John Donovan

Research has indicated that delays in product introdoction significantly redusce ompany's profit. The current Sequential Probability Ratio Test (SPRT) method of assessing product reliability prior to introduction is analysed. The results of previous reliability demonstration lcsts support the hypothesis that SPRT is no longer appropriate. Its statistical basis is questioirable and it introduces dyfunctional behaviour in project teams. A more coherent approach is advocated which emphasises reliability and quality improvenlent rather than measurement. This paper proposes a SURGE (Stress Unveiled Reliabilily Growth Enhancen~ent) process. Emphasis is placed on the control of the development process. Monitoring is performed by appropriate stress testing designed La precipitate all potential defects. By siniultaneouslv correcting both product and process defects, quality is increased atid more reliable products are developed. A new reliability growtli model and stopping rule are introduced lo assess this growth in reliability and deterniine when tuting should terminate. Simulation has shown that it provides a closer fit to failure data than existing models, while also reducing the inlpacl that reliability testing has an the crilical path. The SURGE process has been validated using three products. Development times were reduced by fourteen percelll while reducing voltlllle ramp up by fifty percent.