the sunlight that falls on the earth [Kurzweil, 2005]. But there are other forms of energy that can be utilized, for instance geothermal and tidal energy that are not dependent on the sun. So, if hydrogen can be used as an energy carrier then devices need to be developed that can transform hydrogen into electricity in a simple, cost effective, and efficient manner. Polymer electrolyte membrane fuel cells (PEMFC) are one such device. The advantages associated to PEMFC technology compared to other types of energy conversion devises are: High efficiency 40-70% depending on the design and application Run at relatively low temperature 25-80ºC Minimal moving parts, fan/compressor and valves. Single step process to go from fuel to electricity Cost of the units can be reduced significantly with economies of scale and future material development due to its simplicity and low temperature. 2

6 1.2 PEMFC stack and system design summary Advances in PEMFC technologies have been rather stunning in the past decade in terms of materials, component design, production, and system power density improvements. However there is a lot to be learned in terms of controls. With the advancements in microprocessor technology it is possible to make reliable and inexpensive control systems for the fuel cell. But first it is important to understand the different types of systems and how they need to be controlled. In this section a description of the basic variations in systems that have been produced to date for automotive, stationary, and portable applications. Each type of system has its advantages and disadvantages Automotive system description Automotive systems are generally more complex, mainly due to the need for high power densities and must function in a rigorous environment. The systems generally have four main subsystem; air management, hydrogen management, water and thermal management and power management. The complexity of the system allows for greater control of the fuel cell subsystems so that it can operate at a wide range of environmental conditions. However this adds to cost and increased number of failure modes. Figure 1.1 shows a schematic of a general automotive fuel cell system and the prototype GM Hy-Wire skateboard design built from the ground up to demonstrate PEMFC technology. a. b. Figure 1.1 a.- PEMFC system schematic. b.- GM Hy-Wire fuel cell automotive platform Stationary system description Stationary system do not a have such a restriction in terms of power density and therefore the fuel cell can be directly air cooled. For instance the Ballard Nexa fuel cell system (see Figure 1.2) has been designed for stationary application and has an air compressor for the reaction air, a fan for cooling, hydrogen purge system, and a humidity exchanger between the inlet and outlet reactant air. This system is fairly 3

7 robust and can be operated indoors without any extra safety sensors. However this system is still fairly complicated in terms of the subsystem components and controls. Figure The Ballard Nexa 1.2 kw fuel cell system Portable system description Portable systems are generally the simplest and usually do not have high power densities. The range of complexity varies from completely passive to a few basic controls. For instance PaxiTech has an air breathing passive system as shown in Figure 1.3. Fig PaxiTech fuel cell system with no active control system The limitations associated to these types of systems are that they are generally limited by the environmental conditions, have a slow response time to changes in power, and have relatively low power densities. So the ideal system would be one that has basic control of the important parameters in the system to minimize these drawbacks. For this reason the Horizon H-100 is chosen which has a reasonable power output of 100 watts and the design can be easily scaled up to 5000 watts or reduced down to 12 watts [Horizon Website, 2008]. The control mechanisms are simple and inexpensive. 4

8 However the system efficiency can be improved by understanding the nature of the transfer of water across the membrane and the water content in the membrane. Figure Horizon H-100 portable fuel cell system 100 Watt nominal power 1.3 The role of water in the PEMFC Water both serves and hinders the functionality of the PEMFC. On one hand, the membrane needs to be saturated for it to be protonically conductive, and on the other hand if there is liquid water in the catalyst layer and/or the gas diffusion layer it will block the diffusion of the reactant gases to the catalyst sites. Ironically, the water generated by the reaction is generally not sufficient to maintain adequately membrane saturation, and thus the reactant gases need to be humidified. Humidifying gases is a costly procedure, not only in terms of extra component and complicated controls, but also in energy required to vaporize water. So, having a system that can "self-humidify" the membrane can simplify the system greatly. The key to designing self-humidifying fuel cell systems is to understand how the water that is generated by the reaction can be kept in the membrane and removing any liquid water from the catalyst, gas diffusion layers and channels. Traditionally, the problem of water management in the membrane has been resolved using a steady state approach. The trouble with this solution is that water tends to build up in some sections of the cell, and/or the membrane tends to dry out in other sections. Given that the reactant gases are fully saturation and at low pressures, as well as at elevated temperatures, any small temperature variation can cause either condensation or excessive evaporation in parts of the cell which may cause instability. Recently, a fuel cell system with a novel approach to solving the water management issues has become commercially available. The Horizon H-Series PEMFC system is a 5

