Experimental and numerical investigation of a PEMFC single cell and stack analysis

1 EXPERIMENTAL AND NUMERICAL INVESTIGATON OF A PEM FUEL CELL. SINGLE CELL AND STACK ANALYSIS Alfredo Iranzo*a, Felipe Rosab, Eduardo Lópezb, Fco Javier Pinoc, Luis Valverdea, Pablo Bermejoa aAICIA, ESI, Camino Descubrimientos s/n, 41092 Sevilla, Spain, [email protected] bINTA, 21130 Mazagón (Huelva), Spain, [email protected], lo[email protected] cEscuela Sup. Ingenieros, Camino Descubrimientos s/n, 41092 Sevilla, Spain, [email protected] *Corresponding author: Tel. (+34) 954 487471. Fax. (+34) 954 487247 ABSTRACT The present paper describes the work being done by INTA and University of Seville – AICIA in the framework of a joint research project in the field of Proton Exchange Membrane Fuel Cells (PEMFC). The main objective of the project is to develop a novel methodology for the characterization and investigation of fuel cell stacks, that combines experimental and Computational Fluid Dynamics (CFD) analysis. A CFD model based on the PEMFC Module implemented in FLUENT software is being developed for a single cell, and the polarization curve obtained is being validated against experimental results obtained at INTA facilities. At a second stage of the investigation, a method for the extrapolation of CFD single cell results to the complete stack is developed. The extrapolation method is being validated against experimental results obtained for the fuel cell stack with different numbers of cells. This paper describes the results of the early stages of the investigation and presents the methodology developed for the project. The developed model would allow researchers to use CFD for the complete fuel cell stack with reasonable computing facilities. 2 KEYWORDS Fuel Cell, Simulation Tool, Stack, Computational Fluid Dynamics. 1. INTRODUCTION PEM Fuel Cell models currently being applied and developed by designers and researchers can be considered as multi-dimensional and one-dimensional models [1]. Multi-dimensional models are based on CFD to solve governing equations, and provide high resolution results of flow characteristics of reactants, by-products, and charge transport. This is particularly advantageous for the analysis of a single cell given the detailed information that is provided by the model, but is limited for a stack because of the enormous computational time that would be required. These models are therefore applied to investigate parts of complex domains such as flow fields of a single cell and to predict the performance of a single cell. By contrast, one-dimensional models are macroscopic models that describe layers in a cell. These simplified models do not provide detailed analytical mechanisms of a cell, but allow representations of a multi-cell stack, and therefore these models are preferred for investigating the complete system. However, as models are based on a onedimensional approach, results must be considered with care as they may involve an important lack of accuracy. The main objective of the research work presented in this paper is to develop a novel method for the characterization and investigation of fuel cell stacks. The method is based on both experimental and Computational Fluid Dynamics analysis. The strategy to accomplish this work is being presented in this paper. It must be noticed that this 3 paper describes the results of the early stages of the investigation, and mainly presents the methodology developed for the project. 2. RESEARCH STRATEGY AND METHODOLOGY The model being developed is aimed to the analysis of any type of PEMFC, and for the initial development and validation the work is being performed on two different stacks. The stacks and cells being modelled correspond to commercially available models from ElectroChem Inc. Each stack consists of seven MEAs (Membrane Electrode Assembly) with Nafion-117 membranes. The MEAs are manufactured based on CCM (Catalyst Coated Membrane) technology. The main technical specifications are detailed in Table 1. Table 1 Technical specifications of the fuel cell stack from ElectroChem Inc. Specifications Average cell Voltage 0.65 +/- 0.05 V Current Density 350 +/- 50 mA cm-2 Total Power 180-280 mW cm-2 Nominal power output 50 W Dimensions 13.5 x 13.5cm x 13.5 cm Net weight 4 kg Electrode Assembly 50 cm2, 7 MEAs Type Solid Polymer Electrode A CFD model based on the PEMFC Module from FLUENT software is being developed for a single cell of each stack. The polarization curves for the single cells are obtained by running several simulations with different boundary conditions. The current 4 intensity-voltage curves obtained are being validated against experimental results obtained at INTA facilities, thus allowing for a detailed comparison between numerical and experimental results. Secondly, an extrapolation method is being developed for the extrapolation of CFD single cell results to the complete stack. This would allow researchers to predict the complete fuel cell stack performance from CFD results of a single with cell, thus with reasonable computing facilities. In order to develop and validate the extrapolation method, experimental results are obtained for the fuel cell stacks with different numbers of cells. The results are expected to allow for the development of an accurate method for the extrapolation of CFD single cell results to the complete stack. 