Software Platform Aims to Reduce Trial and Error in Materials Selection

“Nobody has the time and the money to go and melt hundreds of prototype materials and test them one at a time and say, aha, we found the one that works,” said Jason Sebastian, president of QuesTek, in an interview with Design News. But he said historically this is largely how engineers would tackle materials selection, by using a trial-and-error method.

“Back in the day, materials were kind of the afterthought,” Sebastian said. That is not possible now because today’s engineers have to take a materials concurrency approach among different disciplines. He gave an example of how mechanical, electrical, and automotive engineers might get together to work on a design where they have to consider the stress that will be placed on a material. “But [they] also need to keep in mind the electrical engineering constraints, such as having to route cables through the frame of the automobile,” Sebastian said.

“Concurrency is all about bringing materials scientists to that table,” he continued. “We want to do this with the mechanical engineering, we want to do that with electrical engineering, [but] it doesn’t look like existing materials are going to get there without some kind of a compromise, [so] let’s design a new material. So materials engineering is happening at the same time as electrical and mechanical,” he said.

There could also be a situation where the material is the constraint to the design of a new component, so there may be the need to source an entirely new material to achieve goals. “In this case, we'll help with the qualification of the new material,” Sebastian said. “Maybe it's an old alloy but you're putting it into a new part, or a new kind of service environment and you want to make sure you have a handle on the variability and properties and performance that you would get in that environment.”

Other times, materials can perform differently when used in an entirely new process, such as additive manufacturing, Sebastian said. “So, we're trying to understand how materials behave when subjected to those kinds of new processes, which is almost akin to a new material design because these processes are so different,” he explained. “The material is both its composition, its recipe, but also its processing—how it's made. Either one of those things could be changing when we talk about a new material. An old material with a new heat treatment is a basically a new material in terms of its properties.

“That’s where people can benefit from the software,” he added, referencing QuesTek’s ICMD, which stands for Integrated Computational Materials Design. “It's a software for material scientists to do computational materials science.”

The Future of Materials Selection?

ICMD is built on computational physics models derived from hundreds of materials science engagements QuesTek has completed on behalf of clients such as NASA, SpaceX, Apple, GM, Boeing, and Audi since 1997.  

It uses AI as well as machine learning when beneficial to solving a particular materials challenge, Sesbastian said, but QuesTek ultimately takes a more hybrid approach. AI can detect trends from the data, which is helpful, he explained, but those trends are best augmented by a real physical space understanding. “You’re going to need to really understand the physics and the metallurgy of what’s going on in the materials science to do alloy design, materials qualification, etc.,” he said.

“But it's better, compared with the old trial and error approach, this machine learning that narrows things down and then we use a physics-based understanding to really understand what's going on,” Sebastian concluded. “This is what we do. This is the future.”