A Real-Time Pseudo-2D Bi-Domain Model of PEM Fuel Cells for Automotive Applications

Abstract

With the goal of on-line diagnosis for automotive applications in mind, a real-time model of polymer electrolyte membrane (PEM) fuel cell is developed. The model draws from the authors’ previous modeling effort in this area and extends its domain to incorporate transport under the lands. Transport in the catalyst and micro-porous layers, which were previously omitted, are also included in the model. Membrane water transport model is modified accordingly. Moreover, a recently developed homogeneous catalyst layer model is used to describe local oxygen transport resistance in the cathode catalyst layer. Computational efficiency is achieved through spatio-temporal decoupling of the problem, which simplifies the handling of the nonlinear terms. This computational efficiency is demonstrated by a set of simulations that resemble operation under conditions encountered in automotive applications. Moreover, simulation results of the model are in qualitative agreement with earlier computationally intensive modeling studies as well as experimental observations. The current modeling study demonstrates a significant potential for using relatively high-fidelity physics-based models on-line to improve fuel cell performance and durability, which can have a profound impact on its commercialization.

Effective Parameterization of PEM Fuel Cell Models—Part II: Robust Parameter Subset Selection, Robust Optimal Experimental Design, and Multi-Step Parameter Identification Algorithm

“…In this work, we use a physics-based model that has been developed recently for automotive applications. [37][38][39] Particularly, the model is pseudo-2D and captures non-isothermal and two-phase phenomena and has been extensively validated with performance and resistance data obtained experimentally from state-of-the-art fuel cell stacks. 39 Further details about the model and its validation can be found in the original publications.…”

“…39 Further details about the model and its validation can be found in the original publications. [37][38][39] The sensitivities of several of the model predictions to a variety of parameters were examined in the first part of this work using an extended local analysis. 40 Particularly, cell voltage, high frequency resistance (HFR), and normalized membrane water crossover were studied for their sensitivity to 50 of the model parameters.…”

“…Accordingly, here we develop a systematic framework for effective model parameterization that addresses some of the current shortcomings of the fuel cell literature. While the framework is generally applicable to fuel cell models described by differential-algebraic equations, we use a pseudo-2D two-phase and non-isothermal model [37][38][39] for this study as an example. Having investigated the sensitivities of various model predictions to different parameters in the first part of this work, 40 here we focus on the problems of parameter subset selection and optimal experimental design.…”

A Mathematical Model toward Real-Time Monitoring of Automotive PEM Fuel Cells

“…where λ * is the dynamic quasi-equilibrium water content for the ionomer used to model the effects of stress relaxation [9,27] (see Table 1). In addition to S ad , both electrochemical and chemical water productions contribute to source term for ionomer water content.…”

“…Here, our specific focus is on capturing temperature and water distributions throughout the cell, which is achieved through a pseudo-2D bi-domain modeling approach. In particular, this work draws from our earlier effort in this area [13,27] and enhances it as follows: i) the bi-domain modeling approach allows for capturing of the in-plane distributions, ii) the microporous layer (MPL) is explicitly accounted for and is no longer lumped with the gas diffusion layer (GDL), while the catalyst layer (CL) is treated as a component with finite thickness rather than an interface, iii) the counter-flow configuration is modeled while maintaining real-time computation capabilities, iv) the model is more efficiently implemented to allow for significant savings in computation time, and v) the model is experimentally validated with performance data from a differential cell and an automotive short stack under a variety of operating conditions. These modifications render the model suitable for real-time monitoring of unmeasured and critical states within the fuel cell stack.…”