The Interactive Highway Safety Design Model: Designing for Safety by Analyzing Road Geometrics
Background: Safety and Geometric Design
The late 1980s saw a renewed interest in safety and geometric design in the United States. The Transportation Research Board (TRB) Committees on Geometric Design and the Operational Effects of Geometric Design launched a five-year series of sessions, beginning in 1988, on the state of the practice of five geometric design topics: sight distance, interchanges, intersections, alignment, and cross sections.(1-3) These sessions were held at the annual TRB meetings in Washington, D.C. At the state level, interest in geometric design was evidenced by a broad range of research problem statements submitted to and funded under the National Cooperative Highway Research Program (NCHRP).
Meanwhile, in March 1988, Federal Highway Administration (FHWA) management designated Highway Safety Design Practices and Criteria as a high-priority research and development (R&D) area. The objective of the program is to develop an integrated design process that systematically considers both the roadway and the roadside in developing cost-effective highway design alternatives. This process will help the highway designer evaluate design alternatives from a safety standpoint. Moreover, in developing the process, objective measure(s) of highway safety will be established.
Figure 1 shows the initial concept of the components of the integrated design process. The researchers developing the program envisioned that this process would begin with a design alternative developed by the highway designer in accordance with agency guidelines. This alternative would be checked for potential safety problems against safety data in each of four modules of a computer system. These modules would be a roadway module (which would essentially cover multivehicle accidents); a roadside module (which would essentially cover single-vehicle accidents; a consistency module (which would be based on speed profiles, since large changes in speed between successive roadway sections are believed to contribute to accidents); and a physics module (which would measure speeds and lateral accelerations based on a computer simulation of the interaction between the vehicle and the roadway). The design would be checked sequentially through these four modules. The designer would have to decide how to solve any potential safety problems identified through the process.
The program's first product was a six-volume synthesis on highway safety research.(4) This study specifically addressed access control, alignment, cross sections, interchanges, intersections, and pedestrians and bicyclists--topics selected based on recommendations from TRB and earlier synthesis studies.(5,6)
These studies pointed out several broad issues that would have to be addressed if the effort to develop an integrated design process was to succeed:
- The effort should integrate all safety research related to geometric design into a usable form for the designer. Past research only dealt with specific problems and gave little thought as to how results would be incorporated into the design process. For example, the synthesis study has data on the relationship between safety (accidents) and geometric features (medians, grades, etc.). The volume on intersections indicates that "intersections with poor sight distance experience higher accident rates." Unfortunately, "poor" sight distance is not quantified.
- The designer should be able to correct problems as the design was being reviewed. The process should be interactive.
- Standard definitions should be developed and adhered to for study variables. There were no common definitions for the variables used in the older studies reviewed; thus, the results of the different studies could not be combined. Some studies dealing with the same problem, such as the safeness of painted pedestrian crosswalks, reached different and opposite conclusions.
- Correct statistical procedures should be followed. Many of the early studies arrived at conclusions that were not supportable due to small sample sizes and questionable analysis.
- Measures other than accidents for evaluating highway safety should be pursued. Historically, attempts to relate accidents and geometric design features have not been successful; the correlation between accidents and geometric design features was very weak. Accidents might not be the best measure of a highway's safety.
These issues clarified the concept of what is now known as the Interactive Highway Safety Design Model (IHSDM). Although the model's exact format was unknown, its purpose wasand remainsclear. IHSDM will provide information on safety and geometrics in a format that a highway designer can use. It will guide the designer in evaluating the safety of the design.
The Geometric design process varies considerable among the 50 states. However, it can generally be divided into two phasespreliminary design often associated with the preparation of environmental impact statements (EIS) and detail design associated with the preparation of plans, specifications, and estimates (PS&E). Because of the type of data available at each phase, two versions of the IHSDM would be needed.
Level 1 IHSDM
When complete, the Level 1 IHSDM will be used during the preliminary phase of a highway to provide managers with data on safety. Specifically, it will determine the expected number of accidents based on such geometric design information as the number of lanes, average daily traffic, speed, urban/rural environment, type of terrain, etc. This general informationtogether with data on other social, economic, and environmental effectswill be used to decide policy issues during the preliminary phase of highway design.
