Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
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 > Using the Computer and DYNA3D to Save Lives

Jan/Feb 2001
Vol. 64 · No. 4

Using the Computer and DYNA3D to Save Lives

by Martin W. Hargrave and David Smith

They occur every day of the year on our city streets, our county roads, our state highways, and our Interstate Highway System. They occur at dawn, during daylight, in twilight, and at night. It can be raining, snowing, foggy, or perfectly clear. "They" are motor vehicle crashes - more than six million each year, resulting in three million human injuries and 42,000 premature deaths each year, every year.

How do we, as a nation, reduce this annual toll? How do we mitigate the injuries and deaths that occur each year on our roadways? A partial answer to this question - at least for run-off-road crashes - may be the expanded use of a revolutionary new motor vehicle crash-analysis tool, DYNA3D.

The Federal Highway Administration (FHWA) has traditionally taken the leading role in reducing the effects of run-off-road crashes. Many of these run-off-road crashes involve a single vehicle leaving the roadway and colliding with a man-made roadside structure or feature. Examples of man-made roadside structures and features include guardrails, roadside signs, overhead lighting supports, roadside ditches, and slopes.

Aggressive driving is the root cause of many of these crashes. Alcohol or drug impairment is often the cause. Sometimes both are factors.

FHWA has a continuing focus on improving these man-made structures and features to reduce the potential for serious or fatal injury to vehicle occupants involved in a collision. Also, continually improving the safety of our nation's roadways is one of FHWA's five strategic goals. Within the past decade, FHWA has led a program focused on employing and expanding the capabilities of using the computer and a new crash-analysis tool, DYNA3D. DYNA3D is a nonlinear finite element code that can be used, in conjunction with the computer, to replicate three-dimensional motor vehicle crashes. When using the computer and DYNA3D, the objective is to make man-made roadside structures and features more crashworthy - that is, to ensure that they are capable of transferring the high-speed collision energy from the colliding vehicle to the structure/feature in a controlled manner so that the vehicle slows or stops, the vehicle remains upright, and the occupants experience minor or no injuries.

This can be accomplished because when the computer and the DYNA3D code are used by highly skilled engineers, a very accurate replication of a real-world vehicle collision can be generated. Thus, with the aid of DYNA3D, the collision is depicted in a virtual manner, and subsequently, it can be slowed down and viewed one frame at a time. When the collision interaction of the vehicle and the man-made roadside structure or feature is less than desired, virtual (computer) changes can be made to that structure or feature, and the collision can then be repeated. This process can be repeated incrementally until a crashworthy result is obtained.

For the past 50 years, costly trial-and-error crash testing has been the only procedure available to develop improved, crashworthy roadside structures and features. However, by using the computer and DYNA3D in this manner, expensive crash testing can be substantially reduced or eliminated.

Because real-world vehicle collision results can be mimicked by the computer, crash testing can be limited and relegated to the role of final proof-of-results determination. However, it cannot be overemphasized that, when using the computer and DYNA3D to replicate real-world motor vehicle crashes, caution must always be used to ensure that the results are accurate.

Use of the computer and DYNA3D provides other benefits as well. Listed below are three primary benefits and one note of caution:

1. Increased Visibility. When using the computer and DYNA3D, it is possible to view the virtual collision and the interaction between the colliding motor vehicle and the roadside safety structure or feature from any angle, any position, one frame at a time. If parts of the vehicle (fenders, bumpers, engine, etc.) prevent observation of the effects of the collision on a particular part of the vehicle, these intervening parts can be "removed" from the picture, one part at a time, until the vehicle contact and its behavior upon striking the structure or traversing the feature can be clearly observed. (The removed vehicle parts are still part of the collision. They have just been virtually removed from the picture.) If, for example, a wheel of the vehicle snags on a guardrail post and is the root cause of an abrupt stoppage of the vehicle's forward motion, this inappropriate - and potentially dangerous - interaction between the vehicle and the guardrail post can be uncovered and clearly viewed.