9 simple self-humidifying system with an open cathode. A single fan is used for both cooling and supplying oxygen for the reaction. A dead ended anode with a periodic hydrogen purge valve removes water and nitrogen (that has crossed over from the cathode) from the anode flow field. Finally, a periodic short circuit of the stack provides water and heat to the cathode catalyst. The idea is to shock the fuel cell periodically which creates water and forces water to be redistributed in the fuel cell. The fan controls the fuel cell temperature, which varies, depending on the power output. However there is still a lack of understanding in how these control actions effect the redistribution of water in the fuel cell. Currently controller for H-Series system stabilizes the fuel cell by purging and short circuiting the stack at a constant interval regardless of the fuel cell conditions. Therefore the system efficiency can be significantly improved with greater understanding of water flux dynamics through the membrane and its dependence on the internal conditions of the fuel cell. 2. Literature review How fuel cells work on the surface seems quite straight forward. Add hydrogen to one side of a catalyzed membrane and oxygen to the other. The catalyst brakes up the two molecules and recombines them back together to create water with a byproduct of electricity, as shown in Figure Electrochemical reaction The electrochemical reaction in the fuel cell occurs on both the anode and cathode electrode simultaneously according to the following basic reactions. Anode electrode reaction: Hydrogen is supplied to the anode flow field where it diffuses through a porous, electrically conductive, rigid material or otherwise known as the gas diffusion layer (GDL). From there it comes into contact with the anode electrode catalyst layer (which is generally made of platinum) where an oxidation reaction splits the hydrogen molecule (H 2 ) into two protons and two electrons. H H e The protons permeate through the ionically conductive membrane driven by the voltage gradient to the cathode side, while the electrons travel through the GDL, conductive plates, and finally through the external load to the cathode electrode. Cathode electrode reaction: On the cathode side, the oxygen (usually from air) is supplied to the fuel cell where the oxygen needs to diffuse through its GDL and reach the cathode catalyst layer (which is also generally made of platinum). There it combines with the proton that 6

10 traversed the membrane and the electron that traveled through the electric load. Thus, the product of the reaction is water and the byproducts are heat and electricity. + 4H + 4e + O2 2H 2O Overall reaction: The overall reaction is hydrogen plus oxygen goes to water and heat. 2H 2 + O2 2H 2O The previous reactions may have several intermediate reactions that may have undesired products but are not necessary to describe the main processes in the fuel cell and will not be included in the discussion. Figure Basic principles of the working of a polymer electrolyte membrane fuel cell For this reaction to occur there are a few technical issues that need to be resolved. The majority of them are associated to water and its distribution in the PEMFC stack. Water affects the fuel cell in two main ways: either to dry or to wet. The membrane used in the standard PEMFC requires a certain level of hydration for efficient conductivity of protons, and if there is to much liquid water in the cell then it hampers the necessary uniform distribution of the reactant gases and its diffusion to the catalyst layers. Because of these numerous processes that take place simultaneously; it is therefore important to understand these processes and their mutual interdependence. 7

11 First fuel cell law: One cannot change only one parameter in a fuel cell. Change of one parameter causes a change in at least two other parameters, and at least one of them has an opposite effect of the one expected to be seen. [Barbir, 2005] 2.2 Mathematical modeling The models in the literature can be categorized into two main groups; fuel cell performance models, system models. PEMFC performance models are generally designed to simulate the physical processes at the cell level. On the other had system models generally simplify the fuel cell and simulate the subsystem interactions Fuel cell performance models Many models have been created to simulate the processes that occur in the fuel cell, which take into account fuel cell design parameters and operating condition. These models will be defined as fuel cell performance prediction models and generally only describe the fuel cell and not the whole system. They are usually steady-state and are developed at the cell level including spatial variations of the parameters. They apply complex electrochemical, thermodynamic, and fluid mechanics principles to predict fuel cell voltage by calculating the individual losses. The main individual voltage losses that occur in the fuel cell are (see Figure 2.2): Kinetics of the electrochemical reaction losses (i.e. activation polarization) Internal currents and reactant gas crossover losses (generally insignificant) Internal electrical and ionic resistance losses (i.e. ohmic losses) Mass transport losses (Concentration polarization) 8

12 Figure 2.2 Approximation of the individual voltage losses in a PEMFC and resulting polarization curve Early researchers intended to determine the limitations in the fuel cell operating range due to water management and reactant distribution, because these issues generally caused the cell to work reliably in a relatively small window of operation [Larminie, 2003]. So, to resolve these range and stability problem most of the early work focused on developing models to manage the water and the distribution of the gases at an optimal steady state condition simulating temperature, oxygen concentration, nitrogen, hydrogen concentration and humidity that enters the fuel cell and thus determining the distribution of the gases and in some cases the liquid water in the cell [You, 2002]. These models establish the fundamental effects of the operating parameters, such a pressure, temperature, flow rates, and inlet relative humidity s on the fuel cell voltage. The publications that calculate the fuel cell resistance to predict the fuel cell polarization curves at various operation conditions are: [Amphlett, 1994],[Mann, 2000],[Thampan, 2000]. The publications that are based on mass transport of the species in the fuel cell can be split into one-dimensional ([Bernardi, 1992],[Springer, 1993]), two-dimensional ([Broka, 1997], [Dannenberg, 2000], [Gurau, 1998], [Nguyen, 1993]) and three-dimensional [Berning, 2003], [Wang, 2003] models. These types of models are used to determine the fuel cell performance and efficiency under steady-state operating conditions and to optimize the fuel cell components. However, due to there complexity, these models generally can not be integrated into control-oriented fuel systems models. The water transport equations in most these of models are based on the ex situ water transport experiments of a Nafion 117 membrane done by [Springer, 1991] and [Zawodzinski, 1991]. The experiments done by these two studies have created a 9

