Time-Variant Reliability-Based Optimization for Lifecycle Cost Reduction

Principal Investigator
Zissimos P. Mourelatos (Oakland U.)

University Researchers
Panos Papalambros, Michael Kokkolaras (U. of Michigan)

Industry
Uday Korde (GM Powertrain)
Mike Sheh (Engineous)

Government
David Gorsich, David Lamb (TARDEC)

Postdoctoral Fellow and Student
Jing Li, Yipeng Zuo (Oakland U.)

The objective of this project is to finish the development of a new reliability-based, time-variant optimization method for series systems, using first order reliability methods, in order to reduce lifecycle cost. A random process (not a random variable) approach is used to characterize the time-dependent uncertainty. The methodology accounts for dependent failure modes. Only stochastic uncertainty is considered. The lifecycle cost will be minimized, considering the benefit of the design, the acquisition and operation costs, and the cost of failure. The methodology will be demonstrated using powertrain and vehicle applications.

Time-variant reliability analysis and design is a critical area in addressing reliability through time. It is necessary for lifecycle cost reduction studies. Furthermore, the time-dependent metamodeling techniques have a wide range of applications beyond the time-variant reliability analysis and lifecycle cost reduction. They can be easily used to characterize and propagate time-dependent uncertainty for both linear and nonlinear systems, using experimental and/or analytical data.