2. Quantitative Data. Not only can the collision process be clearly observed, but the values of deformations, loads, and stresses (to name a few) of any part of the safety structure can be quantitatively determined at any point in time during the collision process. Engineers are trained to use numbers such as these to determine where and when a structure may fail - for example, where and when a guardrail may fracture during a vehicle collision. If a guardrail fails by allowing the colliding motor vehicle to pass through and behind, the resulting collision with whatever is behind the guardrail can be much, much more dangerous than the collision with the guardrail itself. As an example, a cliff or a steep downward slope may be behind the guardrail, and if the guardrail fails, the resulting fall and impact with the ground at the bottom could be life-threatening.

3. Design Optimization. Computer-generated crashes can be used to determine the potential improvements in the crashworthiness of man-made roadside safety structures under several circumstances: when the structure is made from a different (hopefully improved) material, to compare alternate designs, and to check the effects of slopes or ditches in front of the structure. (Actual crash tests are usually performed only on flat and level terrain.) Using the computer and DYNA3D, roadside safety structures can be improved in a cost-effective, step-by-step manner to obtain the optimum in crashworthy roadside structures, while reducing the need for expensive real-world crash tests.

4. Reality Checks Are Needed. With virtual crashes, spurious results are always possible if the motor vehicle and roadside structure models contain errors or inaccuracies. Therefore, after completing the initial roadside safety structure model on the computer, researchers must test the model by generating a virtual motor vehicle crash into the structure, and the virtual crash must mimic a real-world crash test previously conducted.

"Prior to the use of the computer and DYNA3D, roadside safety structures were crash-tested under very limited conditions because of the expense of testing," said Michael F. Trentacoste, director of FHWA's Office of Safety Research and Development. "These tests were very restricted by the terrain and physical conditions of the test location. Also, only a few vehicle types, speeds, and angles of approach could be used. In addition, researchers were limited to analyzing visual data obtained by high-speed cameras. Things were often missed. Cameras would not show everything that needed to be seen because of poor angles of view or because the motor vehicle obscured critical interactions. In addition, quantitative data associated with the crash were extremely limited, so researchers had to rely a great deal on common sense and personal judgment. The computer and DYNA3D changed all of this."

How FHWA Came to Use Virtual Crash Testing

The DYNA3D code was developed originally by the Lawrence Livermore National Laboratory (LLNL) at Livermore, Calif., in the 1970s to analyze the effects of underground nuclear explosions and the potential for intercontinental ballistic missiles to penetrate hardened silo enclosures. Today, LLNL maintains a version of this code for use by both the laboratory and other authorized independent collaborators.

About 1988, the primary developer of DYNA3D left LLNL to continue development of DYNA3D independently at a new start-up company. That company became Livermore Software Technology Corporation (LSTC). As the developer continued work on DYNA3D, he created a vastly improved version, which he called "LS-DYNA."

LS-DYNA continues to be improved, particularly in the arena of motor vehicle crash analysis, and it is the version that FHWA and its funded researchers use under the generic name "DYNA3D."

"Experts at FHWA were among the first to realize this amazing crash-analysis tool's potential for assisting in the design of improved highway guardrails, bridge supports, signposts, and other roadside safety structures," said Trentacoste.

On Oct. 1, 1992, FHWA and NHTSA jointly established, under a competitive contract, the National Crash Analysis Center. NCAC is operated by the George Washington University at their campus in Ashburn, Va., for the purpose of examining vehicle crashworthiness by using LS-DYNA. Since 1992, this center has served as the primary source for computer-based LS-DYNA motor vehicle models for both FHWA and NHTSA. Many of these NCAC-developed motor vehicle models have been used in FHWA's DYNA3D crash-analysis program. Now, many of these FHWA-funded motor vehicle models are available for use by researchers via NCAC's Web site (www.ncac.gwu.edu).

University Centers of Excellence Make DYNA3D Crash Analysis Available

After the completion of a few initial DYNA3D motor vehicle models by NCAC and other researchers, FHWA in 1994 initiated a set of six cooperative agreements with universities to develop computer-based models of six different roadside safety structures. The desired outcome was to improve the crashworthiness of these roadside safety structures by using DYNA3D. FHWA also intended to make these roadside structure models available to other researchers on an as-needed basis via the Internet.

In 1994, all of the universities contracting with FHWA initiated work using the LLNL version of DYNA3D. However, by 1996, all had independently converted to the LSTC version, LS-DYNA, because of LS-DYNA's superior capabilities.