13 baseline for the industries and show the relationship between electro-osmotic drag (EOD) and water diffusion through the membrane with respect to membrane water activity and membrane temperature. Most of the issues associated to system design and control can be attributed to water not being in the right location [FCHandbook, 2002]. Thus, the flowing section will describe the experimental work done to determine the coefficients needed to evaluate water transport, which are water diffusion through the membrane and EOD Fundamental water transport experiments The research published on the two main transport mechanism of water across the fuel cell membrane will be examined; EOD and diffusion. Hydraulic permeation of water will not be discussed. This phenomenon has been found to be an order of magnitude smaller than the two main transport phenomena [Husar, 2008]. Another reason that it can be neglected is due to the nature of the design of the system in question. The open cathode configuration with forced flow should minimize the liquid water in the cathode GDL and the pressure differential across the membrane is relatively small (in the order of less then 0.5 bars). It has been found that water transport by both EOD and diffusion is dependent on membrane water content (or membrane activity) which is defined as the number of H 2 O molecules per SO 3 -. SO 3 - is the ion conducting molecule in the perfluorocarbonsuflonic acid (PFSA) ionomer, which is the typical ionomer used in most PEMFC. Water content is not a variable that can be measured directly when a fuel cell is assembled. Relationships have been developed to relate water content with water vapor concentration in the gas channel and membrane resistance. The studies will be split into two groups, ex situ and in situ experiments. Ex Situ Experiments: Most models in the literature use the water diffusivity and EOD coefficients determined by [Zawodzinski, 1991] and [Springer, 1991]. The ex situ measurement of water transfer were done on a Nafion 117 membrane. The membrane was fully hydrated and placed in equilibrium with liquid water. The experiments were done at 30ºC with a few data points at 50ºC. The number of water molecules dragged per every proton (electro-osmotic drag) was determined to be 2.5±0.2 H 2 O/H + - with a membrane water content of 22 H 2 O/SO 3 at 30ºC. For a membrane which is not fully saturated (11 H 2 O/SO - 3 ) the EOD coefficient was found to be approximately 0.9 H 2 O/H +. Therefore the EOD coefficient was determined to be a function of the membrane hydration. Another study by the same author [Zawodzinski, 1993], showed the relationship between the intra-diffusion coefficient and membrane water content using a pulsed-field gradient spin-echo nuclear magnetic resonance (NMR) technique. Another study by [Zawodzinski, 1995] shows that the EOD coefficient depends on the membrane water equilibration state, with liquid or vapor. Even though the diffusivity and EOD data for the membrane were accurate, direct application of such data to a real fuel cell may not be appropriate due to the fact that this data was collected ex situ, and can not be considered constant because they may vary depending on other operating conditions like temperature. 10

14 [Fuller, 1992] reported a water transport number which is similar to the EOD coefficient. Its value is 1.4 for a membrane equilibrated with saturated water vapor at 25 C, decreases slowly as the membrane is dehydrated, and falls sharply toward zero as the concentration of water approaches zero. This corresponds well with the other works and show the relationship between EOD and membrane water content. The study however, is not complete because the tests were not done at standard cell operating temperatures. [Morris, 1993] indicated that the diffusion coefficient of water in Nafion 117H is a strong function of membrane water content. It was also observed that the rate of water desorption will differ from the rate of absorption. The study also showed that the membrane resistance increases sharply when the water content of the membrane is less then 6 H 2 O/ SO 3 -. In Situ Experiments: [Dubar, 2007] used magnetic resonance imaging to measure water distribution in an assembled fuel cell using an MEA assembled from a Nafion 115 membrane. This study found that, at low current densities, water was transported from the cathode to the anode. This signified that diffusion forces appeared to dominate over the electro-osmotic forces. [Trabold, 2006] used in situ neutron radiography to investigate how and where water accumulates in a flow field and how different parameters such as humidification of reactants affected the water accumulation. They demonstrate that water tends to accumulate in the 180 bends of the serpentine anode and cathode flow fields and that the current density and cathode stoichiometric ratio affects the quantity of water accumulated in the fuel cell. A net water transport study was performed by [Choi, 2000] for a Nafion 115 membrane, which resulted in a water transfer coefficient due to both EOD and diffusion. They showed that the net transfer coefficient, with humidified gases, decreased sharply from 0 to 200 ma/cm 2 and was nearly constant from ma/cm 2. However, when the cathode was not humidified and with low current densities, most of water generated at cathode was absorbed by the membrane. On the other hand at high current densities, the water exiting the cathode decreased proportionately, this translated into a larger net water transport coefficient. [Janssen, 2001] presented the measured results of the net water drag coefficient for a Nafion 112 membrane under a wide range of operating conditions including temperature, pressure, stoichiometry and current density. The net water transfer through the membrane was measured using a condenser after the cell outlets and it was found that the humidity and the stoichiometry of the inlet gases had a large affect on the water transport. A net water transport study was done by [Yan, 2006] through Nafion 117 membrane at various temperatures and humidification levels. This study measured the effects of concentration difference across the membrane as well as pressure difference across the membrane on the net water flux. They found that the net drag coefficient depended on current density and humidification of feed gases, the EOD ranged from 0.02 to 0.93 H 2 O/H +. They determined the EOD coefficient depending on the operating conditions and it ranged from 1.5 to 2.6 H 2 O. They also showed that the pressure difference across the membrane had less effect on the net water flux than concentration differences and EOD. 11

