08 Handling Multiple Failure Modes & Exercise

Data* is available for 30 electronic devices that were installed in typical service environments. Two failure modes have been observed. “Spike” failures were caused by an accumulation of randomly occurring damage from power-line voltage spikes resulting in electronic component failure. “Wear” failures were caused by normal product wear and began to occur after 100,000 cycles of use. The data is found in Dataset_5.MTW.

1. Set up the data and identify distributions that best describe each failure mode (Note: this step may be skipped if desired. For the following steps use: lognormal for “spike” and smallest extreme value for “wear”)

2. Generate the reliability curve for each failure mode separately as well as the reliability curve for the device that includes both failure modes

3. What is the system reliability at 100,000 cycles? What is the 95% lower confidence bound on R(100,000)?

4. Analyze the data without segregating the failure modes. At what times to the model predictions differ the most?

*From Statistical Methods for Reliability Data, Meeker & Escobar p.106