FHWA researchers quickly appreciated that DYNA3D offers powerful new capabilities for analyzing the multifaceted interactions associated with a motor vehicle crashing into a roadside structure. FHWA wanted to view the complex interactions between the motor vehicle and the roadside structure, ideally at various speeds, and to determine the resulting motions of the vehicle and the deformations of the vehicle's occupant compartment during a collision. FHWA also wanted to better understand and determine the attendant energy-absorption mechanisms that came into play during the collision, both those associated with the motor vehicle and those associated with the impacted structure. DYNA3D provides a much better understanding of the complex crash event than any model previously observed by FHWA researchers.

However, even as engineers marveled at the wonders of DYNA3D, it was evident to FHWA researchers that even more sophisticated models of both the motor vehicles and the roadside safety structures would be needed in the future. And if, in future efforts, the virtual collision and the subsequent trajectory of the vehicle after the collision were to accurately reflect a real-world crash, then even more highly skilled and trained DYNA3D researchers would be required.

Thus, it soon became obvious that DYNA3D was a very sophisticated crash-analysis tool; that even greater skills and abilities would be required by engineers using this tool in the future; and that without a very specialized education, a great deal of training, and a great deal of experience in using DYNA3D, a perfectly capable engineer could not and would not successfully use this new tool.

Over time, this increased understanding of the potential of DYNA3D and the requirements for effectively using this new crash-analysis tool led FHWA to expand the program. In late 1997, four FHWA-funded "centers of excellence" were created to develop new and improved roadside safety structures for other customers, including state and local departments of transportation, roadside safety structure manufacturers, and other federal agencies in need of these services. The centers were also created to expand, over time, the number and the skill level of researchers capable of applying this new crash-analysis tool. The four centers of excellence, established through a competitive contract, are located at Texas A&M University, the University of Nebraska at Lincoln, the University of Cincinnati, and Worcester Polytechnic Institute in Massachusetts. Each of these four well-established engineering institutions had specialized engineers who were willing to invest significant amounts of time and effort to learn how to accurately and effectively apply DYNA3D modeling, using their supercomputers. Each also had a staff of skilled engineers who were committed to enlisting and training students and other faculty members in the use of DYNA3D to improve the crashworthiness of roadside safety structures and features.

The highly experienced staffs at these institutions build DYNA3D models of the safety structures, and then they design, conduct, and analyze controlled virtual collisions using DYNA3D motor vehicle models developed by NCAC. New roadside safety structure models, substructure component models, and other specialized models have been and continue to be created at these centers. Once created, these models are placed in a virtual inventory at a Web site located at NCAC for repeated use, modification, and manipulation as needed or requested.

DYNA3D in Action in Texas

Roger Bligh is a prime example of the outstanding engineers with extensive practical knowledge that these FHWA-sponsored centers of excellence bring to the expanding program of vehicle crash analysis. Bligh is the director of the Center of Excellence for Transportation Computational Mechanics at the Texas Transportation Institute (TTI) located at Texas A&M University. Since 1994, he has been directing efforts to use the computer and DYNA3D to replicate vehicle crashes. His growing experience with the evolving DYNA3D crash-analysis tool is coupled with an even greater experience in roadside structure development and vehicle crash testing at TTI. He also serves as manager of TTI's Highway Safety Structures Program and has been involved in roadside safety structure research and crash testing for the past 14 years.

"The computer simulation research team at our Texas A&M center has been conducting many simulated crash tests per year, and that number may be increased as research requests increase," said Bligh. "Here at TTI, we conduct more than a hundred crash tests per year with each test costing approximately $25,000. Obviously, a savings is realized when the computer and DYNA3D are used to simulate some of these crash tests instead of actually conducting them here at our crash test facility."

All of the centers of excellence perform sponsored research. While FHWA sponsors some of the research being performed on roadside safety structures, particularly roadside barrier testing, the centers also have research grants from and contracts with state agencies and private industry. For example, TTI is conducting a research project in collaboration with the Washington State Department of Transportation (DOT). This sponsored research must be won by the centers on a competitive basis.

