40 Years of Success Stories

From NASA to Pervasive Simulation for Design and Production
Engineering simulation emerged a number of decades ago out of the need to understand processes in which physical measurement was impossible (nuclear industry>structural analysis) or full-scale preliminary tests were not feasible (aerospace>computational fluid dynamics). Leading industries (aerospace, automotive, energy, chemicals) and pioneering companies quickly came to realize the potential benefits of virtual investigations. The first engineering simulation codes were commercialized in the early 1970s.

Rapidly increasing hardware performance quickly enabled a variety of simulation and design of experiment (DOE) techniques that could be performed. The popularity of numerical simulation in the academic and industrial worlds resulted in an explosion of research to develop reliable and robust models able to accurately predict behavior of ever-more complex structures. This trend initially encompassed only single-physics models of advanced fluids, structural dynamics or electromagnetics. The market quickly perceived the value of combining different physics into single models, just as reality encompasses multiple domains. Multiprocessor and multi-core computers opened the door to parametric simulation (DOE) and optimization, empowering designers to go beyond prediction of complex system behavior. Simulation became a tool used to test future solutions against any number of scenarios that could be experienced during a product’s lifecycle — a technology that could systematically explore a range of solutions to identify the best one. Engineering simulation moved far beyond research to development, production and manufacturing. 

While all industries are not at the same computational maturity, they are all following the same path of progressively adopting simulation to model components or entire systems, to challenge existing solutions via what-if scenarios, and to virtually optimize a process before deploying it across the company. With the increasing power of high-performance computing (HPC), designers are embarking on including the probabilistic nature of materials and processes. Specific material properties are never the same, but they vary slightly around an average value. The same observations apply to geometry and operating conditions as well. Although these small variations may sound insignificant, combining them can lead to dramatic consequences. Thus was born the need for design for six sigma (DFSS). Modeling appears to be the only serious option to combine an optimized solution with a probabilistic approach — a concept that literature refers to as robust design optimization (RDO), the new frontier and an important objective.