The researchers decided that the initial attempt to develop the Level 1 IHSDM should involve the newly developed Highway Safety Information System (HSIS).(7,8) The HSIS data base currently contains the linked files of five statesIllinois, Maine, Michigan, Minnesota, and Utah. Data from other states are now being added to the data base. The linked files contain traffic, geometry, and accident data. It was hoped that the linked files would provide enough detail to establish general relationships between accidents and geometric features suitable for the preliminary design phase. Unfortunately, this is not the case. While the linked files are a considerable improvement over earlier data bases, they have a number of weaknessfor example, the lack of uniformity among states on what data elements are collected.
Tentatively, the Level 1 IHSDM will be based on accident data from Minnesota as well as from California and Washington, which are now being considered for inclusion in HSIS in 1994. Three states are being used because the linked data bases in each state have different strengths. For example, California has an excellent intersection data base. The initial effort is to develop a Level 1 IHSDM for two-lane roads. Supplemental field data on both geometric and operational features will be gathered during the 1994-95 period. Two-lane roads were chosen because they represent the best chance for success. If a Level 1 IHSDM cannot be developed for two-lanes roads, there is little chance of success for multilane roads. The Level 1 IHSDM for two-lane roads is being developed as an R&D staff study. The additional field data will be collected as soon as the contract is awarded.
Level 2 IHSDM
The Level 2 IHSDM will be used to evaluate and finalize geometric design details during the development of the PS&E. The synthesis study represents our best current data on the relationships between accidents and geometric design features. A review of these documents clearly demonstrates the difficulty of establishing a definitive relationship between accidents and geometric design features. With a few exceptions, the correlation between accident and geometric features has been very difficult to establish within the bounds that would encourage use of predictive models. A legitimate question would be why will the Level 2 IHSDM succeed when previous attempts have failed. Several factors should contribute to the success of the IHSDM.
First, the development of the Level 2 IHSDM is a coordinated effort with a clear objectivedevelop a process that the designer can use. Much of the information needed for the Level 2 IHSDM is not currently available, but will be developed in the near future through competitive contracts. To coordinate this effort, FHWA has established a Geometric Design Laboratory at the Turner-Fairbank Highway Research Center (TFHRC). Laboratory staff will provide consistent definitions, ensure the integration of different modules, coordinate with public and private partners, handle statistical reviews, etc.
Second, the HSIS provides a broad base upon which studies can be designed to answer specific questions. HSIS allows the development of experimental designs with sufficient data that statistical analysis can be conducted before expensive field studies are started. HSIS analysis coupled with pilot field studies should increase the chances of success. Statistical procedures for analyzing data bases to develop relationships between accidents and geometrics have also improved in the last decade.
Third, technology has provided the designer with the ability to view, analyze, and change designs using desktop computer technology. The most important one for the development of the Level 2 IHSDM is the computer-aided design (CAD) systems used to design highways. Rapid changes in alignment and grade can be made with the click of a mouse. Standardized drawings can be stored and copied in seconds. More importantly, the CAD technology provides a means of integrating other tools that the designer has not had available.
Finally, the Level 2 IHSDM is now viewed as a shell that provides an interface between the design alternative(s) and eight, rather than four, modules. All of these modules represent tools (in the form of computer subroutines) that the designer needs in evaluating proposed designs. In addition to improved accident/geometric feature relationships, the Level 2 IHSDM will provide the designer with new capabilities to develop safer designs. Note: these capabilities will include the consistency module, a subroutine checking design consistency (does the design violate driver expectancy?); the vehicle dynamics module, a subroutine allowing the designer develop data on speed profiles and lateral accelerations and to visually inspect the design (three-dimensional view from a driver perspective); and the driver module, a subroutine that eventually may lead to new design procedures based on human factors. Other capabilities will be discussed in the following section on current status.
Current IHSDM status
Following is a synopsis of current activities related to the Level 2 IHSDM as shown in figure 2. As noted earlier, the Level 2 IHSDM is a shell (computer software) that provides an interface between the eight modules shown in figure 2 and the commercial CAD software. Conceptually each of the modules (or subroutines) may be a pulldown window on the computer screen. Subsequent Public Roads articles will provide more detail on the individual modules.
The Level 2 IHSDM with its CAD core will work as follows. The designer, using agency guidelines stored in a commercial CAD package, will develop a highway design alternative. This design will then be evaluated by each of the modules in sequence. (In the future, the design may be evaluated concurrently by two or more modules, but right now it is unclear how the modules would interact to optimize design.)
The following illustrates how the Level 2 IHSDM might be used. The designer activates the IHSDM program. The designer decides to visually inspect the design alternative and to develop data on speed profiles and lateral accelerations based on the dynamics of the design vehicle operating on the proposed roadway. To do this, the vehicle dynamics module is activated, and the designer selects one of the design vehicles on which the roadway geometry is based. This routine is used to develop a speed profile and data on lateral accelerations. If the designer encounters a geometric feature that does not meet the speed and lateral acceleration criteria, the location is noted, and the module is exited. Any adjustments to the design will be made using the CAD package.