"With the experience and working knowledge that our research team has accumulated to date in applying the computer and DYNA3D to vehicle crash analysis, we can be very competitive," Bligh said. "Private industries will find the crash-analysis capabilities of our center to be of great practical value. Furthermore, as awareness of the center's capabilities grows, more proposals for both crash analysis using the computer and DYNA3D and for final proof-of-results crash testing are emerging."

This is equally true of the other centers of excellence. Private industry and the state and local highway agencies are encouraged to take advantage of these resources and the specialized expertise available.

The Texas DOT has been working closely with the Texas A&M center on the development of new roadside safety structures that address particular problems and conditions found in Texas. For example, the center has been working with the state to improve safety on roadway bridges.

"Use of a W-beam tubular bridge railing is the second most common bridge railing on Texas state roads and highways, and our state DOT is working with the center to determine how well this bridge railing performs, from a safety standpoint, when used with a box culvert bridge," Bligh explained. The concept is that when a motor vehicle collides with the bridge rail, the rail will redirect the vehicle, allowing it to remain on the bridge, while the rail itself breaks away from the bridge support, thereby reducing damage to the bridge structure. This also allows the bridge rail to be constructed with less support and at a lower cost. After a great deal of DYNA3D crash analysis, redesign, and some real-world proof-of-result crash tests for verification, a successful design for this bridge rail has now been approved.

illustration of a barrier system
A model of a tie-down portable concrete barrier system.
According to Bligh, another frequent use of the DYNA3D crash-analysis tool involves varying the roadside terrain conditions surrounding roadside structures and then determining the resulting crash performance of these structures. Because Texas is so large and the topography throughout the state varies greatly, the terrain surrounding actual roadside safety structure installations varies considerably. (Examples include steep adjacent grades and unusually narrow curves.) The laboratory topography on which roadside safety structures are crash-tested is completely flat, level, and straight; therefore, it does not duplicate the actual conditions of installation. However, by simulating the vehicle crash conditions using DYNA3D, researchers can virtually re-create these typical topographical conditions. Accordingly, a safety structure can be virtually crash-tested through a series of what Bligh calls "aparametric investigations." These are simulated vehicle crashes in which the terrain conditions are varied prior to each collision to examine the response of the vehicle and the resulting potential for injury to vehicle occupants - that is, the crashworthiness of the roadside structure. This assists the Texas DOT in the development of special installation methods or structural modifications needed for these differing topographical conditions.

Bligh reports that the Texas A&M center has accumulated a library of validated DYNA3D models of different roadside safety structures. These DYNA3D structural models are also made available to other centers of excellence and to other DYNA3D researchers after approval by FHWA. By using this library, the Texas A&M center can respond much faster to a new customer, thereby saving time and money.

In addition, the Texas A&M Center of Excellence maintains a dynamic interchange with NCAC, which develops the DYNA3D motor vehicle models, for occasions when a specific modification of the motor vehicle model is needed to execute a specific crash analysis. NCAC can modify the model to reflect a wheel spinning, changes to a suspension system, or a finer "mesh" of the vehicle model in the area of collision. (DYNA3D is a nonlinear finite element code, and the mesh size refers to the size of the elements that compose the vehicle model.) These specialized vehicle models can be inserted into the DYNA3D model to improve the fidelity of the computer-generated collision in relation to an identical real-world collision. As the capabilities of the center increase, Bligh envisions that many more and varied motor vehicle models, representing the characteristics of vehicles commonly found in Texas, will be used to evaluate and improve the roadside safety structures.

For example, "Collisions involving roadside safety structures and motor vehicles commonly used in Texas, such as full-size pickup trucks and SUVs [sport utility vehicles], may have very different effects and collision results [than] collisions involving only typical automobiles. This is because these vehicles are much different in construction and are higher off of the ground," said Bligh.

DYNA3D in Action in Nebraska

The FHWA-sponsored Center of Excellence for DYNA3D Vehicle Crash Analysis at the University of Nebraska at Lincoln (UNL) serves a somewhat similar role as the Texas A&M center.

John Reid, the director of the UNL center, is also an associate professor of mechanical engineering at the university. Reid has been involved in DYNA3D crash analysis for 10 years - first at General Motors Corp. and subsequently at UNL. Because of his prior education and work experience, Reid brings an extensive understanding of applying the computer and DYNA3D to vehicle crash analysis.

