Forward and Backward Uncertainty Analysis - Tools for Engineering Design
Gianluca Iaccarino
Stanford University and Cascade Technologies Inc. – USA
Curriculum
Gianluca Iaccarino is the co-founder of Cascade Technologies and has 15 years’ experience in development and application of CFD methods.
He acts as a consultant on a variety of projects involving industrial fluid flow simulations. He is also leading the development of a novel tool for uncertainty quantification and robust design.
Dr. Iaccarino received his PhD from the Politecnico di Bari with a thesis on Immersed Boundary Methods. He holds the position of Assistant Professor in the Mechanical Engineering Department at Stanford University.
Abstract
Uncertainties are inevitable in the design and operation of engineering devices. Recent advances in computational mathematics and statistics are now beginning to enable cost-effective numerical analysis under uncertainties. The classical
viewpoint is to assess the effect of input uncertainties (operating scenario, material properties, etc.)
on the outputs of interest, such as a device performance metric - forward uncertainty analysis. A novel paradigm is the backward analysis in which a tolerance on the outputs is specified a priori while the compatible input uncertainties are the final results. Backward analysis enable to identify the primary aspects of the device that need to be controlled.
The talk will provide a description of both forward and backward methodologies with examples drawn from aerospace and automotive applications. A new software tool devoted to uncertainty analysis is under joint development at Cascade and Enginsoft and will also be briefly introduced.
