An Application of Weibull Analysis to Determine Failure Rates in Automotive Components

This paper focuses on the automotive component failure rate detection using Weibull analysis and other survival analysis techniques. Detailed attention is paid to three areas: 1) overall failure rates are described statistically first, and data cleaning and definitions of ‘failed’ and ‘censored’ data within the research time or warranty period are made; 2) Kaplan-Meier life curves and Log-rank tests are used to compare the component reliability over time and explore risks factor related to the component failure; 3) Weibull regressions, with two and three parameters, are applied to fit real-world reliability data from different test conditions, and to predict the automotive component failure trend over future time. The analysis results agree well with real-world test data, and provide reasonable prediction of future failure trends.