Reid reports that for the past 10 years, eight midwestern states have been pooling research funds to sponsor the development of a varied list of improved roadside safety structures. This work is being accomplished at the Midwest Roadside Safety Facility (MwRSF), which is also located at UNL.

Each year, the participating state DOTs propose projects to be investigated. And each year, an advisory group representing the states works with the engineers at MwRSF to select five of the projects for investigation. The selection criteria are the abilities of the MwRSF staff to do the work successfully and within the budgeted cost.

"The Midwest Roadside Safety Facility and the University of Nebraska Center of Excellence coordinate their respective activities to promote the best use of the talents and experiences of the two professional staffs and to ensure that the roadside safety structures developed here are crashworthy," Reid said.

The research team at the center develops a DYNA3D model of either a component/subassembly of the roadside safety structure under investigation or, on many occasions, the entire roadside safety structure. The team then ensures that these computer-based models replicate either a prior indoor laboratory test or a prior outdoor vehicle crash test, even if that prior test is considered a failure. If the computer-based model does not replicate the prior test, changes that are consistent with real-world physical principles are made to that model until it does replicate the actual test.

If the computer-based model is a model of an entire roadside safety structure, when fidelity of the model is confirmed by a previous test, additional computer-generated vehicle collisions are conducted and subsequently analyzed. Design changes are made to this computer model of the roadside structure after each vehicle collision until the vehicle collisions are deemed crashworthy - mitigating serious or fatal injury to vehicle occupants. The final computer-generated design of the roadside safety structure is then replicated in the real world by the MwRSF staff, and a proof-of-results crash test is subsequently conducted.

illustratio of impact with a barrier
A simulation of a pickup truck impacting the side of a bullnose barrier.

The UNL center obtains DYNA3D models of the motor vehicles used from NCAC. The center also collaborates with NCAC to obtain specialized changes to these existing vehicle models when they are required to ensure the fidelity of computer-generated crashes. However, in many cases, the center's research team has the skills and abilities to modify these vehicle models to fit their needs or even to develop new vehicle models when required for specialized or custom testing.

Reid reports that a recently completed project jointly conducted by the university center and MwRSF is a Bullnose Guardrail System for use in protecting support structures located in the middle of divided highways, particularly the bridge piers of overpasses.

"This guardrail system is in the shape of a partially flattened cylinder or oval," Reid explains. "And development of this design involved the analysis of two complete models of the guardrail system at the university's center of excellence, accompanied by nine full-scale crash tests conducted by MwRSF."

This project demonstrated how the DYNA3D crash-analysis tool was used successfully in combination with physical crash tests to create breakthrough safety structure hardware. The Bullnose Guardrail System has been submitted to FHWA for approval for use on the National Highway System. Approval by FHWA certifies that the system is crashworthy. Upon this approval, the eight Midwestern states will begin to deploy the bullnose system.

The latest challenging project, according to Reid, is working with the Indianapolis 500 Motor Speedway to identify ways to "soften" the concrete speedway track by improving the collision barriers. This project may lead to more work with the Indy Racing League.

"These motor racing projects demonstrate another feature of the combined University of Nebraska Center of Excellence and the Midwest Roadside Safety Facility," said Reid. "We are open for project proposals from any and all sources. Roadside structures, no matter how unique, can be virtually designed using the computer and DYNA3D by the University of Nebraska Center of Excellence, and the structures can then be proof-of-results crash-tested at the Midwest Roadside Safety Facility. Both the center and the safety facility want to be challenged to produce far-reaching, useful safety products."

Objectives of the FHWA DYNA3D Program

From FHWA's standpoint, the primary objective of the DYNA3D research program, according to FHWA's Trentacoste, is to "develop improved roadside safety structures and geometric features that mitigate harm - either fatal or serious injuries - to motor vehicle occupants involved in run-off-road crashes into roadside safety structures or on roadside geometric features."

A second objective is to develop and verify DYNA3D models of many of the more common roadside safety structures and geometric features and "to place [these models] on the shelf" for future use. Following acceptance of these models by NCAC, the models developed are intended for use by other researchers who may have a need to use them in subsequent vehicle crash-analysis assignments. This library of models will reduce start-up times associated with future projects.