15 [Husar, 2008] published that the EOD of a Nafion 115 membrane increases significantly with temperature and current density. The EOD increased from 0.25 to 0.4 H 2 O/H + at fuel cell temperatures of 40ºC as the current density was increased from 0.24 to 0.8 A/cm 2. The EOD for a fuel cell temperature of 60ºC the coefficient rose from 0.65 to 1.2 H 2 O/H + as current density increased from 0.35 to 1.0 A/cm 2. This indicates more than a two fold increase in the EOD when the fuel cell temperature is increased from 40ºC to 60ºC. They also indicated a possible increase in EOD with respect to current density. However the results of this study were inconclusive and more research needs to be done. The study also measured water diffusivity through the membrane at different temperatures. The diffusivity increased from 3.5E-07 to 1.2E-6 cm 2 /s as the temperature was increased from 45ºC to 75ºC. They also showed that the water diffusivity across the membrane did not vary at different water vapor concentrations, however were lower than the results published by [Springer, 1991] and [Nguyen, 1993]. In summary, the EOD and diffusion coefficient vary significantly depending on the membrane and operating conditions. Therefore to have an accurate numerical model it is necessary to determine the coefficients in the lab through experimentation on a specific PEMFC. [Zawodzinski, XXXX] nicely layout the general insight into water transports: H 2 O diffusion coefficients, relaxation times and protonic conductivity all increase with increasing water content. H 2 O and H + diffusion coefficients are similar at the lowest water contents but diffusion of protons is greater than the diffusion of water as membrane water content increases. EOD drag coefficients are roughly 1 for membrane water content greater then 14, independent of temperature. For immersed membranes, EOD is roughly 2-3 water molecules per proton at 30ºC and increase with increasing temperature. EOD is similar for many different PFSA and other immersed sulfonate membranes System models Steady-state system models are typically used for component sizing, static trade-off analysis, and cumulative fuel consumption and hybridization studies. These models generally include all the major components the fuel cell system. The models have the potential to be used as tools for the development and improvement of the components and determine the interdependencies between the components in the system. The models can be used to describe the fundamental phenomena that occur in the system to predict performance at different operating conditions and optimize the design and operation of the system [Yao, 2004]. The models in these studies generally characterize each component in the system such as the compressor or blower, heat exchanger, and the fuel cell with performance maps. [Barbir, 1999] presented a model and trade-off analysis of the entire system that calculated the system component parameters for various operating pressure, temperatures and power outputs. [Boettner, 2002] describes a PEMFC system model 12

16 for automotive applications that includes all the major components. The model has been integrated into a vehicle performance simulator that determines fuel economy using two distinct control strategies. The study showed that significant system efficiency improvements may be possible through proper control of the subsystem components. [Kunusch, 2008] developed a dynamic multi-input multi-output model by linear identifying two input variables (stack current and oxygen flow rate) with two output variables (stack voltage and oxygen total pressure). The model can be applied to control design, online analysis, and error detection. 2.3 Fuel cell system operation and control Since the open cathode, forced flow PEMFC system is relatively new, there are few studies that show the effect of operating conditions on the performance. However, there is one interesting study on a forced flow, open-cathode, self-humidified and dead-ended hydrogen fuel cell system. The study shows the effects of fan flow rate, ambient temperature and hydrogen pressure on the performance of the fuel cell. In this system the fan had a dual function, to supply the oxygen for the reaction and to cool the stack. The test showed net power output of the system increased dramatically from 2.0W to 8.9W when the fan was incorporated to the system. The study also shows that at a constant fan voltage the fuel cell performance decreases with the rise in ambient temperatures. However these tests were only done for ambient temperatures between 30ºC and 45ºC and do not mention what the relative humidity was. The main culprit for the deceased performance was associated to membrane dehydration [Santa Rosa, 2007]. So, with proper water management study the performance and range can be improved. Other relevant studies show the role of ambient conditions on the performance of passive fuel cell (free breathing). One early study by [Chu, 1999] showed that the ambient relative humidity affected the fuel cell performance greatly. The minimum ambient %RH that the stack could operate was 30% and as the humidity was increased the performance increased. An interesting part of the study showed an improvement in performance when hydrogen was humidified. This indicates that the performance can be improved by controlling %RH on the anode side. [Fabian, 2006] investigated the relationship between ambient conditions vs. fuel cell resistance with impedance spectroscopy for a passive fuel cell. The study identified three regions of fuel cell operation which were characterized by the membrane hydration level. It is listed in order of increasing current density, which meant increasing cell temperature as: partial membrane hydration, full membrane hydration with gas diffusion layer flooding and membrane dry-out. The critical fuel cell temperature was determined to be 60ºC, were the water removal due to evaporation balances the water generation rate. Another interesting finding was that the ambient conditions impacted all three major voltage loss components (activation, ohmic, concentration). However the resistive losses were most strongly affected by the ambient conditions. 13

