| Alain de Rouvray, CEO and founder of ESI Group and winner of the Spencer Tracy lookalike contest, at ESI Summit in Milpitas, Calif. |
All those who are shocked by the results of the recent presidential election should have listened to Alain de Rouvray.
The president and founder of ESI Group knew something was amiss with the polling that was forecasting Hillary Clinton as the winner. De Rouvray noticed that the polling relied on an unrepresentative sample size, mostly the urbane. The great majority of the U.S. was not being accounted for. When you corrected the predictions for the missing data, the result was Donald Trump. De Rouvray recognized this two days before the election.
It’s not the first time de Rouvray has found an answer that has eluded others. Another, the answer to how materials behave in previously unpredictable situations, led to the company he founded and still leads, ESI Group.
ESI was able to prove that it could successfully simulate seemingly insolvable situations. An early example was protection against solar wind with particles traveling an order of magnitude faster than previously analyzed—a situation that remained unsolvable even by the best minds at Los Alamos National Laboratory.
Now a 1,100-employee, publicly traded firm headquartered in Paris (de Rouvray is French, though he now lives in San Francisco), ESI Group has been anything but a smooth-sailing venture.
“It was a bit rough in the beginning,” said Amy de Rouvray, his French-born, UC Berkeley–educated daughter who heads ESI’s marketing.
De Rouvray had to fight off threats from newcomers whose methods, de Rouvray decided, were a little too similar to his own. After a seven-year legal battle, the courts agreed that this was unfair competition.
The short, 73-year-old de Rouvray will remind you of Spencer Tracy—with an accent. He, along with his family, retains 30 percent of ESI.
ESI is best known for PAM-CRASH, a specialized CAE program that lets auto makers predict what happens to cars—and their passengers—as they meet immovable objects.
“It was a problem nobody else was solving,” said de Rouvray. Or perhaps they could not. The way that sheet metal crumpled and parts cracked and how pieces went flying—it was all beyond the limits of small-deflection, stability and elasticity theories, around which all finite element analysis codes are built.
Classical solutions rely on continuum mechanics, said de Rouvray. That doesn't apply across discontinuities—like across cracks or when pieces fly off.
That was the sort of problem that no one deemed possible of solving. Except de Rouvray.
Once metal analysis was tamed, de Rouvray decided he would solve fabric behavior, specifically how a car’s airbags react. Think of the analysis of fabric that is crumpled into a wad then literally explodes onto a human. How do you solve for that? Impossible, said the entire auto industry, which was spending hundreds of thousands of dollars testing with real bags and fake humans.
“We solved the airbag problem,” said de Rouvray, with characteristic finality.
ESI IC.IDO virtual reality program immerses users into potentially large data files, such as aircraft, trains or assembly lines. (Image courtesy of ESI Group.)
“We were so good at it that Audi stopped testing [for airbags] and relied on ESI simulation,” said de Rouvray. It can run thousands of simulations in the time it would take just to get instrumented for a single crash test, according to de Rouvray.
ESI seems to have built on its expertise with crash test simulations and added a variety of other analyses, including modeling and viewing solutions.
ESI MINESET using a combination of machine learning and visualization to explore simulation results. (Image courtesy of ESI Group.)
De Rouvray is not done solving puzzles that others cannot. At ESI’s latest summit, held in the company’s recently opened Silicon Valley office, I saw several technologies being nurtured or in full bloom. ESI’s IC.IDO (get it?) is VR technology for full 3D immersion into the large models and simulation results. Its MINESET program will let engineers explore large datasets, such as those resulting from multiple simulation runs, to detect patterns and relationships. Last year, ESI acquired CIVITEC for its sensor simulation platform, Pro-SIVIC, which will simulate self-driving cars before or instead of road or track testing.
By incorporating new technologies to solve new problems, ESI Group seems to be continuing to keep up with its overall goal of supplying “virtual prototyping” solutions.