Joint modeling of laboratory and field data with application to warranty prediction for highly reliable products

ABSTRACT To achieve a successful warranty management program, a good prediction of a product's field return rate during the warranty period is essential. This study aims to make field return rate predictions for a particular scenario, the one where multiple products have a similar design and discrete-type laboratory data together with continuous-type field data is available for each product. We build a hierarchical model to link the laboratory and field data on failure. The efficient sharing of information among products means that the proposed method generally provides a more stable laboratory summary for each individual product, especially for those cases with few or even no failures during the laboratory testing stage. Furthermore, a real case study is used to verify the proposed method. It is shown that the proposed method provides a better connection between laboratory reliability and field reliability, and this leads to a significant improvement in the estimated field return rate.