Reliability analysis from field data and prediction models for customer risk assessments: Case studies and strategy

Abstract

In the automotive industry, risk analysis are performed as soon as quality incidents with the same defect signature are reported from the field. The tools used may be the defect distribution modeling, used for the failure rate estimations required by ISO26262 standard, but adapted to the stronger constraints of quality issues. One of these constraints is the low number of failing parts on which a modeling has to be performed. Some case studies seemed to indicate that it is possible to assert the assumption of a constant failure rate when there are a small number of defects. A novel strategy to assess risk has been developed from this number of defects.