Aug. 18, 1998 Cincinnati -- Guardrails designed and built to protect the cars and trucks rolling off assembly lines several decades ago aren't always effective at protecting drivers and passengers inside the sports utility vehicles and mini-vans on the road today.
So, engineers at the University of Cincinnati are turning to the power of supercomputers to help. Ala Tabiei, assistant professor in the department of aerospace engineering and engineering mechanics, and graduate students Jin Wu and Alan McGowan are using the latest techniques available at the Ohio Supercomputer Center in Columbus to develop simulations of full- scale guardrail crash tests.
Supported in part by a grant from the Federal Highway Administration, Tabiei's work makes it easier to test a number of different designs and variables before the expensive crash testing takes place.
"I can change the angle of impact. I can increase the speed of impact ... change the vheicle. I can change the posts. I can change the spacing of the posts. I can recommend other answers that improve the crash-worthiness of this system," said Tabiei, noting that each one of those tests on a full-scale system would cost roughly $100,000.
Engineers at the Federal Highway Administration are concerned about the increasing number of light trucks, mini-vans and sports utility vehicles on the highway. The vehicles are taller which makes it more likely that the wheels will snag on the guardrail.
"Snagging will cause the vehicle to roll over upon impact," said Tabiei. "This system was designed to redirect the vehicle and absborb the energy of the vehicle until it stops. This is not happening all the time, because this system was designed 20, 30 years ago, and the vehicles then were different than the vehicles today."
Tabiei, McGowan, and Jin Wu are looking at a particular type of guard rail system known as a "strong post" system. It is used along 150,000 miles of highway in this country. Then, the engineers use actual experimental results from crash tests run at the Texas Transportation Institute and build a computer version of those tests.
"I take these data and build a model similar to the test set- up, exactly what they have. The initial conditions, everything, is the same. So, this truck is given a velocity of 60 miles per hour at an angle of 25 degrees and impacting this rail," said Tabiei pointing out the details on a computer screen.
The original simulation shows a pickup crashing into a guardrail. A short movie of the crash simulation can be seen on the World Wide Web at http://www.ase.uc.edu/~atabiei or at http://www.osc.edu/Research/SC97_web/Tabiei_SC97.html. The truck has all the components of the real truck. Even the engine is modeled, and virtual accelerometers can measure the forces experienced by those inside the truck.
The next phase of the research is validating the model, both qualitatively and quantitatively. Frame by frame, the supercomputer simulation must match the actual crash test results. Tabiei said that qualitative validation is relatively simple compared with the quantitative part.
"The most important thing is the quantitative validation," said Tabiei. "They collect acceleration data throughout the test. I have to match the data from the simulation with the data from the test. Once I get my acceleration to match, I can say my model is validated. My model is capturing reality."
Using models developed by other researchers, Tabiei will then be able to simulate crashes using a variety of vehicles that vary by weight and size. "We have different classes of vehicles, and this [guardrail] system must work for all of them," emphasized Tabiei.
Tabiei presented preliminary results from the project during the November 1997 meeting of the American Society of Mechanical Engineering. He uses a special computer code known as LS-DYNA 3D, originally developed for defense purposes, and will discuss that part of the work during the LS-Dyna Conference in Detroit, Michigan in September of this year.
Other social bookmarking and sharing tools:
Note: Materials may be edited for content and length. For further information, please contact the source cited above.
Note: If no author is given, the source is cited instead.