Abstract:
In the recent decades, modelling multiple failure data which was a largely neglected analytical issue until the development of sophisticated techniques has gained a greater enthusiasm among the statisticians, which, can be advocated to address the precise nature of multiple failure type data. This study employs methodologies for the analysis of multiple failure modes for a dataset with two tyre categories namely Cross ply and Radial. Each tyre category had two types of tyres and five distinct modes of failures. Type of tyre and mode of failure are considered as explanatory variables. The lifetime (time to failure) of the tyres, which is the core response of interest, was modelled as a function of the explanatory variables and the fitted models were used to assess the validity of the prevailing warranty period. Two approaches of parametric survival regression were used for regressing the lifetime as a function of the failure mode and tyre type which were the explanatory variables considered in the study. Under the first approach, one model each was fitted for Cross ply tyres and Radial tyres treating all failure modes together, whereas in the second approach, one model for each failure mode of each tyre category was fitted. Reliability estimates for lifetime were obtained and compared across the two approaches of model fitting. A significant difference among the two approaches was not seen with respect to survival estimates. However, with respect to residual analysis, accommodate data dependencies and provision for adjusting the shape flexibility to parameter can be regarded as improvements of the second approach, which fitted separate models for each failure mode. In conclusion, the methods deployed in this study enable better decision making with respect to warranty setting and identifying failure modes and failure timings precisely.