17 From the above literature review it is obvious that a comprehensive understanding and modeling are needed for the control of water transfer across the membrane for fuel cell system with minimal parts and controls. The field of fuel cell system development is evolving quite rapidly and there is a movement toward less complex systems especially for portable applications. But for these systems to work robustly, efficiently, stably and be able to operate in a wide range of conditions, active control of water transfer needs to be implemented. Therefore, to take advantage of some of the technologies in passive systems and only add simple and fast acting control mechanisms to extend the operating range and power densities is the direction the technology needs to go. These types of systems like the Ballard Nexa and the Horizon H Series have shown to have good performance; however there are outstanding issues associated to durability that maybe due to load cycling, which has been linked with the corrosion and degradation of the catalyst [Schmittinger, 2008]. Durability issue will not be considered in this study. For the proper design of these simple fast acting control mechanisms, an in depth understanding of water transport through the membrane, water content in the membrane and temperature needs to be studied statically and dynamics. This study will be performed with simple control actions combined in different ways to maintain specific internal variables constant. Model the transport of water through the membrane based on experimental identifications of coefficients. Integrate the water transport model into a comprehensive model of the complete system for future development of controllers. 3. Objective and scope The thesis will study the behavior and the effects of the control actions on the transport and membrane water content for an open cathode, self humidified, hydrogen purged PEMFC system with the following simple control mechanisms: single fan for both cooling and air supply, a hydrogen purge valve, and periodic short circuit of the stack. 3.1 Thesis objective The objective will be to develop guidelines and control objectives to improve performance and stability from insights gained through experimental testing and modeling of the dynamic water transport through the membrane and membrane water content. The specific objectives are: Determine the effect of the operating conditions on the fuel cell performance and stability, specifically focused on water vapor concentrations. o Ambient temperature o Ambient relative humidity 14

18 o Electric load Determine the effect of the three control actions on transport and water content in the membrane, and its consequences on the PEMFC performance and stability, at a wide range of operating conditions (electrical power drawn, ambient conditions). This study will be performed with the three control actions combined in different ways to maintain specific internal variables constant. o Fan velocity o Hydrogen purge frequency o Short circuit frequency Model the transport of water through the membrane based on experimental identifications on the principal coefficients. o Membrane water content o Water diffusion coefficient o Electro-osmotic drag coefficient o Thermal coefficients Integrate the water transport model into a comprehensive model of the complete system for future development of controllers. 3.2 Approach to the thesis Hypothesis Statement: The management of membrane hydration of an open cathode, self-humidified PEMFC system may be improved through fast acting control mechanisms (e.g., through changes in fan velocity, hydrogen purging rates, and stack short circuit frequency). Thesis Phases: The first step will be to determine the water diffusion through the membrane as a function of temperatures and membrane resistance. The second step will be to determine the electro-osmotic drag as a function of temperature, membrane resistance, and current. The third step will be to combine the two functions and complete the model. This model will also take into account the effects of the environmental conditions on the PEMFC system. The fourth step will be to determine the effects of the three different control mechanisms on the transport of water and how they can be combined to work together to control the fuel cell stack resistance. Limits to the thesis: 1. Detailed electrochemical reaction equation will not be contemplated. 15

19 2. Durability or system failure analysis will not be considered. 3. A specific controller will not be designed or implemented 4. Power conditioning will not be developed. This is strongly depended on the application. Outstanding Issues: 1. How does the reduction of the active area due to liquid water buildup on the catalyst affect the membrane resistance? 2. How to take advantage of the H 2 purge valve and the short circuit to generate information on the state of the water in the fuel cell? (mainly location, anode GDL, membrane, cathode GDL) 3. What is the velocity of the water movement in and out of the membrane/gdl (5 layer MEA) 4. Can the 5 layer MEA be split up into 5 discrete volumes? (anode channel, anode GDL, membrane, cathode GDL, cathode channel) 5. Can the trend of a given measurement be implemented in the control strategy? 4. Methodology The approach that will be taken to understand the movement of water across the membrane in the fuel cell stack will be done in parallel with experimentation in conjunction with numerical simulation. Initially, the water vapor diffusion coefficient, the electro-osmotic drag coefficient, and the heat transfer properties (heat transfer coefficient and effective conductivity) will be experimentally determined. This data will define the unknown variables in the model. The next step will be to determine the effects of the three control mechanisms, which are hydrogen purge valve operational frequency, fan velocity and stack short circuit frequency on the operation and the mass transport of water across the membrane. The experimental layout, the model description and finally the various experiments to be performed are explained in this section. First the fuel cell test station will be described then the numerical model and finally the various experiments will be explained. 4.1 Fuel cell test station The test station will be designed to independently control, simulate and monitor the various phenomena that affect the fuel cell system performance. The system can be either operated with the manufacture supplied controller (see Figure 4.1) or with an internally developed LabView control scheme. The main focus will be on the transport of water across the membrane. This will be a unique test station because of the simple nature of the fuel cell system. The simplicity comes from its open cathode and self-humidified design, however this simplicity comes with a cost, the system will be relatively sensitive to the surrounding environmental conditions. Therefore the fuel cell will be tested in an environmental chamber that will have the capability to control and measure the air temperature, relative humidity, and oxygen concentration. The environmental chamber will have a dual purpose. The first will be 16

