Model-based fuel cell design
Data processing – the essential first step
Fuel cell designers and system manufacturers face the challenge of interpreting large volumes of experimental data in order to design and optimise their products.
The problem is compounded by the fact that taking measurements from within the membrane electrode assembly (MEA) is difficult, and embedding sensors within the MEA alters its performance.
Advanced process modelling provides an easy and reliable way to quantify micro-scale phenomena occurring within the MEA based on macro-scale experimental data as an essential first step in fuel stack and system design.
Using macroscale data to understand microscale phenomena
Using high-fidelity models coupled with experimental data results in a significant improvement in the understanding of phenomena at both the macroscopic and microscopic level.
This reduces R&D risk by enabling the design of more effective experiments which target the cause of performance and life time loss.
The data shown here is from a single-cell experiment, courtesy of Toyota Motor Co..
How it works
The diagram shows the workflow for estimating model parameters from experimental data.
The steps involved are as follows:
1. Conduct experiments
Experiments are typically conducted using a single coin-sized cell (4 cm2) if available, operating at overflow conditions at anode and cathode.
Current is varied over time, with all other operating conditions maintained constant, to generate polarisation curves.
2. Construct experiment model
A model of the stack or single cell used in the experiment is constructed in gFUELCELL, using standard library models. Where parameter values are known, these are entered in the specification dialog box. Unknown parameter values will be estimated in the next step.
3. Estimate parameters
gFUELCELL’s parameter estimation capability is used to estimate key model parameters. Multiple parameters can be estimated from multiple steady-state and dynamic experiments simultaneously.
4. Analyse micro-scale phenomena
Once parameters have been determined, the model can be used to analyse microscale phenomena, providing invaluable information about phenomena such as overpotential and liquid water production.
Parameter estimation also provides a measure of the uncertainty inherent in the parameter values, in the form of confidence intervals.
This uncertainty information can be used to determine where to focus experimentation in order to minimise risk in the final design, thus providing an effective tool for managing technology risk over the development lifecycle.