Application of metal foams as flow field in medium scale polymer electrolyte membrane fuel cells

This thesis presents a research focused on investigating the feasibility of using metal foams (MFs) in cathode flow fields of Polymer Electrolyte Membrane Fuel Cells (PEMFCs), especially open cathode PEMFCs. Increasing growth in world’s energy demand, political unrest, as well as concern of greenhouse gas effects have led the world to search for alternate sustainable energy solutions. Fuel Cell technology is considered clean and sustainable as an alternative to fossil fuel-based energy systems for both portable/mobile and stationary systems, if hydrogen is sourced using renewables. Despite enormous improvements in materials and components in the last decades there are still some aspects of PEMFCs that need attention. For example, temperature disribution, water and thermal management, air/oxygen supply and distribution, water management, hotspot formations, water accumulations, weight and volume of the cells, etc. are general issues that become more challenging in open cathode PEMFCs. MFs are engineered porous structures that possess excellent thermo-electrical and structural properties. Moreover, MFs are lighter in weight as well as highly porous as a material that gives it the privilege of having high specific surface area. Because of these special features, it was hypothesised that MFs can help enhance the performance of PEMFCs by addressing some of above-mentioned challenges.<br><br>For a better performance, a PEMFC needs uniform oxygen supply, uniform temperature distribution, proper stack assembly, and reduced parasitic power. All the above-mentioned parameters are linked to gas flow fields and gas diffusion layer (GDL) in a PEMFC. Unfortunately, conventional channel (Ch) flow fields inherent gradients in oxygen supply and temperature along gas flow direction, which reduce closed cathode PEMFC performance significantly, unless special care is taken. In open cathode PEMFCs, cathode gas (air) flow rate is usually higher than closed cathode PEMFCs that can contribute to drying out or flooding of membrane electrode assembly (MEA), the heart of a PEMFC. Secondly, high flow rate of air does not compensate loss in oxygen diffusion rate due to low air pressure compared to closed cathode PEMFCs. Thirdly, water accumulation and blockage caused by the water in the cathode flow field results into non-uniform gas distribution as well as thermal hot spots. Incorporating air-cooling system using common air as both coolant and cathode gas in the same direction through parallel channel flow fields in open cathode PEMFCs increases possibility of drying out of MEA. In the absence of a proper thermal management system, both nullifying conditions such as flooding and drying of the MEA could happen due to low operating temperature and low specific heat of air.<br><br>By using MF as flow field in PEMFC, this important matter should be taken into consideration that a MF based flow fields in a PEMFC of medium scale are confined flow passages of high aspect ratio. However, existing theories/models on pressure drop and convection heat transfer are for bulk MFs. Moreover, experimental data on the topics for confined MF flow passages is scarcely available. Therefore, mathematical and experimental investigations on pressure drop and convection heat transfer through confined MF flow passages were required as pre-requisites to continue the investigations on PEMFCs.<br><br>This thesis seeks to contribute to surmounting the aforementioned issues in PEMFCs by investigating the effects of MF application in cathode gas flow fields of medium scale single celled PEMFCs. This study was conducted for low operating temperature of PEMFCs, i.e. below 50 °C, to be able to observe any positive effect of MF over Ch flow fields. It also tries to contribute to understanding flow and heat transfer characteristics through confined flow passages filled with MFs, which is helpful in designing heat sinks for different applications, such as PEMFCs, electronics, etc. Hence, the objectives followed by this thesis are to:<br><br>1. understand the physics and governing equations describing the thermo-electrical performance of PEMFCs, and pressure drop and heat transfer in MFs, through conducting a comprehensive study on the available literature;<br><br>2. study, both mathematically and experimentally, the pressure drop along the airflow direction as well as convection heat transfer in confined MF and parallel channel (Ch) cathode when used as flow fields in PEMFCs;<br><br>3. experimentally investigate and quantify the benefits and challenges associated with using MF and parallel channel as cathode flow fields in PEMFCsby focusing on electrical performance of both open and closed cathode PEMFCs;<br><br>4. make recommendations and introduce a roadmap for further research and development in using MFs as gas flow field in PEMFCs. <br><br>Based on these objectives the following research questions have been answered by combining the outcomes of both theoretical and experimental investigations:<br><br>1. What are the challenges associated with conventional airflow channels in PEMFCs and what potentials MFs can offer to address them to enhance the thermal and electrical performance of these fuel cells?<br><br>2. How does a confined MF flow passage affect the pressure drop of air through it in the context of a PEMFC?<br><br>3. How do the MFs affect the overall heat transfer performance and pressure drop compared to conventional parallel channel flow fields?