Featuring developments in Federal highway policies, programs, and research and technology.
|This magazine is an archived publication and may contain dated technical, contact, and link information.|
|Federal Highway Administration > Publications > Public Roads > Vol. 64 · No. 4 > LS-DYNA: A Computer Modeling Success Story|
LS-DYNA: A Computer Modeling Success Story
by John D. Reid, Martin W. Hargrave, and S. Lawrence Paulson
In September 1998, what had seemed like an open road to federal approval for the bullnose guardrail system suddenly developed a major barrier.
The bullnose system is one of three guardrail types that have been traditionally used to protect against median hazards such as a bridge support. The U-shaped bullnose guardrail wraps around the hazard. State highway departments like the bullnose guardrail because it is considered an effective median safety device and because compared to crash cushions (rigid barriers with cushions on each end) and open guardrail systems, it is relatively inexpensive.
However, before the bullnose system could be used on federal-aid highways, it had to meet the crash-test requirements of National Cooperative Highway Research Program (NCHRP) Report 350, also known as "Recommended Procedures for the Safety Performance Evaluation of Highway Features." The report was adopted by the Federal Highway Administration (FHWA) as a required standard for roadside safety features such as guardrails. Report 350 recognizes the growing popularity of light trucks and sport utility vehicles, which are heavier and higher off the ground than cars, and specifies that crash tests must include light trucks - up to 2,000 kilograms (4,400 pounds) - as well as passenger cars.
In 1997, the Midwest Roadside Safety Facility began a program to develop a bullnose guardrail system that would meet the requirements of Report 350. To pass the crash tests, the system had to deflect - or in the case of head-on collisions, "capture" or "trap" - vehicles hurtling into the barrier at speeds of 100 kilometers per hour (62 miles per hour).
The first two crash tests conducted by the testing facility, involving head-on collisions, had mixed results. The barrier captured a small passenger car, but a small truck plunged right through the rail. A follow-up test had the same results.
The project engineers decided to enlist the help of LS-DYNA, a complex computer analysis system whose predecessor, DYNA3D, was originally developed in the 1970s at the Lawrence Livermore National Laboratory to simulate underground nuclear tests and determine the vulnerability of underground bunkers to strikes by nuclear missiles. LS-DYNA, which uses nonlinear impact finite element code to simulate vehicle crashes, allowed engineers at the University of Nebraska-Lincoln (UNL) Center of Excellence, where the simulations were run, to re-create the head-on collision and analyze the elements of the crash - about 10,000 of them - in an attempt to determine what caused the failures. (See "It's a Jungle Out There: Using the Bullnose Guardrail to Protect the Elephant Traps," Public Roads, January/February 1999.)
LS-DYNA helped engineers find the culprit in the barrier design: longitudinal slots cut into the depressions of the three-hump beams, known as thrie beams, that constitute the guardrail. The simulations showed that the guardrails ruptured because of stresses in the top two humps of the thrie beams. The solution was to reinforce the thrie beams with two cables, a successful design change that was confirmed by a later field test that showed that the reinforced barrier withstood the collision and provided protection for the truck's occupants.
Engineers were very optimistic going into the next test - a light truck hitting the guardrail's critical impact point, which is the point where it is not known whether the barrier will trap the vehicle or redirect it. The general feeling was that the cable reinforcement of the guardrail had solved the barrier system's design problems. This was going to be a no-brainer; the system would pass.
Because the engineers were under some time constraints and they were so confident of success, they did not conduct a simulation of the critical impact-point test, known as Test 6. They ran the crash test and were surprised when the test was a failure because the truck overrode the barrier system.
The vehicle was neither redirected nor trapped. Actually, it was launched by the barrier. In the words of the official report on the test, "Vehicle trajectory behind the test article was unacceptable as the test vehicle vaulted and became airborne in the median area behind the bullnose." And when the vehicle hit the ground, it rolled over.
It was time to go back to the drawing board - or, rather, back to LS-DYNA.
LS-DYNA to the Rescue
Before running another crash test, the researchers at the UNL Center of Excellence began to do some simulations. However, simulation of the critical impact-point test turned out to be extremely difficult and time-consuming because the nature of the impact was much different than previously simulated crashes. "There were a lot of things we hadn't taken into account because it was so much different than a frontal impact," said one researcher.
Concerned about further delays, the Center of Excellence engineers decided to forgo a full simulation. They came up with a design for Test 7 that they thought would work, but because of the simulation problems that they experienced, they didn't have a detailed simulation to verify the design. To further hedge their bets, the engineers made four modifications to the guardrail design - mainly involving the posts holding up the guardrail - that they thought would strengthen the system. However, Test 7 was also a failure.
At that point, the UNL engineers knew that the only prudent course of action was to seek the approval of the project sponsors to allow more time so that the design could be studied in far more detail. The sponsors agreed.
The issue seemed relatively straightforward. As the official report on the test stated, "The lack of tension and lateral resistance allowed the pickup truck to penetrate into the guardrail with increased rail deflection and rotation and without the vehicle being captured or redirected. This combination turned the guardrail into an effective ramp for the impacting pickup truck to climb up and roll over. As a result of the failed test, design changes were necessary to allow the successful containment or redirection of the pickup truck. The thrie beam rail would need to remain upright and functional long enough to capture the front of the impacting vehicle, thus preventing vehicle climbing, vaulting, and rollover. The changes required that the rail tension and lateral stiffness be increased without adversely affecting the head-on impact performance of either the pickup truck or small car impacts."
Finding the solution, however, proved to be a complex and painstaking process that caused a delay of several months in the project. This was primarily because the frontal impact simulation that had already been developed for LS-DYNA had to be converted into a simulation of the critical impact-point collision. This required a major modeling effort.
