Flow Regime Evolution in Long, Serpentine Microchannels with a Porous Carbon Paper Wall

An important function of the gas delivery channels in Proton Exchange Membrane (PEM) fuel cells is the evacuation of liquid water created at the cathode. The resulting two-phase flow can become an obstacle to reactant transport and a source of parasitic losses. The present work examines the behavior of two-phase flow in 500 μm x 500 μm x 60 cm channels with distributed water injection through a porous carbon paper wall to gain understanding of the physics of flows relevant to fuel cell water management challenges. Flow regime maps based on local gas and liquid flow rates are constructed for experimental conditions corresponding to current densities between 0.5 and 1 A/cm 2 and stoichiometric coefficients from 1 to 4. Flow structures are analyzed along the entire length of the channel. It is observed that slug flow is favored to plug flow at high air flow rates and low liquid flow rates. Stratified flow dominates at high liquid flow rates. Along the axial flow direction, the flow regime consistently transitions from intermittent to wavy to stable stratified flow. This progression is quantified using a parameter of flow progression which characterizes the degree of development of the two-phase flow toward the stable stratified condition. This parameter is discussed in relation to fuel cell operating conditions. It provides a metric for analyzing liquid water removal mechanisms in the cathode channels of PEM fuel cells. INTRODUCTION The management of water in a proton exchange membrane (PEM) fuel cell is a performance-limiting concern in these carbon-free energy-conversion devices. The performance of the membrane depends on ample humidification. However, liquid water in the microporous layer, gas diffusion layer (GDL), and/or gas delivery channels impedes the flow of reactant gases to catalyst sites and cripples cathodic reaction rates. Two-phase flow phenomena have been identified as concerns for flooding-induced performance degradation, cell voltage hysteresis, increased parasitic pumping losses, and the absence of predicted performance improvements [1-5]. While microchannels are proposed for use in PEM fuel cells with advantages of increased convective transport, flexibility of geometric design, and decreased resistive path lengths, these benefits have not been observed in practical systems [4]. One explanation for the unrealized performance improvements is an increased propensity toward flooding in smaller channels, where surface tension becomes significant compared with other fluidic forces. The present work provides fundamental insight into the physics of two-phase flow in geometries common to practical fuel cells which is critical to their performance. Some two-phase flow characterization has been developed for in-situ fuel cell applications. In fuel cell components lacking optical access, neutron imaging, magnetic resonance imaging, and X-ray tomography allow for identification of liquid water through opaque surfaces. Neutron imaging is a promising technique for real-time imaging of liquid water in an operating fuel cell [8-10]. This technique is limited to two dimensions; delicate analysis is necessary to distinguish flow structures between the anode and cathode. X-ray tomography and 1 Copyright © 2008 by ASME magnetic resonance imaging have been used to create three dimensional maps of water concentration in the membrane and GDL layers with limited resolution [11, 12]. Opaque bipolar plates, layers of porous materials with various surface properties, and non-uniformity of water production rate and location are obstacles to the analysis of flow regimes in fuel cell channels. Fuel cells with modified geometries allow optical access into the gas distribution channels and have been used to correlate the appearance of liquid water with increasing in gas pressure in the channels and voltage drops in the fuel cell [13-19]. These fuel cells may demonstrate atypical reaction distribution because the electrical pathways are distorted when conductive bipolar plates are replaced with non-conductive glass plates for optical access. Much insight can be drawn from these studies in the correlation between liquid water and fuel cell performance; however the precise flow conditions at any particular location in the cell are difficult to quantify due to non-uniformities in reaction distribution and subsequent uncertainties in local flow rates. Absent the electrochemistry of an operating fuel cell, ex-situ visualization experiments can quantify the local flow rates of both phases. Experiments with a porous GDL wall have been used to simulate the GDL-channel interaction of a fuel cell for mini-channels. Single-phase gas flow profiles have been examined in small channel sections using particle image velocimetry in order to characterize the interaction of the channel and porous layer and the flow around corners [20-22]. Su et al. studied flow in 5 mm square cathode channels by injecting water into the channels through a carbon paper layer; the large channel dimensions were chosen to reduce the effect of surface tension on the resulting flow [23]. Others have examined droplet departure from a GDL into a gas channel using short channel segments [24, 25]. Two-phase flow patterns are characterized in microchannels for lab-on-a-chip and microchannel cooling applications [26-29], however these studies do not replicate the distributed liquid water introduction, long channel lengths, and porous boundary conditions of a PEM fuel cell. This study bridges the gap between in-situ experiments on operating fuel cells and ex-situ two-phase flow experiments in different geometries. Metered amounts of liquid water are introduced into the channel through the porous carbon paper gas diffusion layer which flanks one channel wall, thereby allowing for the characterization of flow regime transitions according to known flow rates of both phases. Unlike visualization of operating fuel cells, this setup allows for precise control of the liquid and gas flow rates and injection conditions. Compared with other ex-situ experiments, this study incorporates channels with smaller cross-sectional dimensions than many previous fuel cell related studies. Compared with fundamental two-phase flow studies in microchannels, this study better replicates the characteristics of fuel cell channels, including longer channel lengths and distributed water introduction. This study provides insight to key two-phase flow structure transitions of importance for performance-critical water management in PEMFC channels. NOMENCLATURE i current density Q volumetric flow rate U superficial velocity λ stoichiometric ratio Subscripts a air (gas phase) w water (liquid phase) EXPERIMENTAL METHOD A single serpentine channel replicates flow conditions associated with fuel cell gas delivery channels. A channel with a 500 x 500 μm square cross-section is machined in acrylic with a 60-cm long serpentine layout incorporating five long segments and four short turns. The fourth wall of the channel is comprised of a 2 mm wide, 190 μm thick ribbon of porous carbon paper GDL which follows the channel geometry and fits tightly into a latex gasket seal. The GDL paper is pressed against the channel structure with an acrylic slab using 44 bolts for even distribution of pressure. Water is injected into the channel at eight locations using a multisyringe pump (Harvard Instruments) via 500 μm diameter holes through the acrylic slab. Figure 1 is a schematic of the structure and its crosssection. Various injection configurations were considered before discrete liquid injection was chosen. Other injection Figure 1. Cross-section of channel assembly. 1⁄2” thick Acrylic Sheet B A A B