Flow distribution in proton exchange membrane fuel cell stacks

Paul Chang, Jean St‐Pierre, Jürgen Stumper, Brian WettonJournal of Power Sources 2006 · DOI

TL;DR

This paper presents a computational model for proton exchange membrane (PEM) fuel cell stacks that accounts for how reactants are distributed among individual cells. By coupling a detailed unit cell model with a flow distribution model, the authors investigate how variations in operating conditions and manufacturing tolerances impact overall stack performance, highlighting the critical role of uniform flow for efficient and reliable fuel cell operation.

What Problem Does This Paper Address?

Proton exchange membrane fuel cells (PEMFCs) are a promising clean energy technology, but scaling them up into practical stacks presents significant engineering challenges. A key issue is ensuring that each individual fuel cell within a stack receives a uniform supply of reactants (hydrogen and oxygen) and that byproducts (like water) are efficiently removed. Uneven flow distribution can lead to "weakest link" scenarios where one cell underperforms or fails prematurely, compromising the entire stack's performance and lifespan. While previous models have addressed flow distribution in manifolds, they often simplified or ignored the impact of reactant consumption within the unit cell channels themselves, and rarely considered the effects of manufacturing variations. This paper aims to bridge this gap by developing a more comprehensive stack model that integrates unit cell performance with realistic flow distribution dynamics, including the influence of parameter variations.

Methods

The authors developed a computational model that combines two main components:

  • Unit Cell Model: This is based on a previously developed model [2] that accounts for the electrochemical reactions, mass transport of reactants and products within the channels, and the resulting cell voltage. It incorporates equations for mass conservation of oxygen, hydrogen, and water, as well as momentum conservation to describe fluid flow. Crucially, it couples cell performance directly to the local channel conditions.
  • Stack-Level Flow Distribution Model: This model applies Mass and Momentum Conservation principles to the headers (manifolds) that distribute reactants to and collect products from the unit cells. It handles flow splitting at tee junctions and recombination. The model considers both U-Type and Z-Type manifold configurations.

The computational strategy involves a nested iterative approach. An outer loop solves for the stoichiometry distribution (which dictates how much reactant each cell receives) using Newton iterations, while an inner loop computes header conditions and unit cell performance. The model allows for independent variation of parameters for each unit cell, enabling the simulation of manufacturing tolerances.

Datasets

Not discussed in the paper. The authors used their own model for computations and compared results against experimental data from a Ballard Mk 7 fuel cell stack.

Key Insights and Ideas

  • Coupling is Crucial: The paper demonstrates that the flow distribution within the unit cell channels (due to reactant consumption) significantly impacts the overall stack performance, a factor often overlooked in simpler models.
  • Manifold Design Matters: U-Type manifolds were found to be preferable to Z-Type manifolds for performance, especially at higher current densities, due to better voltage distribution.
  • Operating Conditions are Sensitive: At higher current densities, stack performance becomes much more sensitive to operating conditions like cathode stoichiometry and inlet pressure. Small changes can lead to significant variations in cell voltage.
  • Manufacturing Variations are Significant: Even small, normally distributed variations in parameters like cathode channel friction factor can lead to substantial cell voltage spreads, particularly in the mass transport limited regime. This highlights the importance of robust manufacturing processes.
  • Anode-Cathode Coupling: The study revealed an unexpected coupling effect where changes in cathode friction factor influenced the anode flow distribution, and vice-versa, due to reactant consumption and electrical current. This suggests that flow distribution in one compartment cannot be analyzed in isolation.

Key Results

  • The model successfully predicted pressure drops in headers and flow distributions, showing good agreement with experimental data for a Ballard Mk 7 fuel cell stack.
  • Simulations showed that decreasing cathode flow (e.g., by lowering stoichiometry) leads to decreased cathode performance and increased anode pressure drop in counterflow configurations.
  • The model quantified the impact of variations: a 20% standard deviation in cathode channel friction factor could lead to cell voltage variations of up to 0.4V in Case B (high current density) for a U-type manifold.
  • The paper demonstrated that U-type manifolds generally lead to more uniform cell voltage distributions compared to Z-type manifolds.

Limitations and Future Work

The authors acknowledge several limitations and suggest future research directions:

  • Two-Phase Flow: The current model does not explicitly account for two-phase flow (liquid water) in the channels and gas diffusion layers, which can significantly impact performance, especially at higher current densities. Future work will incorporate these effects.
  • Direct Flow Measurement: The paper notes the lack of a direct method to measure flow distribution in stacks, relying on indirect measurements and empirical formulas. Developing such a method would be highly beneficial for validation.
  • Electrical and Thermal Coupling: While the model considers electrical coupling to some extent, a more comprehensive inclusion of thermal effects and their interaction with flow distribution is planned.
  • Manufacturing Data: Obtaining validation data that directly links manufacturing parameter distributions to observed performance variations is crucial for establishing robust design guidelines.
  • Hydrogen Evolution: The model's negative cell voltages under certain conditions are an artifact of not accounting for hydrogen evolution, which would limit voltage to approximately 0V. This is a known limitation that will be addressed.

Summary updated

BibTeX
@article{chang2006flow,
  author = {Paul Chang and Jean St‐Pierre and Jürgen Stumper and Brian Wetton},
  title = {{Flow distribution in proton exchange membrane fuel cell stacks}},
  journal = {Journal of Power Sources},
  year = {2006},
  doi = {10.1016/j.jpowsour.2006.06.081},
}
Figure