Lifetime Estimation Using Only Failure Information From Warranty Database

Knowledge about product lifetime obtained from actual usage (field data) is of great importance to manufacturers who want to get information about the true reliability of their products. Warranty data are frequently used to estimate reliability characteristics because such databases are automatically generated and updated at no additional cost from repair claims during warranty coverage. For engineering purposes, usage time (e.g., mileage) is more relevant, and lifetime parameters measured in usage time is an integral part of reliability analysis using warranty data. Usually, warranty data consist of only failure information. Censored data (e.g., mileage of non-failure automobiles) are not obtainable, of which usage time distributions are different from those of failed ones. Effective usage-based estimation thus requires supplementary information about usage accumulation for non-failure units such as follow-up studies, or a usage time distribution which includes both failure, and non-failure products. However, sometimes usage-based data for non-failed units are expensive and difficult to obtain. Thus, the unavailability of the usage time of censored units makes it difficult to estimate the usage-based lifetime distribution of products. This paper deals with that problem, and discusses how to estimate the lifetime distribution using warranty data which consist of only failure information. The practical consequence of this finding is that supplemental information is not needed to obtain correct estimates of the lifetime parameters.