Track 1 Session 3
1:00 to 2:00 p.m. Wednesday March 26, 2008

How Can Failure Rate Prediction Be Made More Accurate?

Failure Rate Prediction (FRP) is the most commonly used measure for product reliability. There exist a variety of standards and tools to predict failure rates or Mean-Time-Between-Failures (MTBF). However, such standards and default component failure rate data are conservative and therefore much worse than the actual field reliability observed based on field returned data. This presentation shows how failure rate prediction can be made more accurate and the results more realistic. The following key areas will be discussed:

- Standards and Tools
- Supplier Data
- Design Derating Data
- Design Stress Data
- Field Data
- Service Life Estimation
- A View of the Future

The lesson learned is that failure rate prediction is not just calculating a number. It is a highly sensitive measure and the process behind shall always be aware of the possible impact on life cycle cost in many areas of the product life. Inaccurately predicted reliability may lead to higher costs (e.g. high supplier warranty and field maintenance), which could be avoided.

Key Words: Failure Rate Prediction, MTBF, Service Life, FMEA, Reliability Standard

Wilfried Kallert Emerson Network Power Germany