20 to maintain very specific inlet conditions for the air delivered to the stack to isolate the water transport phenomena through the membrane, and second will be to emulate a wide range of environmental conditions for the purpose of investigating the combined effects of the fuel cell systems thermal and mass balance with respect to performance and the control mechanisms. In the next two subsections the fuel cell test station and the numerical model are explained. Figure 4.1- Basic schematic of the PEMFC system that will be used with manufacturer supplied controller Fuel cell test station design and specifications The main subsystems of the test station can be split in five different subsystems; the environmental chamber, the cathode, the anode, the fuel cell stack, and the data acquisition/control sub-systems as shown in Figure

21 Figure Schematic of the fuel cell test station with the environmental chamber A custom designed environmental chamber will be used, which will have the capability of controlling the chamber temperatures from 5ºC to 70ºC, relative humidity s from 10%-100% and controlling the oxygen concentration from 19% to 25% per volume. The environmental chamber is equipped with all the necessary safety systems for use with a flammable gas (explosion proof electronics with a explosion pressure relief system, hydrogen sensor, nitrogen purge and over temperature shut-off). The only active component in the cathode sub-system will be a fan; the voltage applied to the fan will set the air flow rate entering the stack. The fan s purpose will be to control the temperature of the stack because the air required for the reaction is relative small compared to what is needed for cooling. Inlet and outlet temperatures and dew point sensors will be installed for the calculation of mass and thermal balances. The fan flow rate will be verified with a air velocity meter, a hall effect current sensor, and a motor angular velocity sensor. These sensors will supply the data necessary to model and verify the fan dynamics and power drawn. The anode sub-system has two modes of operation to control the hydrogen flow rate: control the flow rate of the hydrogen with a mass flow controller or controlling the flow of hydrogen by controlling the inlet pressure in conjunction with a purge valve. In the first mode of operation, which will be used for identifying the water transport coefficients, the mass flow controller will be set in the control mode and in combination with a humidifier to control the inlet dew point temperature. In this mode both the forward pressure regulator and the hydrogen purge valve will be by- 18

22 passed. In the second mode of operation the mass flow controller will be set to measure the flow. The forward pressure regulator will maintain the pressure in the anode while the purge valve is closed. So as hydrogen is consumed the pressure will decrease and the regulator will open to refill the system. There will be temperature and dew point sensors at both the inlet and outlet to calculate the mass and thermal balances. Gas line heaters in combination with the dew point sensor will ensure that all the water is vaporized for accurate measurements. The commercially available Horizon H-100 fuel cell stack will be used (see Figure 4.3). The stack is specifically designed to operate at ambient relative humidity with an open cathode and a periodic purge of the anode gas. The stack configuration is as follows: 21 cells with an active area of 19.2 cm 2, with a nominal power output of 100 Watts, at 12 Volts and 8.5 Amps. The stack will be equipped with thermocouples to measure the individual cell temperature as well as individual cell voltages. A manufacturer installed thermocouple measures the internal stack temperature. An electrochemical impedance spectroscopy (EIS) system analyzer will be wired to the stack and cell voltages which will allow to measure the resistance of either the stack or individual cell voltages. Figure cell PEMFC stack with integrated fan for cooling and air for reaction, 100 Watts nominal power The controller that is supplied with the fuel cell will be used to generate the baseline performance and efficiency data. The end results will be compared to the data using this controller. The controller has several different functions: starts up the system, adjusts the fan voltage based on the power drawn from the cell, opens the hydrogen purge valve every 10 seconds, and short circuits the stack every 10 seconds out of phase with the purge valve. The safety features include over temperature shut down at 65ºC, low stack voltage shutoff at 11 volts, high current shutoff at 20 amps and hydrogen inlet shut-off valve. The system shuts down by closing the inlet hydrogen valve and disconnecting the electric load. The controller receives all its power from a independent 13 volt power supply which provides the power for all the electronics (fan, H 2 purge valve, H 2 inlet valve, control electronics). This means that the systems parasitic losses are not integrated into the fuel cell system. 19

23 The baseline will be established, with the manufacturer s fuel cell system controller, in all the experiments the following measurements will be made: membrane resistance, individual and stack temperatures, inlet and outlet relative humilities, fan angular velocity, air inlet velocity, current, cell and stack voltage and hydrogen flow rate. This data will provide a measure (an index) of the efficiency, stability, and performance baseline for the system. The current data indicates that the control mechanisms for the fuel cell can be greatly improved. The fuel cell system efficiency will be calculated by using cell voltage, hydrogen consumption and parasitic losses. The stability will be calculated by the variation from the mean voltage, and the performance will be calculated using the average voltage and current to create a polarization curve of the fuel cell. These will be the three main indicators of improvement in the system. Figures 4.4 and 4.5 show the initial current step experiment and the polarization curve for the H-100 fuel cell system using in manufacturer s supplied controller. Figure Voltage and temperature response to different currents for a constant fan speed 20

24 Figure Polarization cure for the H-100 using the manufacturer s supplied controller. The data uses the average voltage and current from Fig The data acquisition and in-house control system will be built in the LabView programming language and will allow for the measurement and control of all the fuel cell sub-systems. Figure 4.6 shows the basic layout of the fuel cell and all the sensors inside the environmental chamber. 21