Once the new model was completed, the center ran a simulation of the failed Test 7. The simulation showed that one tire was hitting the ground line strut and causing the vehicle to vault. Then, the center's engineers ran the simulation without a ground line strut, and it made a big difference in the design. So, they figured out how to make the bullnose without a ground line strut.
The engineers admit that it is unlikely that they would have singled out the ground line strut for attention without the computer simulation. However, they did test and verify the value of other changes.
For example, the research team thought chamfered [grooved] blockouts would help the rail go under and capture the car better. The blockouts worked in the simulation, and they were incorporated into the design. Other changes, all of which were tested in simulations, included a decrease in the distance between posts for a portion of the guardrail system to add strength and the addition of double blockouts to reduce tire snag and hold the rail higher for a longer period of time as the post rotates during impact.
The simulations weren't fully predictive. Some simulation problems remained and prohibited a complete simulation run. However, the engineers were eager to confirm the value of the changes that were substantiated by the simulations, and so, they moved ahead with Test 8 in September 1999.
The official report summarized Test 8 by noting, "The bullnose barrier successfully contained and stopped the test vehicle in a controlled manner. ... The vehicle remained upright during and after collision, and the vehicle's trajectory did not intrude into adjacent traffic lanes. Vehicle trajectory behind the test article was acceptable as the test vehicle was captured in the median area behind the bullnose." It was, in other words, an absolute success. And the bullnose guardrail system, having passed all of its Report 350 tests, is now awaiting final FHWA approval for use on federal-aid highways.
Using LS-DYNA and DYNA3D Code
As the bullnose guardrail experience makes clear, LS-DYNA simulations are not perfect; there is still a lot of trial and error involved in analyzing complex events and identifying causes and effects. However, without computer modeling, the only way to test possible modifications after a failed test is by running another actual crash test, and at between $15,000 and $25,000 per test, that's hardly cost-effective. It is apparent that performing iterative crash tests without modeling can easily become a prohibitively expensive exercise.
Another major advantage of LS-DYNA simulations is that specific factors of a crash - for example, a wheel hitting part of a highway barrier - can be isolated and examined. The computer system is perfectly suited for examining "what-if" scenarios that simply cannot be tested under real-life conditions and for identifying potential problems, such as the ground line strut, that may not be discovered without the computer.
Nevertheless, LS-DYNA is definitely not a user-friendly program. The actual physics itself is so complicated that you cannot expect the software to be any less complicated. The DYNA3D code is very powerful, but it is also very complex. An expert user of DYNA3D has spent an extraordinary amount of time just learning to use the code. It requires an expertise developed by mastering a number of courses in nonlinear computational mechanics; solid mechanics; and fluid flow, including both noncompressible and compressible liquids.
When FHWA got into DYNA3D, FHWA engineers thought that they could help develop this code and then give it to the state highway departments. However, over a period of time, they determined that this just was not possible. Understanding DYNA3D requires a level of dedicated study (and the subsequent development of expertise) that exceeds the resources and capabilities of even a very good engineer at a state highway department or at a private company that manufactures and markets roadside safety structures.
Centers of Excellence
This special expertise in DYNA3D is what makes the centers of excellence such valuable resources.
There are four centers with LS-DYNA capabilities: UNL, Texas A&M, Worcester Polytechnic Institute, and the University of Cincinnati. To date, 10 states - Iowa, Kansas, Minnesota, Missouri, Nebraska, Ohio, Pennsylvania, South Dakota, Texas, and Wisconsin - have worked with the centers on a pay-as-you-go basis on the design or redesign of roadway safety structures.
A fifth center, at the Ashburn, Va., campus of The George Washington University, houses the National Crash Analysis Center.
Besides the states, county and local departments of transportation and the manufacturers and marketers of roadside safety structures can also contract with the centers of excellence for tests and analysis.
The relationship between the centers of excellence and the universities with which they are affiliated is mutually beneficial. In addition to providing services, the centers also serve an educational function, and because they are at universities, they involve graduate students in the process. In that way, the base of LS-DYNA expertise is continually being replenished and expanded.
Dr. John D. Reid is an assistant professor of mechanical engineering at the University of Nebraska-Lincoln (UNL). He is also the director of the DYNA3D Center of Excellence at UNL. Before joining the faculty at UNL in 1993, he worked at General Motors Corp. for eight years -- the last three in safety and crashworthiness. He received his bachelor's degree, master's degree, and doctorate in mechanical engineering from Michigan State University.
Martin W. Hargrave is a research mechanical engineer on the Roadside Team in FHWA's Office of Safety Research and Development. He conducts and manages research associated with FHWA's DYNA3D finite element research program. Before joining FHWA in 1979, he worked for 17 years in various engineering assignments for private sector companies. He received a bachelor's degree in mechanical engineering from the University of Alabama, a master's degree in engineering from Pennsylvania State University, and a master's degree in civil engineering from The Catholic University of America.
S. Lawrence Paulson is a partner in Hoffman Paulson Associates, a writing/editing and public relations firm in Hyattsville, Md. He has written and edited numerous studies for the Federal Highway Administration, Federal Transit Administration, and National Highway Traffic Safety Administration. He also spent seven years covering Congress as the Washington bureau chief of a national daily newspaper, The Oil Daily.
For additional information about LS-DYNA and its role in supporting a new bullnose guardrail design, contact Martin Hargrave at (202) 493-3311 (firstname.lastname@example.org) or John Reid at (402) 472-3084 (email@example.com).
Page Owner: Office of Corporate Research, Technology, and Innovation Management
Scheduled Update: Archive - No Update
Technical Issues: TFHRC.WebMaster@dot.gov