Cooper Standard Automotive Inc. found an opportunity most research and development departments dream about.
Little investment, quick turnaround and massive time savings.
That’s what the firm accomplished with its proprietary artificial intelligence-based polymer compound development software. According to Chris Couch, Cooper’s vice president of innovation and product groups, the technology has led to an 80 percent reduction in the time it takes for the firm to develop new materials.
The technology was recently named a 2019 Automotive News PACE Award finalist. Automotive News is a sister publication of Rubber & Plastics News, but RPN is not involved with the judging process of the award.
“This is another example of us investing in material science to develop superior products,” Couch said. “This time it’s in the process we use to develop those, not the materials themselves, but it’s equally relevant.”
It is the second straight year Cooper Standard has been honored by the PACE Awards, winning one in 2018 for its Fortrex compound.
“It’s evident that our innovation culture is real,” Couch said. “It’s evidence that our innovation culture is delivering results for the business and results for our customers. The company strategy is focused on innovation. We’ve been investing in that and we will continue to invest in it. Last year was a story about a specific material and this year is a story about the process we use to develop materials. They’re both relevant and both meaningful to our business and customers.”
The firm entered this project with a clear objective: To reduce the time it took to bring new materials to the market. The old way involved trial and error, testing and repeating a process until the material desired was good enough for the customer. Couch said the number of repeats range from 40 to 120, which can easily span months and in some cases more than a year.
Cooper Standard sought to take the desired inputs—their recipes—and match them with desired outputs, or properties. The AI system would then learn the relationship between the recipes and the properties and guide its chemists to the new material.
Couch said using AI for polymer development is not a new concept—his research went back to the 1990s—but those experiments didn’t yield any breakthroughs.
“It fell out of favor because nobody really got value out of it,” Couch said. “But it’s now 2018, not 1990. The computers are faster, the data is better and the algorithms are better. So we decided to give it a try and see if we could make it work. It was very effective.”
The firm brought in outside AI experts, and Couch said from there things moved at lightning speed. It took about 12 weeks to launch its first project, then about another 12 weeks to get though that. He added that the cost for this project was “very low.”
The results, on the other hand, were very high. Cooper Standard ran a one-to-one test, giving a complex compound development problem to a team of chemists to approach the traditional way, and then used the AI in parallel. The firm found an 80 percent reduction of trial and error loops with the software.
Not only was the AI faster, but Couch said the end result was also superior. He stressed that the tool is not designed to replace its innovation team, but rather be used to enhance their abilities by freeing up more time.
So far the software has been applied to seven projects, and Couch said the AI is consistently bringing the same kind of improvement of up to an 80 percent reduction compared to the past trials.
“We’re really excited about these first projects,” Couch said. “It’s no longer just a one-off project. Our chemists are embracing this way of using the tool and are really starting to use it as a core part of our material development process and philosophy.”