Development of a New Guardrail System

72

Paper No. 970355

TRANSPORTATION RESEARCH RECORD 1599

Development of a New Guardrail System JOHN D. REID, DEAN L. SICKING, RONALD K. FALLER, AND BRIAN G. PFEIFER The W-beam guardrail system has been the standard in the United States since the late 1950s and has proved to perform reasonably well under most impact conditions. However, in recent years the vehicle fleet has changed to include a relatively large percentage of light trucks, such as pickups, vans, and sport-utility vehicles. These vehicles have a higher center of mass and bumper mounting height than conventional automobiles and have been shown to have higher rollover and injury rates during guardrail accidents than conventional automobiles. Standard W-beam guardrails were not designed to capture the bumper of many of these vehicles. In recognition of the potential safety problems associated with light-truck accidents, safety performance standards were recently changed with the publication of NCHRP Report 350, Recommended Procedures for the Safety Performance Evaluation of Highway Features. These performance standards require all new safety hardware to be tested with a full-size three-quarter-ton pickup to ensure acceptable performance for most vehicles in the light-truck category. In recognition of this, a guardrail system capable of capturing and redirecting a larger range of vehicle types and sizes was developed. A new guardrail system, called the Buffalo Rail, was designed with a new cross-sectional shape with an effective depth of 311 mm (compared to 194 mm for the W-beam), a rail thickness of 13 gauge, and a post spacing of 2500 mm. The safety performance of the Buffalo Rail was found to be acceptable according to the procedures and criteria recommended for the three-quarter-ton pickup truck at Test Level 3 in NCHRP Report 350.

The W-beam guardrail system has been the most widely used roadside barrier in the United States since the late 1950s and has proved to perform reasonably well under most impact conditions. However, in recent years the vehicle fleet has changed to include a relatively large percentage of light trucks, such as pickups, vans, and sportutility vehicles. These vehicles have a higher center of mass and higher bumper mounting heights than conventional automobiles. Standard W-beam guardrails were not designed to capture many of these vehicles due to a small effective rail depth and a relatively low mounting height. Further, accident data has shown that light-truck impacts with guardrails are more likely to result in vehicle rollover and occupant injury than accidents involving conventional automobiles (1). Finally, the potential for W-beam guardrail to cause high center-of-mass vehicles to roll over was clearly identified when a full-size van experienced a violent roll over during a crash test involving a steel-post W-beam guardrail (2). In recognition of the potential safety problems associated with light-truck accidents, safety performance standards were recently changed with the publication of NCHRP Report 350 (3). These performance standards require all new safety hardware to be tested with a full-size three-quarter-ton-pickup to ensure acceptable performance for most vehicles in the light truck category. FHWA officially adopted NCHRP Report 350 and has ruled that all safety hardware installed after August 1998 will be required to meet the new standard. In 1994, the Texas Transportation Institute (TTI) conducted fullscale crash-testing of strong-post W-beam guardrails under NCHRP Midwest Roadside Safety Facility, Department of Mechanical Engineering, University of Nebraska, P.O. Box 880656, Lincoln, Neb. 68588.