Next, the designer decides to search for combinations of geometric elements that violate driver expectancy. The designer activates the design consistency module (or subroutine). This routine might use the speed profile (developed in the vehicle dynamics model) and look for combinations of geometric design elements (from the CAD package) that result in significant speed changes. The designer will again use the CAD package to make any necessary design adjustments.
Finally, the designer wishes to develop an accident profile of the design and overall accident statistics. The designer activates the accident predictive module that contains the relationship between accidents and design features (accident studies). This data is compared with the design alternative geometry (from the CAD package), and the accident statistics are computed.
In the completed IHSDM, the design will be checked against the remaining five modules and adjusted as necessary. As of this writing, however, four modulesthe vehicle dynamics, consistency, accident predictive, and roadside safety structure modulesare now under development.
CADHeart of the Level 2 IHSDM
All of the modules shown in figure 2 can be interfaced with data found in current computer-aided design (CAD) packages.
There are several major advantages in using CAD technology as the core of IHSDM:
- Most state highway agencies currently use some sort of computer-aided design packagee.g., IGRDS, INROADS, GEOPACK, etc.to develop their highway plans. They are, therefore, familiar with CAD techniques, protocols, and outputs. Moreover, several states have incorporated their standard design elements into these packages. Additionally, mostif not allof the design tables contained in the American Association of State Highway and Transportation Officials' A Policy of Geometric Design of Highways and Streets, called the "green book," have also been incorporated into the CAD package.
- CAD technology is relatively standardized and well-defined. Most third-party CAD packages use either Intergraph's MICROSTATION or Autodesk's AUTOCAD. This makes it easy for the states and add-on vendors to add features to the basic CAD package.
- CAD technology has established output formats. For example, Intergraph uses a DGN format, and Autodesk uses a DWG format. These formats permit design of interfaces between modules and the CAD package using a common technology. A DXF format can be used as a go between.
- The use of the CAD package greatly simplifies the interface between Level 2 IHSDM, the eight modules, and the CAD packages. All of the interfaces must use the outputthe x, y, and z coordinates systemfrom the CAD packages. A contract will be awarded in 1994 to define the system interfaces as part of the development of the driver module.
Vehicle dynamics module
Two contracts were awarded in September 1993 to develop the vehicle dynamics module. The vehicle dynamics module will contain the design vehicles listed in the AASHTO green book. The vehicle dynamics of these design vehicles will be evaluated as rigid bodies. The IHSDM will than provide the link between the design vehicle and the roadway geometry (stored in the CAD package). When completed, this module will allow a designer to "drive" the design vehicle through the alternative design and develop a speed profile and data on lateral accelerations. This ability to travel through the design will give the designer a visual method of looking for poor design situations. In addition, the vehicle dynamics module can be used by researchers to evaluate current ditch and slope design criteria based on current motor vehicles. In April 1994, a prototype vehicle dynamics module will be demonstrated at the TFHRC.
The AASHTO document A Policy on Geometric Design of Highway and Streets presents the national policy for geometric design. The policy has evolved since the 1930s and today focuses on the design speed concept. AASHTO defines design speed as "the maximum safe speed that can be maintained over a specified section of highway when conditions are so favorable that the design features of the highway govern."
Research in the United States and abroad has shown that the use of the design speed concept in selecting geometric elements can result in designs that violate driver expectancy.(9) In general, design speed and operating speed should be reasonably close, and operating speeds between successive design elements should not vary widely. A design where the operating speeds are consistent with the driver's expectations is desired. Designing according to the design speed values does not necessarily ensure consistent design.
Consequently, much research has recently been initiated on the topic of consistency. For example, at the January 1994 TRB session on speed, four papers based on FHWA-sponsored consistency research were presented.(10-13) All four of the papers emphasized the importance of operating speedthe speed selected by users when free of traffic or regulatory constraints. The results of studies on two-lane, rural roads show that drivers do exceed design speeds on highways. Additionally, an international workshop is being planned for the summer of 1994 on this topic.
Accident predictive module
The accident predictive module of the Level 2 IHSDM will estimate the expected number of accidents for any design alternative. To do this, the accident predictive module has been divided into four separate submodels: roadway segment, intersections, interchanges, and roadside. The overall accident rate would be the sum of the accident rates for each individual road segment, intersection, interchange, and roadside area that make up the design alternative.(14)
HSIS will be the starting point for the development of these accidents rates. To supplement the existing data contained in the HSIS data base, large field studies will be conducted.
