Machine learning and combining data with physics also is useful in making more accurate models.
And time is of the essence.
"You need a tire model that will respond quickly because you can't run a bunch of simulations that will run overnight to tell me what will happen if I turn the steering wheel," Ebbott said. "It has to happen in real time."
Goodyear uses CD Tire3d/RT technology, which is similar to other offerings on the market. It can be geared toward the ride and handling portions of tire response, and has enough freedom in the full model to give good predictions on force and motion, he said.
"Historically, the way these models have been produced is by training them with testing," he said. "Of course, if you're looking at a prototype you haven't built yet, you need another way of evaluating that. If I have to build and test a prototype tire, that kind of reduces the utility of having all of this modeling put together."
Looking forward, Ebbott said future tire development likely will look different than it does now, he said. It may involve virtual collaboration that includes upfront engineering and target-setting, followed by one or two virtual tire submissions. That would be followed by one physical submission for track evaluation to make sure targets are met, and then ultimately a physical test tire for final tuning.
And all this should lead to even shorter times to market, as a smaller number of physical test tires will need to be built.
"It is difficult to quantify the cost, but on the timing, what took us three years to develop a decade ago we can now accomplish in 18 months," he said.
"And that figure should continue to decrease, although as our capabilities improve, the questions also continue to get more difficult as the market and demands continue to become more complex."