PREDICTION OF WARRANTY PERFORMANCE UNDER TWO-DIMENSIONAL COVERAGE

Automobiles arc often warranted for a certain time and mileage. whichever comes first. Any failures occurring in the two-dimensional warranty window will be repaired at the expense of the manufacturers, and thus the failure data are known. Manufacturers usually use these failure data to model and predict the mean cumulative number of warranty repairs for comparing different designs and evaluating the effectiveness of fix in design and production. This paper describes nonparametric estimates of the warranty repair rate and the mean cumulative number of warranty repairs. The relationship between the warranty repair rate and the failure intensity function is established and then employed to estimate the mean cumulative function (MCF). A nonhomogeneous Poisson process (NHPP) is used to model the failure process. The model parameters are estimated using the maximum likelihood method, and the confidence intervals for the model parameters and other quantities are developed. Using the NHPP model, we predict the warranty repair rate and the mean cumulative number of warranty repairs at the end of warranty period. An automotive engine example is presented to illustrate and validated the developed method.