Fuel Cell Power Research for Transportation

This paper describes the fuel cell research activities conducted in the National Tsing Hua University in Taiwan. A series of fuel cell power research for transportation were based on the academic fundamentals of the molecular dynamics (MD), fluid dynamics and system dynamics, scoped from micro-scale, messo-scale, macro-scale analyses to the system level of fuel cell vehicle design. Various power ranges of fuel cells, such as proton exchange membrane fuel cells (PEMFC), direct methanol fuel cells (DMFC) and solid oxide fuel cells (SOFC) were studied. The electrolyte composition, flow channel configuration and system design variations of the fuel cell stack were parametrically investigated. A general purpose, multi-scale analysis software and a real-time hybrid electric vehicle simulator were developed in this paper. Fuel cell scooters and a personal mobility vehicle were the first target to build up and also for verifying the software. This paper will explain the design concept and the related fundamental research. Key-words: multi-scale analysis, molecular dynamics, fluid dynamics, system dynamics, PEMFC, DMFC, SOFC INTRODUCTION National Tsing Hua University conducted a series of research on electric vehicles in the 1980s after the first world energy crisis [1]. The outcome was a success technically, but not commercially. Similar to those happened in the world that the bottleneck was the battery performance. Limited energy density and long recharging time of the battery make the electric vehicle (EV) impractical. Only few EVs were built up for demonstration and for pilot study. In the 1990s, the research direction was switched to hybrid vehicles and electric scooters. The hybrid vehicles were aimed to extend the mileage and enhance the performance by implementing an efficient internal combustion engine. The electric scooter research was encouraged by the government due to the concern of air pollution problems in the urban area. It turns out to be a successful product due to the reason that the scooters don’t need so much power for auxiliaries like in passenger cars. The mileage ranges from 40 to 60 km, that is a normal traveling distance for a scooter rider in a week. However, the recharging time which takes 6 to 8 hours is still a complaint from the consumers. This is the reason why we switched to the fuel cell power research after 2000. The fuel cell research task assigned to the university was to develop the simulation and design tools for the industry. The university research was not aimed to produce a commercial product, that was led by the Industrial Technology Research Institute (ITRI) near the campus [2]. Also the Taiwan Institute of Economic Research (TIER) was in charge of commercialization [3]. This is the reason why we concentrated on the fundamental research and aimed to investigate the proton exchange mechanism (microscale molecular dynamics), fuel and oxidizer transport phenomena (messo-scale computational fluid dynamics), and system design optimization (macroscale system dynamics) for fuel cell power from the academic theories. MOLECULAR DYNAMICS The fundamental research on the fuel cell power can be started from the ion diffusion mechanism study which takes place inside the electrolyte. The electrolyte is the major component of the fuel cell, its ionic conductivity may decide the fuel cell performance. Since solid electrolytes are more welcome by engineers because of easy assembly and packaging, we start the molecular dynamics simulation of the ion transportation mechanism on the SOFC, PEMFC and DMFC. SOFC ELECTROLYTE – The electrolyte in SOFC is a solid, nonporous, metal oxide. The most common one is usually the yttria-stablized zirconia (YSZ), which is composed of pure zirconia (ZrO2) and some yttria (Y2O3). Fig. 1 shows the basic molecular structure of the YSZ. Pure zirconia is in monoclinic structure at room temperature. It is stabilized into the cubic florite structure as YSZ after being doped some yttria. The cations (Zr and Y) form a face-centeredcubic (FCC) lattice and oxygenions are in the tetrahedral sites of cations in the cubic fluorite structure. Oxygen vacancies are produced to maintain charge neutrality in YSZ, and two Y would form one oxygen vacancy. Fig. 2 shows the trajectory of discrete hopping of an oxygen ion at operating temperature of 1273 K. Due to the fact that higher temperature provides more kinetic energy, it helps ions to jump longer distance and visit more sites. The results lead to the increase of diffusion coefficient and also the ionic conductivity because of higher temperature. We can dope more Y2O3 to increase the oxygen vacancy from engineering viewpoint. However, from the molecular dynamics simulation, the