25 Figure 4.6-3D depiction of the experimental setup 4.2 Description of experiments This first section will describe how the model will be calibrated based on the initial experiments for the purpose of applying them to the second section where the dynamics of the system will be defined based on the three control mechanisms Water transport experimental characterization There are two dominant transport mechanisms for water vapor across the membrane: diffusion which is driven by difference in water vapor partial pressure and electroosmotic drag which is driven by protons traversing the membrane. As discussed previously in the hypothesis section the relationship between stack temperature, membrane resistance and current will be examined. The results from these to sets of experiments will allow for the completion of the 5 layer MEA water balance submodel Diffusion coefficient experimental design The effective water vapor diffusion coefficient and its dynamic evolution through the membranes in the stack under open circuit voltage (zero load) conditions will be 22

26 determined by isolating this water transfer phenomenon across the 5 layer MEA (which include the two gas diffusion layers the two catalysts and the membrane) in relation to stack temperature and membrane resistance. The experimental model is laid out in Figure 4.7. Figure Diffusion coefficient experimental diagram This coefficient is dependent on the material properties of the components in the fuel cell (gas diffusion layer porosity, membrane thickness, and exc.). The dynamics of temperature and water content will be established experimentally. The controlled variables in this experiment are the water vapor partial pressures at the inlets and the temperature of the fuel cell stack. Experimental procedure is to independently vary the water vapor partial pressures and fuel cell temperature and measure the water vapor partial pressure at the outlets. The environmental chamber indirectly controls the fuel cell temperature as well as the inlet dew point temperature and air temperature. The anode reactant inlet water vapor partial pressure is maintained by a membrane based humidifier which is controlled with a dew point sensor. The measured variables will be the anode and cathode outlet dew point temperatures and the membrane resistance. The dew point temperatures will be measured using a humidity sensors and the membrane resistance will be measured using a continuous high frequency EIS. The experiment will provide insights in the variability of the effective diffusion coefficient under a variety of conditions, temperatures from 10ºC to 60ºC and relative humidity s from 10% to 100%. 23

27 This experimental setup will be used to evaluate the diffusion coefficient variability with respect to membrane water content which is directly related to membrane resistance and stack temperature. This data then will be analyzed and incorporated into the model Electro-osmotic drag coefficient experimental design In general the electro-osmotic drag is considered dependent on temperature and water content in the membrane. The proposed experiment will examine the effects of temperature, current density, and water content on the coefficient while isolating it from diffusion. Electro-osmotic drag will be affected by the catalyst layers since a large volume fraction of the catalyst layer is ionomer. This means that electro-osmotic drag will also occur in the ionomer portion of the catalyst layer. To accurately calculate water diffusion transfer through the membrane, one must use the water contents of the membrane at the anode and cathode sides as boundary conditions. Yet covered with the catalyst layer and the GDL, it is almost impossible to know the water contents at the membrane boundaries. Supersaturated hydrogen will be supplied to the stack to ensure saturation throughout the anode flow field. The cathode reactant will be saturated as well. This will minimize the transfer of water due to diffusion. Thus, the water that is transferred from the anode to the cathode should depict the water transfer due to electroosmotic drag. The stack temperature will be maintained to be equal to the temperature in the environmental chamber with the fan so there will be a limitation to the amount of current the can be drawn from the stack do to the temperature gradient that will be produced. Another restraint for the test will be to ensure that both reactant streams leave the stack fully saturated relative to the stack temperature. This will indicate that there is minimal water transfer due to diffusion. The intended ranges of conditions that will be tested are temperatures from 10ºC to 60ºC and currents from 1 to 8 amps all with full saturated flows. Measuring the humidity at the exit streams will then allow for the calculation of water transfer across the membrane due only to electro-osmotic drag. The measurement of the membrane resistance will give an indication of the water content. The goal will be to main the water constant fully saturated at all the temperature and current density to get a map out of the changes in the electro-osmotic drag coefficient due to the previously stated variables. An equation will be fitted to this data that will be a function of temperature, current density, and water content which will be incorporated into the system model. 24

28 Heat transfer constants The coefficient for heat transfer through the fuel cell will be split into two coefficients: heat transfer coefficient, and the effective thermal conductivity coefficient. Both will be experimentally examined and an introduced into the model. The convective heat transfer coefficient will be experimentally determined between the air, hydrogen and the stack Effect of the 3 control mechanisms The 3 control mechanisms of the simple PEMFC system that will be examined are the fan velocity, hydrogen purge valve frequency and stack short circuit frequency. The purpose is to understand how these mechanisms affect the transport of water across the membrane at various operating conditions. There is an indication from the preliminary data that the effects that the hydrogen purges and the short circuits will vary depending on the current and temperature. Figure 4.8 indicates the average change in the voltage for a hydrogen purge and a short circuit done every 10 seconds verses the current. It indicates that the hydrogen purge has a negative effect on the voltage at low current and increasingly greater effect as the current increases. On the other hand the change in the stack voltage due to the short circuit tends to increase slightly with increasing current. The temperature effect has not been incorporated in this data and should be significant. 25

