System-Level Reliability-Based Design Optimization and Robust Design Optimization of Highly Nonlinear Systems with Correlated Input Variables

Principal Investigator
K.K. Choi (U. of Iowa)

University Researchers
Edwin Hardee, Liu Du (U. of Iowa)

Industry
Lingyun Pan (Caterpillar)
Ren-Jye Yang (Ford Motor Company)
Tom Curry, Paul Weal (LMS)

Government
David Gorsich, Wesley Bylsma, Paul Decker, David Lamb (TARDEC)

Students
Ik Jin Lee, Yoojeong Noh (U. of Iowa)

It is well known that the first-order reliability method (FORM) is not accurate to predict reliability when the performance measure of a multi-dimensional (i.e., more than five random variables) design problem is a nonlinear function of the input random variables. In addition, for many structural Reliability-Based Design Optimization (RBDO) problems, the input random variables such as the material properties, fatigue properties, etc. are often correlated. To solve the RBDO problems with correlated input variables, it is necessary to obtain a joint PDF or CDF of input variables. However, it is very much challenging in real engineering design problems to identify correctly the joint probability density function (PDF) or commulative density function (CDF) of the correlated input variables using limited experimental data. Finally, it is very well known that the RBDO result of the problem with correlated input variables is significantly different from the RBDO result of the problem with independent input variables.

The objectives of this project are to develop a system-level Reliability-Based Design Optimization (RBDO) and Robust Design Optimization (RDO) methods for the multi-dimensional system with highly nonlinear performance measures and/or correlated random input variables, so that it can be applied to multi-disciplinary automotive design optimization for broader performance measures such as durability, NVH, crashworthiness, and other design concerns in Army and industry applications.

Along with the results of the accompanying the University of Iowa project “Development of Parallelized RBDO Software System on TARDEC HPC for Durability Optimization of HMMWV System,” the results of this project will be able to support Army’s Reliability-Centered maintenance (RCM) and Condition-Based maintenance (CBM) by identifying locations where the sensors should be mounted (i.e., durability hot spots) for effective data collection.