A prediction model of degradation rate for membrane electrode assemblies in direct methanol fuel cells

Abstract

This article proposes a new prediction model to describe the nonlinear performance degradation paths of membrane electrode assemblies (MEAs) in direct methanol fuel cell (DMFC): a bi-exponential model with random coefficients. The bi-exponential model is constructed on a mathematical basis representing Second-order kinetics. Performance variation between MEAs is incorporated by random coefficients in the proposed model. A likelihood ratio test is sequentially executed to select random effects in the nonlinear random-coefficients model. Analysis results indicate that the reliability estimation can be substantially improved by using the nonlinear random-coefficients model to incorporate two heterogeneous degradation characteristics of MEA performance during continuous operation of DMFC. Confidence intervals of failure-time distributions are obtained by the parametric bootstrap method.