General Motors says virtual design will save time, cut costs | Rubber News

FINDLAY, Ohio—General Motors is working to boost the amount of virtual design, development and validation that goes into its vehicles, and is looking for help from its suppliers to meet its goals.

GM doesn’t have plans to shift its entire development system to virtual at this point, but it “looks at a range of tools and innovations to further enhance our development capabilities, and that helps make our teams even more productive throughout the process,” a GM spokesman told Rubber News. “We work with our teams and tools to find the right balance of safety, quality and speed to do so.”

The vehicle maker said that program timings are speeding up, and it needs to move faster with development. Besides saving time, GM also expects that moving to more virtual design will help trim costs from its traditional trial-and-error process of testing to get vehicles and their components ready for production.

And working with its supply base will play a big part in the design evolution moving forward.

“We have been partnering with the supply base to identify the gaps and put together the plans to achieve our ... virtual plans,” Matthew Wieczorek, GM engineering group manager, told attendees of a recent conference in Findlay hosted by Endurica L.L.C.

Now Wieczorek’s profile on LinkedIn may show that he’s managing fuel injection systems for the vehicle giant, but he assured the crowd that rubber is still his area of expertise. He spent 13 years at the Clevite Elastomers Group of Tenneco, before leaving that unit in 2012 to be a resident engineer at GM for Tenneco. He then made the jump in 2014 to work full time for GM.

“We have to keep pace with the rest of the industry,” Wieczorek said. “Hardware tests are always kind of limited in scope. You go to prototype vehicles. You go out and you test them. One fails and the other doesn’t, what do you do? Do you react to it? Do you not react to it? What caused the one to fail and not the other?”

Besides, he said that GM will not make hardware and vehicles available for development and testing once it makes the move to virtual.

And by going virtual, the car maker and its supplier partners can represent a broader range of production variations and boost the number of learning and iteration loops in the development process.

Most of the components are developed during the design phase, and that’s also when it’s cheapest to fix them.

“When things are still ideas in our heads and bits on the hard drive, it’s real cheap to change the design of a suspension bushing,” Wieczorek said. “But as soon as we start cutting tools and actually building parts, the costs just start to skyrocket. The closer you get to launch the higher that cost is.”

Accurate durability predictions also are a must, according to the GM official. After all, most parts need to last the life of the vehicle.

“In the past, we’ve seen everything from ‘kinda close’ to ‘wildly inaccurate’ ” in terms of durability predictions, he said. “That gets really hard when you’ve only got virtual data to work on. We all need to get better. General Motors. The supply base. Everybody.”

Wieczorek showcased a study that GM, Endurica and Tenneco have been working on to develop a better system for full load calculations for bushings. He said they could have tested a plain bushing, but they wanted to push further. So they opted for a bushing that would test a number of areas, such as bump stocks, surfaces, shear rates and other variables that tend to trip up the finite element analysis tools.

The test included six degrees of freedom: three forces and three moments. They ended up testing 215 case factors, with each factor having 11 events, or points. That meant a total of 2,750 load cases were examined.

Each of the cases were examined in two ways, predicting when the bushing would crack. Once with Abaqus Explicit software, and the other with an interpolation software developed by Endurica.

Both systems came up with almost identical durability predictions on how many cycles before the bushing cracked.

But the Abaqus tool took 1,445 hours of computing time to come up with its answer—equal to 60 days—while Endurica’s software took just 17 hours to come to the same conclusion.

“Reducing that time down to less than a day lets us simulate more variation within the part,” Wieczorek said. “That’s huge for us.”

He added that suppliers have to do material models, and while GM won’t dictate what software solutions to use, it will be incumbent on the supply base to prove that their solution works.

Hey, Subscriber! Did you know you can access the latest edition of Rubber News online? Check out our latest and greatest digital editions here.

He was very specific in what GM needs from its suppliers.

Accurate static rate curve predictions, both linear and non-linear, are key, along with end-of-travel data to aid in generating vehicle level loads.

Another growing need is for multi-axis measurements and predictions.

“Imagine you apply the brakes and you hit a pothole,” Wieczorek said. “Or you’re in the middle of a turn and you hit a speed bump or a pothole. We need to be able to simulate or at least get the data on how the bushing behaves in those sorts of events.”

Being able to make accurate dynamic rate predictions also is becoming more vital.

“Electric vehicles have a much lower noise floor than an ICE vehicle,” the GM group manager said. “There’s no engine noise to hide behind, so every little noise that gets transmitted up from the ground through the tires, through the suspension is heard by the vehicle occupants.”

Going forward, he said suppliers will need to be able to model dynamic stiffness up to 3 kilohertz.

GM will take this data from suppliers, Wieczorek said, and use it in its multibody dynamic models, which are used to generate loads and provide early vehicle performance data. Loads into the vehicle structure are highly dependent on bushing and mount rates.

He told the Endurica audience of a case study where a virtual model definitely would have helped. In the example, the Mullins Effect was causing extreme input amplitude sensitivity. The loss in spring rate of the stabilizer bar bushing in question was reducing the stabilizer bar system efficiency, bringing it to an unacceptable level.

“General Motors actually ended up changing the size of the stabar right before a launch, which cost a whole bunch of money. It needed bigger bend radii and other things had to move,” Wieczorek said. “If we had been able to model this kind of behavior up front, we would have had a bigger bar from the get-go and we wouldn’t have had to deal with this.”

Editor's note: This article has been updated to reflect that General Motors will not be moving 100 percent of its vehicle design and development to virtual by 2025, as previously stated.

Please enter a valid email address.

Please enter your email address.

Please verify captcha.

Please select at least one newsletter to subscribe.

Rubber News wants to hear from its readers. If you want to express your opinion on a story or issue, email your letter to Editor Bruce Meyer at [email protected].