3. STACK MODEL DEVELOPMENT The different research milestones to be accomplished during the research work are described in the following sections. 3.1. Single cell model development and calculation 3.1.1. Model description A CFD steady state model for a single cell based on the PEMFC Module from FLUENT software is being developed. The model is solving not only the reactant gases flow and diffusion, but also the electrochemistry and current continuity, the liquid water generation and water transport, the membrane humidification, electric losses, heat transfer and other relevant phenomena. The equations used in the model are described in 5 [2]. The PEMFC model is based on the correlations proposed by Springer et al. [3], developed for Nafion-117 membranes. 3.1.2. Geometry and material properties The geometry of the Bipolar Plates has been obtained from ElectroChem Inc., and the physical properties necessary for the model have been obtained from ElectroChem Inc., DuPont Technical Data Sheets (for Nafion-117), SGL Group (GDL provider) and from the literature [4,5,6,7]. A summary is listed in Table 2. There are still unknown values that will have to be obtained from the literature or experimentally. In a first approach, two different Bipolar Plates geometries are being analysed: a straight flow and a serpentine flow pattern, as shown in Figure 1. Table 2 Physical Properties of the fuel cell materials. Data Units Value Source Material Properties Bipolar Plate Material: Isotropic graphite ElectroChem Density kg m-3 1990 ElectroChem Specific Heat Capacity J kg-1K-1 710 [4] Thermal Conductivity W m-1K-1 120 ElectroChem Electric Conductivity ohm m 1.08E-05 ElectroChem Gas Diffusion Layer Density kg m-3 320 SGL Group Specific Heat Capacity J kg-1K-1 * Thermal Conductivity W m-1K-1 0.3 [5,7] Electric Conductivity ohm cm2 0.015 SGL Group 6 Porosity - 0.82 SGL Group Viscous Resistance m-2 1.0E+09 [5] Contact Angle deg * Thickness mm 0.42 SGL Group Catalytic Layer Density kg m-3 * Specific Heat Capacity J kg-1K-1 * Thermal Conductivity W m-1K-1 * Electric Conductivity ohm m * Porosity - * Viscous Resistance m-2 * Surface-to-Volume ratio m-1 * Contact Angle deg * Thickness mm * Membrane Density kg m-3 1980 DuPont Specific Heat Capacity J kg-1K-1 2000 Thermal Conductivity W m-1K-1 0.13 [7] Electric Conductivity ohm m Model [3] Equivalent Weight kg kmol-1 1100 [6] Thickness mm 0.175 DuPont PEM Model Parameters H2 Species Diffusivity m2 s-1 3.0E-05 [4] O2 Species Diffusivity m2 s-1 6.0E-05 [4] H2O Species Diffusivity m2 s-1 3.0E-05 [4] Pore Blockage Saturation Exponent - * * denotes unknown values. 7 Figure 1: Geometry of the Bipolar Plates: serpentine flow pattern (left), straight channel flow pattern (right). 3.1.3. Mesh generation Previously to the work presented in this paper, a different cell with a different MEA and Bipolar Plate have been studied. The MEA was a Celtec®-P1000 from BASF-PEMEAS, which is based on high temperature polymer polybenzimidazole (PBI) and phosphoric acid for a high temperature operation (120ºC-180ºC). However, the models of the FLUENT PEMFC Module are based on correlations derived for Nafion-117 membranes [2,3], and therefore the software cannot be directly applied to the resolution of cells with PBI membranes. However, the simulations performed with this cell were used to assess the meshing requirements needed for successful CFD simulations of fuel cells. A mesh sensitivity study has been performed in order to ensure that CFD results are independent from mesh spatial discretization. The procedure applied to accomplish this has been performed by following standard CFD Best Practice Guidelines [8]. The pressure drop of the flow along the channels was selected as monitoring variable. The geometry of the Bipolar Plate consisted of a triple serpentine flow pattern as shown in Figure 2, with an active surface area of 45 cm2. Channel width and depth, and rib width 8 is 0.764 mm. The description of the meshes generated for the study is presented in Table 3, and results obtained are presented in Figures 3 to 5. Figure 2: Geometry of the Bipolar Plate used for the mesh discretization study. Table 3 Description of the meshes used for the mesh sensitivity study. Mesh ID Ne (x1E3) Nn (x1E3) Ne (d channel) Ne (w channel) Ne (L channel) Ne t GDL Mesh 1 420 490 5 5 40 4 Mesh 2 875 980 7 7 40 5 Mesh 3 1925 2125 10 10 60 5 Ne – Number of elements. Nn – Number of nodes. d channel – Channel depth. w channel – Channel width. L channel – Channel length. t GDL – Gas Diffusion Layer thickness. From the results obtained in this study, a Fuel Cell Meshing Best Practice Guidelines procedure has been developed with the following conclusions, applicable to cells of dimensions similar to the cell studied: 9 1. The channel cross-section must be resolved with at least 7x7 elements. 2. The channel length must be resolved with at least 40 elements. If non-conformal meshes are to be used due to a cross-flow bipolar plate configuration, the minimum number of elements must be 80. 3. The thickness of the different layers of the MEA: gas diffusion layers, catalytic layers and membrane, must be resolved with at least 5 elements. Figure 3: Mesh study results. Channel pressure drop versus number of mesh elements.