result shows that 8 mol% of Y2O3 concentration provides the maximum value of the ionic conductivity (see Fig. 3). The reason was due to the Columbic force balance that restricts the vacancy position even more Y2O3 being doped into the molecular structure. The ionic conductivity was measured by experiments and proved that the molecular dynamics simulation results were correct [4]. Fig. 1 Molecular structure of the YSZ Fig. 2 Discrete ion hopping trajectory for a single oxygen ion Fig.3 Ionic conductivity prediction using MD technique PEMFC (or DMFC) ELECTROLYTE – The proton exchange membrane (PEM) was used as the electrolyte for both hydrogen and direct methanol fuel cells. The basic structure is composed of sulphonated fluoropolymer, usually fluoroethelyne. The most wellknown commercial product is Nafion by DuPont. The hydrogen ions were combined with water and transport along the clusters of hydrophilic sulphonate side chains (SO). Fig. 4 shows the simulation of the transport mechanism of two H3O ions along with the two side chains and water clusters. The dimension in this simulation is in nano-scale. Fig. 4 Ion transport simulation in the Nafion In the DMFC system, the critical operating problem is the carbon dioxide (CO2) venting in the anode side. The CO2 bubbles may block the fuel passage, and also the flow channels. The bubble dynamics inside the porous diffusion layer can be simulated by treating the porous medium as blocks inside a micro channel. The dimension is in messo-scale and the fluid dynamics is in two phase flow. It is a perfect case by employing Lattice Boltzmann simulation [5]. Fig. 5 and Fig. 6 show the bubble transport along a micro channel with and without blocks. The bubble is blocked by obstacles if the gap is either too small or the fluid-wall interaction is not significant. The problem can be solved by studying the parametric relations between the buoyancy, surface tension of the fluid and fluidsolid wall interactions. The technical skill can be used to design a gas-fluid separator. Fig.5 Bubble transport along a micro channel without blocks. Fig.6 Bubble transport along a micro channel with blocks on the wall. FLUID DYNAMICS Computational fluid dynamics (CFD) is a useful tool in designing the fuel cell stack and system configurations. The mathematical model is based on the conservation laws of mass, momentum, energy and species. Table 1 displays the governing equations with different source terms in the flow channel, diffusion layer, catalyst and the membrane. Although the NavierStokes equation is valid in all sub-systems, the governing equation inside the porous diffusion and catalyst layers can be replaced by the Darcy law to reduce the computation time. Fig. 7 shows three types of typical flow channel designs, and the corresponding oxygen concentration fields generated inside the diffusion layers are illustrated in Fig. 8. The parallelthree serpentine design (right-hand side) shows the best diffusivity of the oxygen gas. The over-potential in this case is proved to be the lowest among those three examples [6]. The simulation can be extended to the anode side, where the anode over-potential is taken account. Fig. 9 shows the comparison of V-I performance between DMFC and PEMFC using the same flow channel configuration. The DMFC degrades the performance apparently because of lower diffusivity of the liquid methanol and also the problem of methanol cross-over. Fig. 10 shows that the single cell structure can be evaluated by CFD simulation whose governing equations are listed in Table 1. The velocity field, concentration field and temperature field are illustrated and the resulting membrane temperature distribution is displayed. From macroscopic viewpoint, transport phenomenon in the flow channel, including the diffusion and catalyst layers, will influence the temperature distribution on the solid electrolyte. That will induce thermal crack and sealing problems in the high temperature operation such as in the SOFC case. The CFD simulation can be further extended to multicell fuel cell stack design. Fig. 11 is an example case of the influence of inlet manifold design on the pressure field of the multi-cell design. Table 1 CFD model governing the fuel cell transport phenomenon Fig. 7 Examples of flow channel designs Fig. 8 Oxygen concentration field at the cathode side Fig. 9 Methanol concentration field at the anode side and the corresponding V-I curve, compared between DMFC and PEMFC Fig. 10 Velocity, concentration and temperature fields of a single cell stack Fig. 11 The influence of inlet manifold design on the multi-cell fuel cell stack performance membrane temperature velocity field concentration field temperature field 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2