Report 350 criteria (4). Tests were performed with three-quarter-ton pickups on both steel- and wood-post W-beam guardrails. During both of these tests, the impacting vehicles deformed the guardrail significantly, experienced significant wheel snagging on the posts, and attained a high roll angle as the vehicle exited the barrier. During the wood-post guardrail test, the vehicle almost rolled over and then returned to an upright position. However, during the steel-post test, the vehicle had a slightly higher roll angle when it returned to the ground, and as a result it rolled onto its side rather than returning to its wheels. The results of these two tests were classified as marginally successful for the wood-post system and a marginal failure for the steel-post system. However, due to the high levels of variability inherent in full-scale crash-testing, it is difficult to use these test results to conclude that the wood-post guardrail performs any better than the steel-post system. The pickups used in these two tests had bumper heights ranging from 622 to 635 mm, which are on the extreme lower end of the range of bumper heights found on three-quarter-ton pickups. A survey of three-quarter-ton pickup bumper heights has been conducted as part of the NCHRP Project 22-11, currently in progress at TTI. This survey has shown that the bumper heights of these vehicles range from 622 to 787 mm. In an effort to develop a better understanding of the three-quarter ton pickup truck test vehicle, the Midwest Roadside Safety Facility has begun to measure test vehicle center-of-mass heights. During this process, it was noticed that the best indicator of center-of-mass and bumper height of the pickup was the number of lugs on the wheels. All of the three-quarter-ton pickups that had lower bumper heights, in the range of 622 to 635 mm, were manufactured by General Motors, had a light-duty suspension, and had a six-lug wheel. Examinations of the films of the TTI tests have shown that the vehicles used in the NCHRP Report 350 crash tests of strong-post W-beam guardrails also had only six lugs. Heavy-duty three-quarter-ton, two-wheel drive Chevrolet pickups and all Ford three-quarter-ton, two-wheel drive pickups had bumper heights and centers-of-mass approximately 65 to 125 mm higher than the light-duty, three-quarter-ton Chevrolet pickup trucks. Further, the pool of available crash-test vehicles in Nebraska and surrounding states has been found to be predominately heavy-duty versions of the three-quarter-ton pickups. Vehicle purchasers have indicated that Ford pickups are most readily available and that fewer than 20 percent of the available Chevrolet vehicles fall into the light-duty category. Therefore, at least in the Midwest region, it can be concluded that vehicles conforming to the size and configuration used during testing of the strong-post W-beam guardrails represent fewer than 15 percent of the total pool of available crash-test vehicles. Another variable in this complex relationship is the actual mounting height of the W-beam guardrail. There are frequently significant variations in the mounting heights of W-beam guardrails during construction, and the effective height of these barriers can change when adjacent traffic lanes are overlaid. A 1985 study conducted at TTI investigated the performance limits of the W-beam guardrail (2). Successful tests were conducted with a full-size sedan on a

Reid et al.

Paper No. 970355

610-mm high system and with a minisize automobile on a 762-mm system. As a result of this testing, an allowable variation in W-beam guardrail mounting heights of ±75 mm was included in the 1989 AASHTO Roadside Design Guide. Based on the relatively marginal performance of W-beam guardrails at the standard mounting heights, there is no reason to believe that these barriers could continue to function adequately when mounted at the lower end of the currently acceptable range. One option for improving guardrail performance is to require thrie-beam be installed in place of the W-beam. The thrie-beam system is more than 60 percent deeper than the W-beam, and is more than capable of redirecting a pickup truck impact. There are several disadvantages to the thrie-beam, however, as it contains approximately 50 percent more material than the W-beam, resulting in a comparable increase in weight and price. The large thrie-beam is also somewhat of an ominous site along the roadside, as it blocks the drivers view and diminishes the aesthetics of the roadway. On the basis of this information, it is evident that there is a need for a new guardrail element with sufficient height to successfully redirect pickup trucks, provide adequate depth to prevent underriding by small cars, and provide reasonable mounting height variations for field installations. It would be beneficial to develop a guardrail with these properties if this can be accomplished without significantly increasing the cost of the system. Further, if the new rail could be designed without significantly increasing the section modulus and cross-sectional area, it could meet the revised crashtest standards for longitudinal barriers and still have the potential to utilize economical end-treatment techniques. The scope of this study includes the development and compliance testing of a guardrail system capable of meeting the criteria set forth

FIGURE 1

73

in Test Level 3 of Table 3.1 in NCHRP Report 350. The development of this system includes the evaluation of a number of candidate shapes through component testing and computer simulation. The compliance testing described in this paper in-cludes impacting the rail with a 2000-kg pickup at 100 km/hr and 25 degrees (Test 3-11). This study concentrates on the development and research of the new guardrail system, and not on the specifics for an FHWA-compliance report for NCHRP Report 350 certification.

