Component-based modelling of PEM fuel cells with bond graphs

A Vasilyev,J Andrews,L.M Jackson,S.J Dunnett,B Davies

doi:10.1016/j.ijhydene.2017.09.004

A Vasilyev, J Andrews + Show 3 more

Open Access

https://doi.org/10.1016/j.ijhydene.2017.09.004

Copy DOI

Abstract

A polymer electrolyte membrane (PEM) fuel cell is a power generation device that transforms chemical energy contained within hydrogen and oxygen gases into useful electricity. The performance of a PEMFC unit is governed by three interdependent physical phenomena: heat, mass, and charge transfer. When modelling such a multi-physical system it is advantageous to use an approach capable of representing all the processes in a unified fashion.This paper presents a component-based model of PEMFCs developed using the bond graph (BG) technique in Modelica language. The basics of the BG method are outlined and a number of relevant publications are reviewed. Model assumptions and necessary equations for each fuel cell component are outlined. The overall model is constructed from a set of bond-graphic blocks within thermal, pneumatic and electrical domains.The model output was compared with the experimental data gathered from a two-cell stack and demonstrated a good accuracy in predicting system behaviour. In the future the designed model will be used for fuel cell reliability studies.

Highlights

Fuel cells (FCs) are a technology for electricity generation that promises to play an important role in the future energy landscape
It is clear that bond graphs are very well suited for modelling fuel cell devices as demonstrated by the reviewed publications and modelling work performed in this paper
The designed model relies on a pseudo-bond graph representation of thermo-fluid phenomena and is implemented in Modelica modelling language

Summary

Introduction

Fuel cells (FCs) are a technology for electricity generation that promises to play an important role in the future energy landscape. Some engineering systems exhibit coupled energy phenomena, such as cooling systems where thermal energy transfer is associated with a moving volume of liquid or gas Such thermo-fluidic systems may require a different set of power variables in order to represent them efficiently. This may lead to an inconsistent representation of power within a model and care must be taken when employing both pseudo- and true bonds in one graph [6]. Pseudo bonds are widely used when modelling thermofluid and linear heat transfer problems because they offer a more intuitive way of representing such processes

Literature review

À 1:3553ach Á þ 1:9467a2ch

O2 H2O

Conclusions