<br><br>4. How does performance of PEMFCs can be affected by using MFs as cathode flow fields? <br><br>In the process of fulfilling the objectives and answering these questions, the research has led to the following original outcomes and contributions to the body of knowledge:<br><br>• The fluid flow (air) and convective heat transfer properties of confined MF samples were characterised experimentally<br><br>Rectangular confined flow passages have aspect ratio of around 50 (e.g. width of 150 mm and height of 3 mm) filled with MF. Four MF samples such as, uncompressed 20 ppi (MF- 1) and 40 ppi (MF-2) aluminium MFs with 88% porosity (in both samples), and compressed 20 ppi (MF-3) and 40 ppi (MF-4) MFs with 83% and 86% porosities respectively were investigated experimentally for pressure drop in the case of using dry airflow through the passage.<br><br>At the beginning of this investigation a well-known flow model of Fourie-Plessis RUC model was used to estimate the pressure drop of airflow through the uncompressed MFs. The model was selected due to its simplistic approach and direct dependency on morphological parameters of MFs that could be measured directly. Upon comparison with the experimentally obtained values the model was found to be overestimating pressure drop gradient through the MF samples by 20% to more than 100% for an airflow velocity range of 0.4-6 ms-1. Further investigation on the behaviour of the RUC model revealed that the overestimation was solely due to long cylinder assumption in drag coefficient relation. This finding was taken as an opportunity to propose a correlation for the drag coefficient as a correction to the RUC model. The proposed modification enabled the RUC model to estimate pressure drop gradient with a good accuracy. Difference between experimental and estimated pressure drop values for MF-1 was ±5% for velocity up to 3 ms-1; however, it increased up to 22% for velocity up to 6 ms-1. On the other hand, for MF-2, the model over-estimated the pressure drop value from 0-20% in the airflow velocity range of 0.4-6 ms-1. <br><br>Encasing walls (especially between shorter side) are the sources of additional drag. Two extreme conditions can appear when the walls are connected with vertical and horizontal ligaments. The drag value will be large if most of the ligaments connect the walls vertically, and it will be small if most of the ligaments are horizontal. In the case of MF, ligaments are mostly randomly inclined due to its geometry. Orientations of ligaments of MF-1 sample were observed using its scanned 3D model and it was found that inclined ligaments at different depths covered more than 70% of planar area. Further investigation on the blockage effect revealed that the critical gap value in both MF-1 and MF-2 were almost half of their ligament’s lengths, which shows the possibility of flow stream reattachment. However, due to random appearance of the cells in MFs, the re-attachment hinders. Therefore, based on above discussion, the effect of encasing walls on pressure drop was considered negligible for the flow passage filled with either MF-1 or MF-2.<br><br>Temperature distribution results from the PEMFC simulating heat transfer experiments confirmed that flow can thermally develop within finite and short (25% of flow passage length in our case) distance, and thermal entry length decreased with an increase in heat flux at steady state condition. Temperature difference between MF surface and air was found smaller than the difference in the case of the Ch flow passage configuration. It indicates a better heat transfer from MF to air in comparison with Ch to air. A heat transfer surface with high heat flux relative to friction-power expenditure is considered as a highperformance surface. Based on this definition the performances followed a sequence such as, MF −1 > CH > MF − 2 > MF − 3 > MF − 4 for any value of Re numbers. However, MF-3 and MF-4 were in very close proximity.<br><br>• A guideline outlining the characteristics of potentially suitable MFs as gas flow field in PEMFCs was developed<br><br>Application-specific performance characterisation was required to find suitable MFs for use as gas flow fields in PEMFCs. Heat transfer performance curves provided very general characteristics of the heat transfer surfaces and could not be directly implemented for designing the flow field. This is because the MFs do not share common characteristic structural details. Hence, a method has been developed to compare the cooling capacities of the surfaces for equal flow passage size. This method compares cooling capacity of the heat transfer surfaces and required coolant flow rate with friction-power expenditure. It was observed that similar cooling capacities exhibited by MF-1, MF-2 and Ch flow passages within friction-power expenditure range of 3-10% based on the abovementioned method and PEMFC simulated experimental data. The cooling capacity versus frictionpower expenditure ratio curves intersected within the abovementioned range. MF-2 showed the highest cooling capacity followed by CH and MF-1 above the range. On the other hand, MF-1 showed the highest cooling capacity below the intersection point followed by CH, MF-2, MF-4 and MF-3. MF-2, MF-3 and MF-4 provided cooling capacity of approximately 0.8 W (35.6 W/m2 based on active area) and 2.6 W (115.6 W/m2 based on active area) at 0.01% and 0.1% of friction-power expenditure respectively, which was 50% of cooling capacities of MF-1 and CH at the friction-power expenditure values.