A quick evaluation method for the lifetime of the fuel cell MEA with the particle filter algorithm. Issue 14 (14th November 2021)

• Record Type: Journal Article
• Authors: He, Luyan; Zhan, Zhigang; Chen, Hong; Jiang, Panxing; Yu, Yuan; Yang, Xiaoxiang; Sun, Yahao; Wan, Xiongbiao; Liao, Liwen; Li, Shang; Pan, Mu
• Is Part Of: International journal of green energy. Volume 18:Issue 14(2021)
• Page Start: 1536
• Page End: 1549
• Publication Date: 2021-11-14
• Subjects: PEMFC -- membrane electrode assembly -- particle filter -- lifetime -- quick evaluating; Power resources -- Research -- Periodicals; Energy industries -- Periodicals; Energy development -- Periodicals; 333.79
• DOI: 10.1080/15435075.2021.1911809
• Languages: English
• ISSNs: 1543-5075

Abstract

ABSTRACT: The proton exchange membrane fuel cell (PEMFC) has bright prospects in energy applications, but its durability and cost still affect its commercialization. In this work, a group of membrane electrode assemblies (MEAs) were tested to study the quick evaluating method for predicting their lifetimes. MEA1 was tested under the real constant load till the end of its lifetime and MEA2 was tested with a quick evaluating method. The data were treated with the particle filter algorithm to eliminate the influence of all kinds of errors, including the system error and random error. Based on this, the lifetime prediction model was developed. For a degradation rate of 10% of the initial working voltage, the lifetime of MEA1 was 6140 h; that of MEA2 was 3575 h if predicted with the test data directly, and 6776 h if the data was treated with the particle filter algorithm. The relative errors for both cases were about 41.8% and 10.3%, respectively, when compared with the lifetime of MEA1; indicating that the prediction accuracy was greatly improved. Furthermore, it was found that the voltage degradation rate was nonlinear when the whole lifetime of the MEA was considered, and the degradation rate increased significantly near the end period. In the quick evaluating method employed in most studies, the lifetime prediction model uses a linear degradation rate, which tends to cause errors; therefore, further research is needed.