Date of Award

8-2012

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Automotive Engineering

Advisor

Pisu, Pierluigi

Committee Member

Omar , Mohammad

Committee Member

Prucka , Robert

Committee Member

Vahidi , Ardalan

Abstract

Of the fuel cells being studied, the proton exchange membrane fuel cell (PEMFC) is viewed as the most promising for transportation. Yet until today, the commercialization of the PEMFC has not been widespread in spite of its large expectation. Poor long term performances or durability, and high production and maintenance costs are the main reasons. For the final commercialization of fuel cells in the transportation field, durability issues must be addressed, while costs should be further brought down. At the same time, health-monitoring and prognosis techniques are of great significance in terms of scheduling condition-based maintenance (CBM) to minimize repair and maintenance costs, the associated operational disruptions, and also the risk of unscheduled downtime for the fuel cell systems.
This dissertation presents a comprehensive on-line supervisory system to address the important issues related to the PEMFC durability, including: 1) diagnosis of critical operating conditions, 2) optimization of the operating conditions, and 3) health monitoring (or damage tracking) and remaining useful life (RUL) prediction. In order to design and implement this supervisory system, a comprehensive fuel cell model is developed that integrates a control/diagnostic oriented dynamic fuel cell model and a prognostic oriented fuel cell degradation model, due to a lack of such models in the existing literature.
To address the first issue, a model-based on-line diagnostics system is developed for fuel cell flooding and drying diagnosis, thanks to the incorporation of the diagnostic feature in the dynamic fuel cell model. The channel flooding diagnostic problem is decoupled from the gas diffusion layer (GDL) flooding and membrane drying diagnostic problem. Simultaneous state and parameter estimation problems are formulated for both cases. Dual extended Kalman filter (EKF) and dual unscented Kalman filter (UKF) techniques are applied respectively to solve the problems. The second issue is addressed by a diagnostic based control design for the air supply of the fuel cell system. The design concept allows selection of the most suitable controller in a controller bank that delivers the best performance under specific operating conditions and that mitigates the faulty condition based on the feedback of the diagnosis results. The control problem is reformulated as an H-infinity robust control problem, the objective of which is to minimize the difference between the desired and actual excess O2 ratio, thus preventing and minimizing oxidant starvation at the cathode. Finally, an UKF-based health-monitoring and prognostic scheme is proposed and applied to the damage tracking and RUL prediction for the fuel cell. The developed aging model is employed as the kernel for this scheme, which utilizes the fuel cell output voltage as the only feature for the prognostic and health management task.

Share

COinS