Study On The Two-Dimensional Data Analysis And Spare Parts Demand Forecastingand Management

In the fierce competing marketplace,the manufacturer of durable products has to not only sell the confirmed products to the customers but also provide warranty to the sold products.The warranty in after-sales procedure is a guarantee provided by the manufacturer to the customers to ensure that the sold products will execute the intended functions in a specified period and under specified using conditions.If a failure occurred in the warranty period,the manufacture has the obligation to repair or replace the failure free of charge to the customers,which means the warranty provides protection to the customers.Based on the differences in the definition of warranty period,warranty is often classified to one-dimensional warranty and two-dimensional warranty.When the failures occur in warranty period,warranty claims are reported to the manufacturer.In the warranty service processes,there are a large amount of data are collected and stored,where there are quality and reliability related information exists.Warranty data analysis(WDA)is a set of methods used to extract useful information from the warranty data.Firstly,the theory and method of product reliability analysis based on warranty data are systematically introduced.Then we studied three questions based on the two-dimensional warrantied products.(1)From the beginning of a new designed product,we study the phenomenon that the reliability grows with the lasted manufacturing days or accumulated amount of products manufactured based on the two-dimensional warranty data.This phenomenon is referred to as learning effects in reliability.Based on warranty dataset collected from an automobile manufacturer,we fitted the warranty data using a log-linear regression model by taking the effects of usage rate on reliability into consideration.Further,some of the reasons to the learning effects in reliability are analyzed simply.The results of this study can improve the accuracy of warranty claims forecasting.Furthermore,the results of this study also benefit the engineer or managers to find the factors that lead to learning effects in reliability and to accumulate the learning procedure.(2)For the two-dimensional warrantied products,the product often deteriorates over both the age and the usage of the products.In the existing works,the usage is often processed as a linear function of the age.Different with the existing works,a stochastic function between the age and usage rate is defined.Furthermore,the marginal distribution function is derived for the age with given usage or the usage with given age.Then a method of failure modeling is proposed based on the bivariate Weibull distribution,which provides new directions of two-dimensional warranty modeling.(3)We study the problem of forecasting on the demand of warranty service spare parts based on warranty data.A usage rate distribution based method is used to predict the number of products in warranty and the method to forecast the demand of warranty service spare parts is proposed considering the effects of usage rate on the reliability of products.As for the effects of usage rate on reliability,both the accelerate failure time(AFT)model and proportional hazard(PH)model are used to simulate the effects.The inventory policy is optimized to minimize the total cost using the dynamic inventory controlling model.Finally,a case study is presented based on the warranty data of a excavator manufacturer.