29 Figure 4.8 Indicates the average change in the stack voltage due to a hydrogen purge or short circuit Fan velocity effect The fan velocity will be varied at different current densities, ambient relative humidity and ambient temperatures in order to achieve different stack temperatures. Varying the above conditions will change the partial pressure of water on the cathode side and thus will effect the transport of water. To study these effects the inlet and outlet humidity s will be measured. The anode side flow rate and relative humidity will be kept constant. The membrane resistance will be measured to relate it to the membrane water content. Secondary effects on the cathode outlet due to the change in fan velocity should be: change the air flow rate through the stack for both cooling and oxygen for the reaction. So with a change in the air flow rate on the cathode side there will be a change in the average oxygen concentration, average water vapor partial pressure. A change in the fan voltage also will change the fan power consumption and thus the system efficiency. Secondary effects on the anode outlet: The change in temperature will change the maximum water vapor partial pressure The duty cycle and period of the hydrogen purge valve Opening the hydrogen purge valve flushes the anode flow field. This removes both liquid and water vapor from the anode side. This reduces the partial pressure of water for a time being and increases the hydrogen concentration. This control action should temporally drive water from the cathode to the anode. A secondary effect is the removal any other impurities such as nitrogen that has diffused from the cathode. The frequency of the purges affects the system efficiency due to hydrogen lost to the environment Stack short circuit duty cycle and period Short circuiting the stack for a very short period of time (in the order of 40 milliseconds) consumes all the hydrogen and oxygen in and around the catalyst layers and produce heat and water on the cathode catalyst. This heat and water will be distributed to both anode and cathode flow fields depending on there temperature and water vapor partial pressure states. The frequency of the short circuit affects the system efficiency due to the hydrogen that is consumed. A dynamic full system model and a fuel cell test station will be constructed to accomplish these experiments. 4.3 Fuel cell system numerical model A lumped parameter dynamic model, based on structure of [Pukrushpan, 2003], [McKay, 2004] and [McKay, 2008], will constructed and adapted to the current system. 26

30 Additional sub-models will be developed and integrated to model in order to simulate other important dynamic behaviors that affect the system performance (i.e. membrane water content and transport, the stack thermal system). The proposed model will describe the mass and energy balance in the fuel cell system. Specifically the developed model will be used to simulate and study the effects of the available control mechanisms on the above mentioned dynamic behaviors, and consequently on the system performance. The model construction will be carried out following a standard methodology of decomposition of the model into its different sub-systems listed below. Cathode fan and flow field Anode forward pressure regulator, flow field, and purge valve Membrane water content and transport. Thermal model of stack. Figure 4.9 depicts the block diagram of the model indicating the various subsystems. Figure 4.9 Model block diagram of the various subsystem 27

31 5. Expected results 5.1 Hypothesis on the effects of the different control mechanisms: The transport of water through the membrane occurs mainly by diffusion and electroosmotic drag. Therefore by controlling the concentration of water vapor and the additional creation of water, membrane hydration and fuel cell "flooding" can be managed. The following are the available mechanisms to influence the above variables. Air flow rate of the fan: Effects on the cathode: The fan air flow rate affects the water vapor and oxygen concentration on the cathode and heat removal rate of the stack. The water vapor concentration and internal fuel cell temperature are intrinsically connected variables that affect each other. The air flow rate also affects the system efficiency as parasitic losses. Effects on the anode: Changing the air flow rate affects the fuel cell temperature which in turn modifies the gas temperature and saturation partial pressure of water on the anode side. The duty cycle and period of the hydrogen purge valve: The hydrogen purge removes both liquid and water vapor from the anode side which temporary reduces the water vapor concentration and increases the hydrogen concentration. This should drive water from the cathode to the anode. The purge also removes any other impurities such as nitrogen that has diffused from the cathode. The frequency of the purges affects the system efficiency due to hydrogen lost to the environment. Stack short circuit duty cycle and period: Short circuiting the stack for a very short period of time (in the order of 40 milliseconds) consumes all the hydrogen and oxygen in and around the catalyst layers and produce heat and water on the cathode catalyst. This heat and water will be distributed to both anode and cathode flow fields depending on their temperature and water vapor concentrations. The frequency of the short circuit affects the system efficiency due to the hydrogen that is consumed. The system model will allow for the investigation of how water vapor concentration is interconnected to the different variables and control actuations. 28

32 5.2 Contribution This analysis will then lead to an in depth understanding of how a system with minimal parts and complexity can be improved in terms of overall efficiency, stability and operating rage. The conclusions from this work will also show how the system reacts to changes in the environmental conditions and thus will enable to develop a proper control strategy with the necessary sensors to ensure that the system operates properly under a wide range of environmental conditions. Propose an experimental methodology to identify key physical parameter for an open cathode PEMFC using an environmental chamber Propose a control oriented model of an open cathode PEMFC system Propose control guidelines to improve performance and stability of an open cathode PEMFC 6. Work plan/timetable 1. Literature Review 2. Define project objectives 3. Design and purchase equipment for test station 4. Layout and construct model 5. Design experimental procedures 6. Test station construction 7. Run experiments 8. Data analysis 9. Final model validation and conclusions 10. Thesis redaction 7. Required resources Table 6.1 The work plan timetable for the thesis The funding for this thesis is provided by MEC under contract number DPI C02-01 and DPI The funding for student PhD. program is provided by CSIC I3P scholarship. 29