Robust design methodology for multi-response systems with degrading components

This thesis is motivated by the fact that poor quality always has a monetary cost associated with it. These costs arise from (a) scrap and rework at the manufacturing stage mainly due to variation in manufacturing processes, and (b) increased maintenance, excessive rates of returns, warranty cost, or worse, disappointed and lost customers mainly due to performance degradation caused by environmental conditions at customers' hands. The growing need for a competitive design toot provides a strong reason. The design tool must produce a reliable and robust product satisfying the short and long-term expectations of the customer with low cost and short product development time. In addition, it must provide an efficient way for predictive maintenance augmenting preventive maintenance.; The objective of this thesis is to develop a new design methodology based on a non-sampling approach that tries to improve system performance reliability of a multi-response system with multivariate, uncertain and degrading components. The design methodology provides a way of obviating problems in current approaches, such as (i) computational inefficiency and poor design flexibility in a sampling-based approach, and (ii) inapplicability to multiple system responses in a non-sampling approach. The methodology comprises (a) system performance reliability prediction, and (b) design for performance reliability improvement using a gradient-based optimization method.; For reliability prediction, component degradation models are assumed to be known and the degradation of the system is related to component degradation using mechanistic system models. Selected performance measures (e.g. responses) are related to their specifications by time-variant limit-state functions. System failure is defined as the non-conformance of any response. Therefore, unions of the multiple failure regions are required. For discrete time, set theory establishes the minimum union size needed to identify a true incremental failure region. A cumulative distribution function of time to soft failure is built by summing incremental failure probabilities. A practical implementation of the theory is manifested by approximating the probability of the unions by second-order bounds. Further, for numerical efficiency probabilities are evaluated by First-Order Reliability Method (FORM).; For reliability improvement, the predicted performance reliability for combinations of design parameters provides robustness measures (mean time to soft failure and its variance, reliability level at a planned time), and a monetary measure (present worth of expected failure cost). Design problems to allocate means and tolerances of design variables are formulated into constrained optimization problems wherein, as objective functions, the robustness measures and the monetary measure are optimized using a gradient-based local optimization method.; The main contributions of the thesis are (a) the first performance reliability prediction method for a general, multi-response system with multivariate, uncertain and degrading components using a non-sampling approach, and (b) the first integrated design method (means and tolerances design) based on a non-sampling approach as available in the open literature.