Optimization of Realistic Accelerated Stress Tests for PEM Fuel Cells Using Standardized Automotive Driving Cycles

Abstract

Proton exchange membrane (PEM) fuel cells have so far not achieved the targeted lifetime necessary to be an economical alternative for automotive applications. This is mainly caused by their high degradation rates. To better understand the degradation mechanisms, the use of accelerated stress tests (ASTs) at the component level is widespread. However, it is still a challenge to transfer their results to real-world applications.To address this problem, the approach of this work is to derive load profiles from standardized driving cycles for passenger cars. In combination with a fuel cell system model that is used to define the operating conditions of the media, such as temperature or pressure, a test procedure is obtained that results in a realistic degradation of the fuel cell. The proposed test procedure exposes the PEM fuel cell (PEMFC) to the most important degradation conditions in mobile applications. Dynamic operation is responsible for deterioration processes within the PEMFC and is the underlying cause of typical hygrothermal and voltage cycling events. Such cycling also occurs during start-stop operations, causing severe damage to the PEMFC catalysts. In addition, idling conditions may prompt open-circuit voltage (OCV) conditions in the PEMFC, resulting in dissolution of the Pt catalyst and subsequent chemical degradation of the PEM. Additionally, high load operation may also lead to catalyst degradation. Finally, the inertia of the auxiliary components, such as the humidifier, during load changes causes not ideal conditions, such as relative humidity changes, resulting in further mechanical stress on the membrane.Due to the targeted PEMFC lifetime of up to 8,000 hours for automotive applications [1], realistic degradation test procedures have a very long test duration, making them particularly challenging and costly to conduct. To overcome this hurdle, the load profile of the proposed test procedure is shortened by using an in-house developed optimization algorithm. For this purpose, four operation modes are defined as degradation-accelerating: voltage cycling, relative humidity (RH) cycling, OCV/idling, and high load operation. Based on state-of-the-art AST procedures, a sub-algorithm is developed for each operating mode. Such sub-algorithms are capable of identifying all occurrences within a given test procedure that may lead to accelerated fuel cell degradation. For each sub-algorithm, the degradation parameters (namely upper and lower potential limits for voltage cycling, membrane tensile stress for RH cycling, fluoride emission rate for OCV/idling, and voltage degradation rate for high load) are specified by the user. This allows the comparison of different fuel cells in terms of their resistance to conventional degradation mechanisms.The developed algorithm can be used to optimize various driving cycles such as the worldwide harmonized light vehicles test cycle (WLTC) for various mobile applications. All cycle events identified as accelerating degradation are connected with a point connector function based on the fuel cell polarization curve. This ensures moderate voltage steps between the identified phases and thus prevents the algorithm from inducing further degradation of the fuel cell.The optimization algorithm obtained can reduce the test duration by up to 40%. The method can be repeated frequently during AST operation by repeatedly measuring the polarization curve to consider the continuous degradation in the test procedure. Decreasing fuel cell performance leads to more severe voltage drops during load changes and thus an increase of degradation-accelerating phases over the test period. Furthermore, start-stop ASTs developed by Bisello et al. [2] are also included due to their strong contribution to fuel cell degradation in automotive applications.Finally, the optimized cycles are analyzed using two simulation tools to evaluate the degradation. The first tool by Ao et al. [3] uses the electrochemical surface area (ECSA) to quantify the proceeding degradation during cycle operation. A characteristic voltage and the voltage change rate are extracted from the driving cycle. For both values, the resulting decrease in ECSA is modeled and aging is defined as the ECSA loss rate. The second tool by Pei et al. [4] examines the driving cycle according to the four degradation-inducing conditions and combines them with experimentally determined degradation rates. Both tools allow an analysis of the degradation derivative between the initial and optimized test procedures. By adjusting the degradation rate for the different degradation mechanisms, the AST procedure is improved in terms of its representativeness for real-world applications.