These studies should avoid many of the problems of earlier field studies by using a two-step approach. First, HSIS will be used to see if the relationship between accidents and geometric design features can be developed for each submodel, the strength of the relationship, and the magnitude of the effort needed to develop a sound statistical predictive model. Second, an experimental plan, with pilot studies if needed, will be developed to gather the additional data needed to ensure that all important variables have been identified. The final step will be the actual data collection and analysis.
In 1992, two contracts were awarded to develop several experimental designs for the accident predictive modules. The contractors are currently working on intersections and roadside (encroachment) submodels. The contractor developing the experimental design for intersections is also the contractor on two NCHRP studies related to intersectionsIntersection Sight Distance and Median Intersection Design. A preliminary analysis has been completed and a pilot study is under way to see if the robustness of the predictive model for estimating intersection accidents can be improved.
The second contractor is working to develop a procedure to measure encroachments. Encroachment occurs when the driver unintentionally leaves the roadway. A roadside accident model could be based on encroachments or based on accidents. Both options have been discussed. The current encroachment work is aimed at improving the encroachment data used in the ROADSIDE program.(15) Initial efforts on developing an encroachment submodel have not been successful. Additional work on alternative ways of estimating encroachments is under way.
It is hoped that the ROADSIDE program can be used as the roadside submodel in the accident predictive module. Both TRB and FHWA are working on various aspects of the roadside problemi.e., through the NCHRP research project on Improved Procedures for Cost-Effectiveness Analysis of Roadside Safety Features and the FWHA study on Development of Preliminary Severity Indices for Roadside Benefit/Cost Model.
Roadside safety structure module
The roadside safety structure module will be used to design roadside safety structures that reduce injury severity. FHWA and the National Highway Traffic Safety Administration (NHTSA) have agreed that the general-purpose, nonlinear, finite element computer program, DYNA3D, will be the finite element method (FEM) the agencies will use in studying crash events.(16) FHWA and NHTSA have entered into an interagency agreement with Lawrence Livermore National Laboratory (LLNL) to further develop DYNA3D to solve crash problems. DYNA3D will allow FHWA to treat roadside crashes as a vehicle/structure design problem. DYNA3D will be used to conduct parametric studies that will provide data on forces and deflections on existing and new roadside safety structures, leading to both better designs for minimizing accident severity and improved criteria for designing barriers. FHWA is using DYNA3D to study design problems associated with the breakaway cable terminal (BCT) and U-channel sign supports. These studies will provide the insight and experience necessary to design new systems.
Work on the driver module will begin in 1994. The first study in the module's development effort will focus on how the driver module will interact with the IHSDM. Another study to be conducted early in the development effort will probably focus on the information drivers obtain from the highway environment in order to select operating speed.
IHSDM researchers and program designers initially thought that a measure for highway safety might be developed based on geometrics alone. As alluded to elsewhere in this article, no way has yet been found to develop such a measure. Accident frequency and accident rate are affected by vehicle operations. Moreover, preliminary data from the development of the Level 1 IHSDM clearly demonstrate the importance of average daily traffic (ADT) in establishing a relationship between accidents and geometric design features. Thus, the IHSDM traffic module will provide data on vehicle operations, notably on average daily traffic, to help establish an accident-ADT-geometrics relationship.
Policy review module
The policy review module will not be initiated until the NCHRP project 17-9 on the Effects of Highway Standards on Safety is completed in 1994. The objective of the NCHRP study is to assess the safety effects of highway design standards given limited resources and other constraints. The policy review module will assist designers in evaluating design elements that are not addressed by the other modules. For example, there may be time when a particular design element does not comply with established design criteria. Exceptions are granted when full compliance with a standard would not be cost-effective. This module will provide a means for explicitly documenting such decisions.
The benefit-cost module is also being deferred until the NCHRP project on the Effects of Highway Standards on Safety is completed because the study is looking at construction costs. The purpose of this module is to determine if incremental increases in construction costs could be justified on the basis of reduced accident costs. Current CAD models are now available, however, that can calculate construction costs based on design data. In addition, FHWA has adopted comprehensive costs for use in benefit-cost analysis.(17)
The discussion here is intended to introduce the reader to the concept of IHSDM. The overall plan has been presented to the TRB committees on Geometric Design and on the Operational Effects of Geometric Design. Both committees have been very positive about the need to conduct the research to develop the IHSDM. In June 1994, the four regional AASHTO Subcommittees on Design will discuss IHSDM. Subsequent articles will expand on the model. By the end of 1994, contract work will be under way on developing the IHSDM shell and five of the modulesvehicle dynamics, consistency, accident predictive, roadside safety structure, and driver. The development of the Level 1 and Level 2 IHSDM is a very ambitious R&D program, but one that can be accomplished.