SYSTEM DETAILS The system developed as a result of this research project is similar in construction to the current W-beam system, that is, the new system, named the Buffalo Rail, consists of a guardrail attached to wood posts that are imbedded in soil. However, major design changes were made to the rail shape, the material thickness, the rail splice, and the post spacing. A description of the developmental system follows, and some of the details of the design are described later. The first priority of the new rail development was to increase the rail’s maximum height and its effective depth or capture region. Increasing the rail’s height raises the resultant force on the impacting vehicle and thereby reduces the propensity for causing rollovers. Increasing the effective depth involves widening the distance between the lower and upper peaks of the rail element. This change will allow the rail element to capture vehicles that might vault over or underride other barrier systems. As illustrated in Figure 1, the primary difference between the W-beam system and the new system is the increased height and effective depth of the rail. The effective depth of the rail elements is

Cross-sectional view of W-beam and Buffalo Rail.

74

Paper No. 970355

TRANSPORTATION RESEARCH RECORD 1599

the region between the two peaks. When a vehicle’s bumper or other component strikes a barrier between the two peaks, it will generally be contained. The effective depth of the guardrail is defined as the distance between the points where the inside slopes of the outermost corrugations go to zero. Any bumper impacting in this region should be captured by the guardrail. As is evident in the drawings, this effective depth is increased from 194 mm on the W-beam to 311 mm on the Buffalo Rail, an increase of more than 60 percent. This increased depth allows a greater range of vehicles to be captured and redirected by the Buffalo Rail. The difference that this makes in the bumper-guardrail interaction can be observed in Figure 2, in which the same pickup is shown next to both a W-beam system and the Buffalo Rail. The bumper meets the W-beam rail just above the upper corrugation, although it is clearly within the effective-depth region of the Buffalo Rail. To maintain the tensile capacity of the rail without increasing the amount of steel required, the thickness of the sheet metal was reduced to 2.28 mm (13 gauge) from the 2.66 mm (12 gauge) used in the standard W-beam. This reduction in thickness allowed for the much greater effective depth, while maintaining the same crosssectional area of material. The steel used in this new guardrail shape conforms to the AASHTO M180 Class A specifications, the same specifications used for the W-beam guardrail. Because of the reduction in material thickness, it was necessary to increase the number of bolts in the splice from the 8 found in the W-beam splice to 12.

The rails used during the development of this system were fabricated on a break press, so it was only possible to make them 5.5 m in length, which covers two post spans. When in production, the rails will be rolled approximately 8 m long so that each rail extends over three post spans. Photographs of the wood-post system are presented in Figure 3. This system consists of standard 152- × 203- × 1829-mm posts with two 152- × 203- × 438-mm wood blockouts. The system was tested with two wood blockouts to minimize the effect of the vehicle wheels snagging on the posts. The main purpose of the blockout in a guardrail system is to increase the distance between the guardrail and the post to minimize the possibility of the vehicle snagging on the post. Two blockouts were used in the design so that the interaction between the pickup and guardrail could be evaluated independently from any snagging that might occur when only one blockout is used. When snagging occurs on a wood post, it occurs over the surface of the post where there are no hard or sharp discontinuities to accentuate the snagging. In addition to this, the wheel can gouge through the wood upon impact, lowering the peak force imparted into the wheel assembly. The post spacing of the Buffalo Rail is 2.5 m, compared with 1.9 m for W-beam. This increased post spacing results in a 25 percent reduction in the number of posts, post bolts, and possibly blockouts needed for a given installation. The new system does require 50 per-

FIGURE 2

FIGURE 3

Comparison of bumper interaction.

Buffalo Rail wood post design.

Reid et al.

Paper No. 970355

cent more splice bolts, and if an agency chose to use double blockouts, it would also result in a 50 percent increase in the number of blockouts. The rail lengths are comparable, 7.5 m for the Buffalo Rail and 7.6 m for the W-beam, so the number of splices required for a given system will be similar. The final system to be submitted to FHWA for NCHRP Report 350 approval is expected to have single blockouts.