<br><br>An increase in specific surface area of MF is preferred but at no significant increase in friction-power expenditure. Material densities of all the MF samples were similar to each other. However, ligament diameters and pore densities were different. MF-1 and MF-2 were uncompressed; however MF-3 and MF-4 were compressed, and their parent MFs were of 6% relative density. The compression was done to achieve higher specific surface area. Specific surface area of MF-4 was 1.33 times higher than MF-3 but the showed similar cooling capacity. Furthermore, at given friction-power expenditure, MF-1 and MF- 2 performed better compared to MF-3 and MF-4. It should be noted that optimisation of ligament diameter and pore density was out of scope of this work. However, the outcomes pointed out its direct impact in heat transfer performance, especially in the case of PEMFC application. Therefore, the author recommends on optimising the physical characteristics of MFs in such applications for future research on this topic.<br><br>• Advantages and limitations associated with confined MFs as gas flow field in PEMFCs were quantified<br><br>Electrical performance of the closed cathode PEMFC experimentally studied increased by ~14% using MF cathode flow field compared to Ch cathode flow field when oxygen was cathode gas. Temperature gradient was observed ~15 times higher in the Ch PEMFC compared to the MF PEMFC. This performance was used as reference in comparing the performances of both open and closed cathode PEMFCs with air as cathode gas.<br><br>With air as cathode gas, the closed cathode PEMFCs performance was lower than the reference case, but the trend of the polarisation curves were similar, i.e. MF PEMFC showed better performance than the Ch PEMFC. On the other hand, the open cathode PEMFCs showed different behaviour. Upon comparison with the reference case, it was observed that the Ch PEMFC performed better than the reference case with Ch flow field above total current of 20 A; whereas, the performance of the MF PEMFC was lower than the reference case with MF flow field, but higher than the closed cathode MF PEMFC.<br><br>Due to design configurations, maximum surface contact area between GDL and flow field was in the closed cathode Ch PEMFC, with 52% of the active area followed by the closed cathode MF PEMFC with 37% of active area. In the open cathode PEMFCs, the surface contact area was around 30% of the active area. Liquid water was observed at the outlets of all the PEMFCs. However, the amount of liquid water in GDL in the closed cathode PEMFC with Ch flow field must be higher than the open cathode PEMFCs due to the largest surface contact area between GDL and Ch flow field. Moreover, there were strong possibility of flow passage blockage by water and slug flow of water in the absence of water management system and low flow velocity. Both of the above reasonings result with non-uniform oxygen distribution to cathodes, especially in the closed cathode Ch PEMFC.<br><br>Further analysis of the V-I curves for the MF-cathode PEMFCs revealed that the effect of MF flow field on the electrical performances of the cells (both types) were similar. The trend of Nyquist plots and V-I curves illustrate that water removal from the open cathode Ch PEMFC was the best among all the cells. Better performance of closed cathode MF PEMFC was due to smaller surface contact area between the flow field and the GDL. At the given air stoichiometry, MF did not show effective water removal for both types of the PEMFCs.<br><br>During the tenure of this research, some opportunities for further expansion of the current study were identified as follows:<br><br>1) It was observed in this study that MF increases oxygen diffusion but water removal from it is poor at low range of operating temperature (i.e.25-40 °C). The low operating temperature was purposely considered to study MF application in worse-case-scenario. However, physics of the PEMFC at higher range of operating temperature is different, thus the fuel cell tests can be conducted at higher range of temperature (i.e. above 40 °C to 80 °C) to study the validity of the results obtained in this work.<br><br>2) Effects of corrosion on the cell over-potential as well as change in mass transfer resistance due to metal foam flow passage can be investigated with the help of proper tool for electrochemical impedance spectroscopy data.<br><br>3) Cathode flow fields can be treated with hydrophobic/hydrophilic (based on the experimental design) treatments to reduce water withholding capacity of both metal foam and parallel channel cathode flow field configuration.<br><br>4) Effect of metal foam on water accumulation in gas diffusion layer and electrode can be investigated with the help of neutron imaging tool.<br><br>5) The third experiment can be repeated with different types and specifications of gas diffusion layer, electrode, and membrane. It will lead to optimise the fuel cell components when MFs are used as gas flow field.<br><br>6) Humidity certainly plays an important role in the cell performance and it is worthy of a detailed experimental investigation to understand how it can affect the findings of this study.<br><br>7) Heat transfer/exchanger performance of other commercially available metal foams can be studied experimentally as compact heat transfer surfaces to enrich data bank on the topic and generalise the findings<br><br>8) An investigation on realistic mathematical model on pressure drop as well as three dimensional flow components through open pore metal foams of any specification can provide enormous advantage in designing different industrial components such as, catalyst support, flow fields in PEMFCs, etc. <br>