(1) Highway Sight Distance Design Issues, Transportation Research Record 1208, National Research Council, Transportation Research Board, Washington, D.C., 1989.
(2) Intersection and Interchange Design, Transportation Research Record 1385, National Research Council, Transportation Research Board, Washington D.C., 1993.
(3) Cross-Section and Alignment, National Research Council, Transportation Research Board, Washington D.C., forthcoming.
(4) Safety Effectiveness of Highway Design Features, Volume I: Access Control (FHWA-RD-91-044), Volume II: Alignment (FHWA-RD-91-045), Volume III: Cross-Sections (FHWA-RD-91-046), Volume IV: Interchanges (FHWA-RD-91-047), Volume V: Intersections (FHWA-RD-91-048), Volume VI: Pedestrians and Bicyclists (FHWA-RD-91-049), Federal Highway Administration, Washington, D.C., 1992.
(5) Designing Safer Roads--Practices for Resurfacing, Restoration, and Rehabilitation, Special Report 214, National Research Council, Transportation Research Board, Washington, D.C., 1987.
(6) Synthesis of Safety Research Related to Traffic Control and Roadway Elements, Volume 1 (FHWA-TS-82-232), Volume 2 (FHWA-TS-82-233), Federal Highway Administration, Washington, D.C., 1982.
(7) F.M. Council and J.F. Paniati. "The Highway Safety Information System (HSIS)," Public Roads, Vol. 54, No. 3, December 1990.
(8) J.F. Paniati and F.M. Council. "The Highway Safety Information System: Applications and Future Directions," Public Roads, Vol. 54, No. 4, March 1991.
(9) R.A. Krammes, R.Q. Brackett, et al. State of the Practice of Geometric Design Consistency, Federal Highway Administration, Washington, D.C., forthcoming.
(10) R.A. Krammes. "Design Speed and Operating Speed in Rural Highway Alignment Design," 940996, paper presented at Session 246, Transportation Research Board annual conference, Washington, D.C., January 1994.
(11) J. Ottesen and R.A. Krammes. "Speed Profile Model for a U.S. Operating-Speed-Based Design Consistency Evaluation Procedure," 940995, paper presented at Session 246, Transportation Research Board annual conference, Washington, D.C., January 1994.
(12) I. Anderson and R.A. Krammes. "Speed Reduction as a Surrogate for Accident Experience at Horizontal Curves on Rural Two-Lane Highways," 940995, paper presented at Session 246, Transportation Research Board annual conference, Washington, D.C., January 1994.
(13) J.P. Tarris, J.M. Mason, and N.D. Antonucci. "Geometric Design on Low-Speed Urban Streets," 940997, paper presented at Session 246, Transportation Research Board annual conference, Washington, D.C., January 1994.
(14) D.W. Hardwood, J.M. Mason, and J.L. Graham. Conceptual Plan for an Interactive Highway Safety Design Model, FHWA-RD-93-122, Federal Highway Administration, Washington, D.C., February 1994.
(15) Roadside Design Guide. American Association of State Highway and Transportation Officials, Washington, D.C., 1988.
(16) D.A. Schauer, F.J. Tokarz, R.W. Logan, et al. Vehicle Impact Simulation Technology and Advancement (VISTA), FHWA-RD-92-111, Federal Highway Administration, Washington, D.C., September 1993.
Jerry A. Reagan is chief of FHWA's Design Concepts Research Division at the Turner-Fairbank Highway Research Center in McLean, Va. Prior to this assignment, he served as chief of the Safety Traffic Implementation Division. He has had a variety of experiences with FHWA, beginning in 1967 as a materials engineer. Later he was assigned to Region 15 as a soils and foundation engineer. In 1973, he transferred to the Office of Environmental Policy at FHWA headquarters where he worked for 10 years. Then he moved to TFHRC as the state programs officer of the National Highway Institute where he was responsible for the NHI short-course program. He has a bachelor's and a master's degree in civil engineering from the University of Tennessee. He is a registered professional engineer in Tennessee and Virginia.