NAFEMS Americas and Digital Engineering (DE) teamed up (once again) to present CAASE, the (now Virtual) Conference on Advancing Analysis & Simulation in Engineering, on June 16-18, 2020!

CAASE20 brought together the leading visionaries, developers, and practitioners of CAE-related technologies in an open forum, unlike any other, to share experiences, discuss relevant trends, discover common themes, and explore future issues, including:
-What is the future for engineering analysis and simulation?
-Where will it lead us in the next decade?
-How can designers and engineers realize its full potential?
What are the business, technological, and human enablers that will take past successful developments to new levels in the next ten years?

Resource Abstract

A typical engineering design optimization goal is to find a better performing design compared to a baseline design. Most of the optimization problems involving engineering simulation are nonlinear and might contain a mixture of continuous and discrete design variables which are subject to constraints. Constraints are typically formed as inequalities. Commercial optimization software tools on the market do not clearly differentiate between expensive and cheap constraints. Expensive constraints depend on simulation outputs and cheap constraints are easy-to-compute mathematical or logical relationships between design variables. Cheap constraints must be satisfied before expensive simulation calls; satisfying these cheap constraints heavily depends on the initial method of sampling.



Expensive constraints are the desired performance criteria set at the beginning of an optimization run. Since there is no a priori knowledge about the relationship between design variables and simulation outputs, how expensive constraints are handled is critically important for the success of the optimization in a given run time budget. Any design point that violates an expensive constraint is considered as an infeasible design and often discarded. This point, however, entails important information that may be useful for the selection of the next design point.



Penalty methods are widely accepted and used for constrained handling for various optimization algorithms. The main idea is to add constraint violation as a penalty function to the objective function value to direct the search into a feasible region. From a practicing engineer’s stand point, a constrained optimization algorithm’s performance could be measured as how many iterations are required to find the feasible space and how good the optimum is in a given iteration/time budget.



This talk will first present our recent research results on developing an adaptive aggregation-based approach (SAKS-TRO) for expensive constraint handling for simulation-based optimization. Then a benchmark mathematical problem will be used to demonstrate the graduate sampling redirection behavior of the expensive constraint in OASIS software. Such a strategy resamples new points that close or on the boundaries of the feasible space with frugal computational costs.



Real-world applications including slurry pump design optimization and crash simulation for automobiles will be given and discussed at the end of this talk.