DESIGN AND ANALYSIS Rail Section Development The most common longitudinal barrier system in use today is the strong-post guardrail system consisting of a corrugated steel beam supported by either wood or steel posts. Since the mid-1950s, several countries throughout the world have developed and tested their own corrugated steel beam shapes, with mixed results, as presented in Figure 4. For many years, the United States has relied heavily on only one corrugated steel beam shape, known as the W-beam. The W-beam

FIGURE 4

Existing guardrail shapes.

75

shape was developed and evaluated for both weak- and strong-post guardrail systems and was based primarily on a significant amount full-scale vehicle crash-testing with both sedans and small cars. During the early 1970s, a larger corrugated steel beam, known as the thrie-beam, was also introduced in the United States. The thriebeam shape was developed to contain heavy vehicles that required a deeper, stronger rail section, and also to aid in the transition of semirigid guardrail to more rigid bridge rails and median barriers. The development of the new rail shape, named the Buffalo Rail, considered several design factors, such as improved geometry, comparable structural capacity and cost to that of the W-beam, and aesthetics. Researchers believed that the general shape of the new rail should be based on improvements or modifications made to the existing W-beam shape. Therefore, it was determined that the new rail should provide approximately the same tensile capacity as the W-beam, but with an increased depth or vertical contact distance to reduce vehicular instability problems encountered with higher center-of-mass vehicles. The design effort began by drawing the W-beam cross section with AutoCAD to determine and verify the geometric cross-

76

Paper No. 970355

sectional properties such as the area, total width and depth, moments of inertia, centroids, section modulii, and length. Once the W-beam shape was calibrated with known values, variations and changes could be made with confidence to the W-beam shape as well as to subsequent shapes to determine the revised geometric crosssectional properties. Nearly 20 cross sections were considered during the design process, which began with the W-beam and iterated through several shapes until the Buffalo Rail shape was obtained. Some of the rail shapes considered during this process, as well as the W-beam and Buffalo Rail shapes, are presented in Figure 5. In addition, the cross-sectional properties for these selected shapes are given in Table 1. Researchers believed that the guardrail’s safety performance could be improved with an increase in both the effective and total depth of the rail. The effective depth is defined as the distance between the points where the inside slopes of the outermost corrugations go to zero and it is also considered to be the minimum capture area believed necessary for containing bumpers and redirecting higher center-of-mass vehicles. As indicated in Table 1, the effective depth was increased 60 percent from 194 mm on the W-beam to 311 mm on the Buffalo Rail, allowing a greater range of vehicles to be captured and redirected by the new rail. The total depth was also increased from 311 mm for the W-beam to 387 mm for the Buffalo Rail. However, to maintain the economic feasibility of the new shape, the thickness of the sheet metal was reduced to 2.28 mm (13 gauge) from the 2.66 mm (12 gauge) used in the standard W-beam. This reduction in thickness allowed for much greater effective and total depths, while maintaining the same 1284-mm2 cross-sectional area of steel used in the W-beam. By using a comparable grade of steel, conforming to the AASHTO M180 Class A specifications, and a crosssectional area equal to that used in the W-beam, the Buffalo Rail maintained a tensile capacity approximately equal to that of W-beam. During the iteration process, it was necessary to change the orientation of the slopes of the legs of the two corrugations. This was done to provide the required effective depth as well as the crush strength needed to cause the rail to interlock with the vehicle during impact. For the Buffalo Rail, the slopes of the inner and outer legs of the corrugations, as measured with respect to the rail depth or 311-mm dimension, were 43.6 and 78.7 degrees, respectively. In addition, the dimensions of the various rail shapes were configured to provide an elastic section modulus, or a measure of a beam’s ability to resist internal bending moments, approximately equal to that provided by the W-beam. The critical, elastic section modulus about the vertical axis for both the W-beam and Buffalo Rail are 22.5 cm3 and 21.5 cm3, respectively.

FIGURE 5

Sample of rail shapes considered.

