OMSP: Failure detection based on small field part and data volumes

Abstract

Increasing functionality and complexity of technical products result in complex damage symptoms and failure modes within the customer use phase. Especially in automotive industry, complex damage symptoms are often the result of multiple failure modes. Therefore the importance of continuous field product observation and field data analysis is an important way for analyzing product reliability in the use phase. For the manufacturer, the goals are the accurate and economic identification of possible failure modes and the knowledge of the product failure behavior based on field complaints at an early stage after product market launch. In addition to statistical reliability analysis based on field data (e.g. Weibull distribution analysis, RAW Concept), the technical analysis of damaged field components is an essential point for the detection and verification of individual failure modes. Especially for this technical analysis, the manufacturer needs damaged field components, which represent the whole supposed failure spectrum in the field. The goals for the manufacturer include: (1) Early and detailed detection and identification of critical failure modes, with the objective of targeted response to critical failure modes (2) Conjunction of requirements for the determination of the regress rate and detection of the critical failures in a comprehensive approach (cf. chapter 2) (3) Optimization of economic aspects of the reliability analysis in terms of sampling procedure, sampling analysis and technical analysis costs (4) Integration into existing technical analysis processes to establish and support an industry-specific standard The temporal aspect results in reduced field monitoring periods and therefore results in small damaged field part and data volumes. This requirement restricts the use of parametrical statistical methods and requires the use of nonparametric statistical methods. The results of field data and technical analysis generate the basis for a targeted roll out of further actions, for example field failure rectification or product optimization. Moreover the initiation of concentrated development-approaches /-strategies for failure prevention with respect to the subsequent product generations (e.g.: COP strategy) is feasible. An industry-wide or cross-industry approach for economic and optimized sampling procedures of damaged components out of the field does not exist. The Chair of Safety Engineering / Risk Management at the University of Wuppertal - in cooperation with manufacturers of the automotive industry - developed the Optimized Multi-Stage Sampling Procedures (OMSP) concept. The OMSP concept is a proposal regarding to analyzing standard for the automotive industry. Key aspects of the OMSP concept are as follows: (1) Early identification and analysis of critical failure modes with reduced amount of analyzed damaged components but with similar detection rate and resolution accuracy (2) Reducing costs of the technical analysis reduce the scope of analyzed damaged components with comparable detection rate of critical failure modes (3) Deselection of selected data areas to reduce the amount of data and the request of considered damaged components allows the verification of the reliability of technical component changes while drawing constant sample sizes (4) Recognition of critical failure modes allows targeted actions for troubleshooting, for example in the field or in the current product generation (5) Integration of the OMSP concept in the FDA process allows the verification of the potential failure modes detected by the statistical reliability analysis This paper outlines the effectiveness and use of the OMSP concept in a near reality case study of the automotive industry. The focus of the case study is the analysis of a shift by wire actuator module (consisting of an electric motor and electrical control unit) including different failure modes. The application of the OMSP concept shows the following essential results: On the one hand, a 50% reduction of requested damaged field components regarding to the technical analysis. On the other hand, the detection rate of critical failure modes based on the reduced analyzed field parts is almost the same in comparison to a full technical analysis of all damaged field components.