An additional objective of the DYNA3D program is to stimulate interest and involvement in highway safety and vehicle crash analysis by a number of universities in the United States. Universities with faculty and staff who have demonstrated skills and abilities in using the computer and DYNA3D to analyze impact problems, who can apply them to vehicle crash analysis, and who can develop working relationships and obtain funding from state DOTs and roadside safety equipment manufacturers have the potential to become future centers of excellence for DYNA3D analysis.

"The longer term, overall objective of this program is to further develop the capabilities of this crash-analysis tool and to create core groups of trained researchers so that both the tool and the researchers become an integral part of the design and redesign process associated with improved roadside structures," Trentacoste said. The enhanced understanding of the crash event provided by this tool goes far beyond what can be currently expected when using existing techniques such as crash testing.

And finally, for the computer and DYNA3D to become an integral part of the design process for roadside safety structures, several tasks - the computer-generated vehicle crash analysis, the attendant design or redesign of the structures, and the proof-of-results crash testing - must be completed within time frames appropriate for the customer's needs.

DYNA3D Is Available to State DOTs and Private Companies

illustration of side impact
An LS-DYNA simulation of the side impact of a vehicle striking a slip-base luminaire pole.

It is important to note that the very capable technical staffs at NCAC and the four existing centers of excellence are available to satisfy customers' needs on a pay-as-you-go basis. State, county, and local DOTs and the many companies that are manufacturing and marketing roadside safety structures can now take advantage of this cutting-edge technology. They can put this technology to work on their roadside safety problems by employing the technical staffs at one or more of these five facilities.

Roadside safety structures with the highest degree of crashworthiness are currently being developed at these five facilities, and this output is expected to grow in the years to come as customer knowledge increases and acceptance of this new technology tool proliferates. In fact, it is clear from results to date that the computer and DYNA3D are making it possible for state DOTs who have taken advantage of this new technology to resolve some of the more difficult roadside safety problems that have been plaguing them over the years.

A good example of such problem-solving is the effort to determine the best way to anchor a guardrail end terminal so that a colliding vehicle of any size and type will not be impaled by the end of the guardrail, will not flip over, or will not pull the terminal out of the ground. The newest generation of guardrail end terminals was developed using DYNA3D. This new terminal brings the impacting vehicles to a sudden controlled stop and reduces the potential for injury to vehicle occupants.

As more and more computer-based crash tests are correlated with real-world tests, confidence in the accuracy and timeliness of computer-generated test results will increase. And as an ever-increasing number of motor vehicle models become available for crash analysis, FHWA and the state DOTs may soon be in a position to evaluate if and when simulated crash tests using the computer coupled with DYNA3D can be used to replace many of the real-world crash tests now being conducted as acceptance tests.

photo of a side impact
The actual side-impact crash test of a vehicle hitting a slip-base luminaire pole.

Saving Lives and Mitigating Injury

At a transportation safety conference held during 1999 in Washington, D.C., former Secretary of Transportation Rodney E. Slater told the attendees that saving lives and preventing transportation-related injuries were at the top of the priority list of the U.S. DOT. He went on to say, "Safety is our North Star by which we in DOT will be guided and judged."

By using the computer in combination with DYNA3D, we can now obtain powerful new insights into motor vehicle crashes that occur every day on our city streets, our county roads, our state highways, and our interstate highways. As a result, we now have at hand the potential to prevent some of the 3 million injuries and the 42,000 premature deaths that occur every year on our nation's roads.

For run-off-road motor vehicle crashes into roadside safety structures, we now have a North Star to guide us to a safer future. That North Star involves the use of the computer and DYNA3D.

Martin W. Hargrave is a research engineer in FHWA's Office of Safety Research and Development. One of his responsibilities is developing and managing research programs associated with FHWA's DYNA3D crash-analysis program. Before joining FHWA in 1979, he worked for 17 years in various engineering assignments in the private sector. He holds engineering degrees from the University of Alabama, Penn State, and the Catholic University of America in Washington, D.C.

David Smith is a frequent writer on transportation policy, issues, and technologies. A graduate of Cambridge University, he is the principal of AMANUENSIS Creative Group, a professional writing and consulting group located in Vienna, Va.

You can learn more about highway research programs through the Web site of FHWA's Turner-Fairbank Highway Research Center (www.fhwa.dot.gov).

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration