Sequential testing decisions in product development

Abstract

Timing and frequency of reliability tests is essential in product design. Testing is modeled as an activity that generates information about technical errors and customer-need related problems that will need to be redesigned. Optimal testing strategies (number and timing of tests) need to balance the tradeoff between several variables, including the cost of a test, the increasing cost of redesign when discovered at a later stage, and the relationship between sequential tests. We propose an optimal sequential testing policy based on the condition that the net marginal value obtained from a test must be higher than the cost associated with it. The optimal strategy is obtained for a general form of accumulated errors and redesign functions. The relation between design errors and rework cost has an important role on determining the optimal testing policy. Linear relation has the most number of tests compared to concave or convex. In addition, earlier tests during the project are preferred for a concave relationship. In convex relation, however, late tests are recommended.