Numerical Analysis Seminar
Basic research for reliability-analysis and reliability-based design optimization (RBDO) methods have been carried out at the University of Iowa in collaboration with the Automotive Research Center (ARC), which is funded by the U.S. Army TARDEC. Comprehensive capabilities are developed including: (1) input distribution modeling for both independent and correlated variables using input test data; (2) variable screening method for high dimensional problems; (3) performance measure approach (PMA) and dimension reduction method (DRM)-based PMA for sensitivity-based RBDO; (4) dynamic Kriging (DKG) and local window method (to mitigate curse-of-dimension) for sampling-based RBDO; (5) confidence-based uncertainty quantification (UQ) and reliability assessment for insufficient input data; and (6) simulation model validation using insufficient output test data for confidence-based UQ and reliability assessment. Some examples are presented to demonstrate effectiveness and accuracy of the developed methods. With successful research and development of reliability analysis and RBDO methods, the Iowa team established a small start-up company to develop a commercial Reliability Analysis & Multidisciplinary Design Optimization (RAMDO) software.