Date on Master's Thesis/Doctoral Dissertation
Usher, John Stephen
Availability demonstration testing; ADT; Consumer risk
Commercial products--Testing; Consumer goods--Testing
Evaluation and demonstration of system performance against specified requirements is an essential element of risk reduction during the design, development, and production phases of a product lifecycle. Typical demonstration testing focuses on reliability and maintainability without consideration for availability. A practical reason considers the fact that demonstration testing for availability cannot be performed until very late in the product lifecycle when production representative units become available and system integration is completed. At this point, the requirement to field the system often takes priority over demonstration of availability performance. Without proper validation testing, the system can be fielded with reduced mission readiness and increased lifecycle cost. The need exists for availability demonstration testing (ADT) with emphasis on managing risk while minimizing the cost to the user. Risk management must ensure a test strategy that adequately considers producer and consumer risk objectives. This research proposes a methodology for ADT that provides managers and decision makers an improved ability to distinguish between high and low availability systems. A new availability demonstration test methodology is defined that provides a useful strategy for the consumer to mitigate significant risk without sacrificing the cost of time to field a product or capability. A surface navy electronic system case study supports the practical implementation of this methodology using no more than a simple spreadsheet tool for numerical analysis. Development of this method required three significant components which add to the existing body of knowledge. The first was a comparative performance assessment of existing ADT strategies to understand if any preferences exist. The next component was the development of an approach for ADT design that effectively considers time constraints on the test duration. The third component was the development of a procedure for an ADT design which provides awareness of risk levels in time-constrained ADT, and offers an evaluation of alternatives to select the best sub-optimal test plan. Comparison of the different ADT strategies utilized a simulation model to evaluate runs specified by a five-factor, full-factorial design of experiments. Analysis of variance verified that ADT strategies are significantly different with respect to output responses quality of decision and timeliness. Analysis revealed that the fixed number of failure ADT strategy has the lowest deviation from estimated producer and consumer risk, the measure of quality. The sequential ADT strategy had an average error 3.5 times larger and fixed test time strategies displayed error rates 8.5 to 12.7 larger than the best. The fixed test time strategies had superior performance in timeliness, measured by average test duration. The sequential strategy took 24% longer on average, and the fixed number of failure strategy took 2.5 times longer on average than the best. The research evaluated the application of a time constraint on ADT, and determined an increase in producer and consumer risk levels results when test duration is limited from its optimal value. It also revealed that substitution of a specified time constraint formatted for a specific test strategy produced a pair of dependent relationships between risk levels and the critical test value. These relationships define alternative test plans and could be analyzed in a cost context to compare and select the low cost alternative test plan. This result led to the specification of a support tool to enable a decision maker to understand changes to a and ß resulting from constraint of test duration, and to make decisions based on the true risk exposure. The output of this process is a time-constrained test plan with known producer and consumer risk levels.
Walker, Randall L., "A cost model for managing producer and consumer risk in availability demonstration testing." (2013). Electronic Theses and Dissertations. Paper 1500.