Comparison between two PEM fuel cell durability tests performed at constant current and under solicitations linked to transport mission profile

https://doi.org/10.1016/j.ijhydene.2007.03.013Get rights and content

Abstract

The first phase of a research program on PEMFC durability was devoted to the test of a 100 W stack operated in stationary regime during 1000 h. The second phase was dedicated to the test of another 100 W stack under dynamical current constraint. In this case, the load current cycle applied was computed from a standardised transportation mission profile and then adapted to the power of the investigated fuel cell. For both ageing experiments, the stack characterisations were based on polarisation curves, recorded for various stoichiometry rates, as well as on EIS measurements performed at regular time-spaced intervals throughout the ageing. Some analysis tools derived from the response surface methodology are employed to analyse and compare the results of the two durability experiments. Some numerical models are proposed for the degradation of the stack performances. They are finally used to optimise the fuel cell operating conditions versus ageing time.

Introduction

Fuel cell (FC) systems are generally considered as promising and environmentally friendly energy converters for future electric vehicles. Indeed, FC can offer high fuel economy, through higher efficiency, and also substantially lower CO2 emissions. Considering the different FC types, which can be encountered, proton exchange or polymer electrolyte membrane FC (PEMFC) has been receiving more attention especially because of its high efficiency and power density. However, some major technical issues are still to be solved for the wide-spread marketing of FC generators into the transportation area. Economical viability depends notably on improving the durability and the reliability of these new embedded generators. FC lifetime requirements vary significantly, from 5000 h for car applications up to 20 000 operating hours for bus applications. In comparison with conventional combustion engines, the FC could appear as being an inherently reliable system because of the absence of moving parts, but the stacks are prone to material degradation and the MEAs are placed under mechanical constraints. Some papers [1], [2], [3] provide an overview of operating conditions which can have significant effects on PEMFC durability, including for instance low reactant gas flows, too high and too low humidification levels, high and low temperatures. A few physical degradation causes can already be mentioned: loss of active catalyst surface, poisoning, loss of proton conductivity in the membrane, deterioration of mass transport properties in the porous layers, ‘hotspots’ and holes in the membrane, sealing degradation, corrosion on plates, etc. Nevertheless, all the failure modes, degradation causes and mechanisms are not yet fully understood. Moreover, it can be noticed that the ‘end-of-life’ (EOL) performances are still not often specified by stack manufacturers because these performances strongly depend on the way the FC has been used.

The L2ES laboratory based on the FC test platform of Belfort currently carries out research on small 100 W PEMFC stacks. The aim of the lab is to ensure for the FC system proper operating conditions, leading to high efficiency delivery as well as high performances in term of reliability and durability. As L2ES and INRETS work on FC by focusing on the system aspect, the main goal of the program is to evaluate and to adapt the technological choices made for the various needed actuators, ancillaries, electronic converters, peak power devices such as supercapacitors, etc. in order to increase stack lifetime. The first step of the program was devoted to the test of a 100 W three cell stack operated in stationary regime at roughly nominal conditions during 1000 h [4]. This first experiment serves as reference test for the other studies that are led on other 100 W stacks and in different environment conditions. The second phase was dedicated to the durability test of another 100 W stack placed during about 700 h under dynamical current constraint linked to vehicle road cycle [5]. It is generally admitted that the cyclic current loading condition is one of the severe environment constraints, which can have strong impacts on the lifetime and reliability of embedded FC systems [6], [7], [8]. Therefore, this major issue has to be investigated, in particular with regard to related strategies of dynamic gas flow management. Besides, as the FC durability experiments are expensive and time consuming, the tests have to be carefully implemented and exploited. Therefore, well-suited methodologies have to be adopted as well as efficient and practical tools for the test analyses. Some techniques derived from the response surface methodology (RSM) can be developed with this aim in view.

In the second and next part of the paper, the experimental set-up will be briefly described. The durability test conditions will be discussed as well. The third part will be devoted to a global presentation of the main experimental results collected. The fourth section of the paper will focus on deeper experimental analyses as well as on comparisons between the two experiments. Finally, an optimisation of the operating conditions as a function of ageing time will be made.

Section snippets

Experimental set-up

The three cell PEMFC stacks used in this study have been assembled with commercial membranes (Gore MESGA Primea Series 5510; active cell area of 100 cm2), gas diffusion layers and machined graphite flow distribution plates. The FC operates at atmospheric pressure (maximal pressure of 1.5 bar abs.). A detailed description of the 1 kW test bench used for the durability tests can be found elsewhere [9], [10]. Many physical parameters involved in the stack can be controlled and measured in order to

For the first ageing test

In the case of the first ageing test performed at constant current, we are able to display the evolutions of stack and cell voltages over the whole test duration. Fig. 2 shows the stack voltage that was recorded continuously. The variable is plotted each 24 h, but not during the characterisation sequences. Except at the very beginning of the test (where a slight improvement of the voltage could possibly be detected), the stack exhibits quite constant voltage during the first 350 h of the

Using the polarisation curve records

Such a graph like Fig. 4 can already provide a quite good and simple visual representation of the FC ageing. The contour plots of the voltage response surfaces in the plane ageing time–load current could be used to compare the two ageing tests. For instance, the response surface displayed for the voltage of the first ageing test could be easily subtracted from the surface obtained for the second test. Nevertheless, this method is not really suitable if the impact of ageing time over the FC

Optimisation of the operating conditions

In this last part, an optimisation of the FC performances is done by taken into account the time, the current and the hydrogen/air stoichiometry rate factors. This study is another interesting application of the models proposed in the previous section. The goal of the optimisation is to find the parameter values of the stoichiometry rates FSA/FSC that lead to the highest FC voltage. Considering the experimental domain explored, the computing problem satisfying some equality constraints shall be

Conclusion and perspectives

In this paper, we have compared the results of two ageing tests performed on similar 100 W three cell PEMFC stacks operated in the first case at constant current during 1000 h, and in the second case under specific dynamical load transients linked to vehicle road cycle during about 700 h. The experimental observations have been made from the stack voltages recorded along the total ageing test durations as well as from the static current–voltage characteristics (obtained for four various FSA/FSC

Acknowledgement

Financial support by the French Government under contract 01Y0044-02 is gratefully acknowledged.

References (18)

There are more references available in the full text version of this article.

Cited by (0)

View full text