Mixed Models for Product Design Selection Based on Accelerated Degradation Testing

• Published in: Proceedings. Annual Reliability and Maintainability Symposium pp. 1 - 6
• Main Authors: Ruiz, Cesar, Dashti, Ali
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.01.2025
• Subjects: accelerated testing; Degradation; Extrapolation; Life estimation; Manufacturing; mixed models; Prototypes; reliability; Reliability engineering; Statistical analysis; Stress; System analysis and design
• ISSN: 2577-0993
• DOI: 10.1109/RAMS48127.2025.10935278

For modern electronic systems, the increase in market competition has led to higher expectations on reliability and long useful life of the system while significantly shortening the development time. These conditions encourage the use of reliability testing under stringent accelerated conditions to speed up failures and select better designs based on their overall reliability. However, reliability prediction for such prototype systems is challenging due to the presence of several significant failure mechanisms and poor manufacturing quality that leads to high unit-to-unit variation during design testing. Effective system design therefore requires isolating the effect of design configurations from unit-to-unit variability. However, simple analysis of failure time data lacks enough information for robust statistical conclusions, which can lead to inaccurate reliability predictions and design choices. We propose a mixed effects multivariate degradation framework to assess the effects of design changes and unit-to-unit variability in the form of fixed effects and variance components, respectively. We illustrate the proposed methodology with an example using accelerated degradation data of multiple configurations of LED systems.