TRANSPORTATION RESEARCH RECORD 1599

TABLE 1

Geometric Rail Properties

Splice Pattern Because of the reduction in gauge thickness of the new rail, the splice pattern between two rails required a redesign. A number of static tests, along with bolted splice calculations, were performed previous to the full-scale crash tests to help determine the number and location of the bolts at the splice. The new splice pattern (Figure 6) consists of 12 bolts, compared to 8 bolts for the W-beam splice. As is evident in Figure 6, the maximum number of bolt holes in a line is three, as compared to four in the W-beam. This reduced number of bolts in a line, combined with the smaller gauge steel, results in an 8 percent increase in the effective cross-sectional area at the splice when compared with the W-beam splice. Computer Simulation Nonlinear finite-element analysis (FEA) was used to aid the design process. The code used for simulation was LS-DYNA3D (5). This section briefly describes some of the simulation effort.

FIGURE 6

Splice pattern.

Reid et al.

Paper No. 970355

77

Creasing and Interlocking

Redirection

One of the critical features of the W-beam is its ability to capture the bumper and crease the side of a vehicle and create an interlock between the vehicle and the rail. This interlock helps prevent the vehicle from vaulting or rolling over the rail. Thus, any new rail shape would have to create such an interlock. To help determine the interlocking capabilities of the W-beam, 300-mm sections were cut and statically compressed until flattened. The average peak crush force from eight physical tests was 31 kN per 300 mm of rail. The energy absorbed in the crushing was 1795 kN-mm per 300 mm of rail. The test data provided an upper-bound target for any new cross-sectional shape. The flattening tests were then simulated using FEA. Figure 7 presents the FEA model and force-deflection curves from a few of the physical tests, along with the simulated response. The relatively good simulation of the W-beam flattening test gave confidence in using simulation to study various cross-sectional shapes. By using FEA, several design options were quickly discarded, thus helping to narrow a large number of design candidates. Actually making several candidate cross sections and then testing those cross sections was not feasible.

Simulation of the redirectional impact was used throughout the design process to get a feel for the new design behavior before any actual tests were performed. Early simulations consisted of a rigid rounded block impacting a shortened guardrail system; the simulation results presented in Figure 8 was one such simulation of an early cross-sectional candidate. Although the simulation had multiple problems (for example, the obvious hourglassing in the posts), insight into the behavior of the new design was gained. Later in the design process the details of the models were improved significantly. The result in Figure 9 was completed five days before the first full-scale test. The model illustrated attempted to replicate the final system design. Because the goal of the project was to develop a new guardrail system, and not to simulate a guardrail system, no attempt was made to improve or validate the final model shown. The simulation had served its purpose. Although FEA did not lead the design for this application, the simulation effort provided valuable insight into the guardrail behavior. This insight, although not quantifiable, played a significant role in the Buffalo Rail design.

FIGURE 7

Rail flattening tests.

78

Paper No. 970355

FIGURE 8

TRANSPORTATION RESEARCH RECORD 1599

Rigid bogey simulation.

NCHRP REPORT 350 TEST 3-11 (2000P, 100 KM/HR, 25 DEGREES) A 1984 Chevy C20 pickup was directed into the Buffalo Rail at 100 km/hr and 25 degrees. The impact point, as determined from criteria in NCHRP Report 350 (3), was located 0.66 m upstream of Post 11. The mass of the pickup was 2040 kg, the vertical center of gravity was 711 mm above ground, and the top of the bumper was 725 mm above ground. The safety performance of the guardrail was evaluated according to three major factors: structural adequacy, occupant risk, and vehicle trajectory after collision. These three evaluation criteria are defined explained in Table 5.1 in NCHRP Report 350. A summary of the test results and sequential photographs are presented in Figure 10. On impact with the guardrail, the right-front corner of the pickup began to crush inward as the bumper was captured by the rail. At 80 msec after impact the right-front tire of the vehicle contacted Post 12 and by 240 msec the vehicle had reached Post 13. The maximum dynamic deflection of the rail was 851 mm, which occurred 256 msec after impact at Post 12. The pickup became parallel to the rail at 344 msec, and reached Post 14 at 360 msec. The test vehicle exited the system 624 msec after impact, at a speed and angle of 42.4 km/hr and 8.3 degrees, respectively. The trajectory of the pickup caused it to come in contact with the rail downstream of the initial impact point, and it came to rest 25.3 m downstream of impact, with its bumper against the rail as illustrated in Figure 11. The overall damage to the vehicle was relatively minor, considering the impact conditions. The right-front corner of the vehicle was damaged, but the wheel assembly was intact and the tire remained inflated. There was no deformation of the occupant compartment. Additionally, damage to the guardrail is presented in Figure 11. Approximately 8 m of guardrail was deformed in the

