No abstract

Prognostics applications on PEMFC are developing these last years. Indeed, taking decision to extend the lifetime of a PEMFC stack based on behavior and remaining useful life predictions is seen as a promising solution to tackle the too short life's issue of PEMFCs. However, the development of prognostics shows some lacks in the literature. Indeed, performing prognostics requires health indicators that reflect the state of health of stack, while being able to interpret them in an industrial context. It is also important to propose criteria to set its end of life. Moreover, to trust any prognostics' application, one should be able to evaluate the performance of its algorithms with respect to standards. To help launching a discussion on these subjects among scientific and industrial actors, this paper addresses some of the issues encountered when performing prognostics of a PEMFC stack. After showing the link between prognostics and decision, this paper proposes guidelines to set the limits of a prognostics approach. The definitions of healthy and degraded modes are discussed as well as how to choose the time instant to perform predictions. Then, three criteria based on the power produced by the stack are proposed as indicators of the state of health of the stack. The definition of the end of life of the stack is also discussed before proposing some criteria to assess the performance of any prognostics algorithm on a PEMFC. Some perspectives of works are also discussed before concluding.

International audienceAlthough, the proton exchange membrane fuel cell is a promising clean and efficient energy converter thatcan be used to power an entire building in electricity and heat in a combined manner, it suffers from a limited lifespandue to degradation mechanisms. As a consequence, in the past years, researches have been conducted toestimate the state of health and now the remaining useful life (RUL) in order to extend the life of such devices.However, the developed methods are unable to perform prognostics with an online uncertainty quantification dueto the computational cost. This paper aims at tackling this issue by proposing an observer-based prognostic algorithm.An extended Kalman filter estimates the actual state of health and the dynamic of the degradation with the associateduncertainty. An inverse first-order reliability method is used to extrapolate the state of health until a threshold isreached, for which the RUL is given with a 90% confidence interval. The global method is validated using a simulationmodel built from degradation data. Finally, the algorithm is tested on a dataset coming from a long-term experimentaltest on an eight-cell fuel cell stack subjected to a variable power profile