FIGURE 9

Buffalo Rail simulation.

impact area, and the maximum permanent set deformation of 567 mm occurred midway between Posts 12 and 13. The post bolts were pulled through the rail at Posts 12, 13, and 14. There was no evidence of rail material failure anywhere along the rail, including the splice locations. The normalized occupant-impact velocities were determined to be 7.7 msec in the longitudinal direction, and 5.3 m/sec in the later direction. The highest 10-msec average occupant ridedown decelerations were 7.5 g (longitudinal) and 7.1 g (lateral). This test performance easily met the recommended criteria set forth in NCHRP Report 350 (3), and was thus determined to be satisfactory.

CONCLUSIONS W-beam guardrail has been the standard guardrail system used in the United States for the last 40 years, and has performed well under many impact conditions. However, this system was not designed to capture the bumpers of light trucks and utility vehicles, which have recently become a relatively large percentage of the vehicle population. In recognition of this, this study was undertaken to develop a guardrail system capable of capturing and redirecting a larger range of vehicle types and sizes. A new guardrail system, called the Buffalo Rail, was designed with a new cross-sectional shape with an effective depth of 311 mm (compared with 194 mm for the W-beam), a rail thickness of 2.28 mm (13 gauge), and a post spacing of 2500 mm. The safety performance

FIGURE 10

FIGURE 11

Summary of Buffalo Rail test.

Post-crash photographs.

80

Paper No. 970355

of the Buffalo Rail described in this study was found to be acceptable according to the procedures and criteria recommended for the three-quarter-ton pickup truck at Test Level 3 in NCHRP Report 350 for Test 3-11. ACKNOWLEDGMENTS The authors acknowledge Buffalo Specialty Products for their sponsorship of this project, specifically Joseph Batley. The authors also thank the MidAmerica Transportation Center for providing support for this research. A special thanks is given to the staff of the Midwest Roadside Safety Facility.

TRANSPORTATION RESEARCH RECORD 1599

2. Buth, C. E., W. L. Campise, L. I. Griffin III, M. L. Love, and D. L. Sicking. Performance Limits of Longitudinal Barrier Systems, Volume I: Summary Report. FHWA/RD-86/153, Texas Transportation Institute, Texas A&M University, College Station, May 1986. 3. Ross, H. E., D. L. Sicking, R. A. Zimmer, and J. D. Michie. NCHRP Report 350: Recommended Procedures for the Safety Performance Evaluation of Highway Features. TRB, National Research Council, Washington, D.C., 1993. 4. Mak, K. K., R. P. Bligh, and W. L. Menges. Crash Testing and Evaluation of Existing Guardrail Systems, Volume XI: Appendix J, Draft Report to FHWA, Texas Transportation Institute, Texas A&M University, College Station, Dec. 1995. 5. Hallquist, J. O., LS-DYNA3D User’s Manual, Livermore Software Technology Corporation, Livermore, Calif. 1995.

REFERENCES 1. Ross, H. E., Jr. Evaluation of Roadside Features to Accommodate Vans, Mini-Vans, Pickup Trucks and 4-Wheel Drive Vehicles (NCHRP Project 22-11). TRB, National Research Council, to be published.

Publication of this paper sponsored by Committee on Roadside Safety Features.