National Academies Press: OpenBook

Recommended Guidelines for Curb and Curb-Barrier Installations (2005)

Chapter: Chapter 2 - Literature Review

« Previous: Chapter 1 - Introduction
Page 5
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 5
Page 6
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 6
Page 7
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 7
Page 8
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 8
Page 9
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 9
Page 10
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 10
Page 11
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 11
Page 12
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 12
Page 13
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 13
Page 14
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 20
Page 21
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 21
Page 22
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 22
Page 23
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 23
Page 24
Suggested Citation:"Chapter 2 - Literature Review." National Academies of Sciences, Engineering, and Medicine. 2005. Recommended Guidelines for Curb and Curb-Barrier Installations. Washington, DC: The National Academies Press. doi: 10.17226/13849.
×
Page 24

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5CHAPTER 2 LITERATURE REVIEW INTRODUCTION Assessing the safety effectiveness of curbs attracted a con- siderable amount of attention in the early decades of roadside safety research. Curbs were thought to be a low-cost method of keeping vehicles on the roadway for at least some impact conditions. In 1953 the California Division of Highways per- formed a series of 149 full-scale tests on 11 different types of curb geometries in order to assess the safety effectiveness of curbs (4). This test series was followed in 1955 by another series of tests using the four best-performing curbs from the first series (7). The conclusion of the researchers was that bar- rier curbs, i.e., vertical curbs, should be at least 10 inches high, have undercut faces, and have a relatively smooth surface tex- ture. Other similar but less extensive studies were performed in Canada, Germany, and the United Kingdom (8, 9, 10). These early crash tests formed the basis of the AASHTO policy described in Chapter 1. Although the vehicle fleet has changed considerably since the time of these early studies, the current version of the AASHTO Green Book contains substantially the same recommendations as the 1965 Green Book regarding the use of curbs. The methods that have been employed for analyzing the safety effectiveness of curbs in earlier research included ana- lytical methods, vehicle dynamics codes, and full-scale crash testing. Each of these methods is discussed in the following sections. Information from selected studies from previous research on curbs and curb–barrier combinations are also pro- vided, followed by a summary of the literature review. ANALYSIS METHODS APPLIED IN THE STUDY OF CURB SAFETY Analytical Methods Most analytical work regarding vehicle impact into curbs has been concerned with either redirectional capabilities of vertical face curbs or their potential to cause rollover. If the impact speed and angle are plotted on a graph and different symbols used to denote redirection and mounting, then a curve like Figure 3 can be developed. Figure 3 shows the characteristics of two particular experimental curbs, the Trief and Elsholz curbs (9, 10). The line describes the boundary between redirective behavior and mounting behavior. Com- binations of impact speed and angle falling to the left of the curve would result in redirection, and those falling to the right would result in mounting the curb. The boundary between redirection and mounting can be described by K = V sin α where V is the impact velocity and α is the impact angle. In essence, this expression indicates that a given curb will redirect the vehicle when the lateral compo- nent of the impact velocity is less than some characteristic value. In his 1973 study of barrier curbs, Dunlap found that the characteristic lateral component of velocity for the Trief curb was 5 km/h and for the Elsholz curb was 14.6 km/h; thus, the Elsholz curb was more effective at redirecting vehicles than the Trief curb (11). Dunlap attempted to extend this basic methodology by treating the impact speed and angle as a random probabilis- tic variable along with the vehicle type. If the distribution of encroachment angles and vehicle speeds for a particular roadway is known, the percent of vehicles that would be redi- rected by each type of curb can be estimated (11). Dunlap used data from a specific roadway in Michigan for the speed distribution and the Hutchinson-Kennedy encroachment data for the impact angle distribution (12). For the specific site in Michigan, Dunlap found that the Elsholz curb could be expected to redirect 70% of the impacting vehicles and the Trief curb could only be expected to redirect 27%. Unfortunately, the curb characteristic lateral component of velocity is also a function of the characteristics of the vehi- cle that strike the curb and the type of curb. Some vehicles will have geometric, suspension, and handling characteristics more prone to mounting the curb than other vehicles. A curb’s ability to redirect a vehicle depends not only upon the speed and angle of impact, but also upon the dimensions of the curb, the surface material of the curb, if it is wet or dry, and the radius of the impacting tire. The boundary line between mounting and redirection shown in Figure 3, therefore, is only valid for a single type of test vehicle impacting a spe- cific type of curb. The dramatic changes in vehicle charac- teristics over the past decade seriously limit the validity of the findings of these early studies. The vehicles of today are lighter, have higher centers of gravity, and have lower pro- file tires. In addition, the passenger vehicle population has become much more diverse, now including pickup trucks, large and small sport utility vehicles (SUVs), and minivans,

as well as the traditional passenger car. Some of these vehi- cle types have proven to be less stable in collisions with traf- fic barriers than traditional passenger cars. While the testing done over the past 40 years provides some interesting insights, the results must be viewed carefully since the vehicle popu- lation of today is much different than it was during the 1960s. An analytical study on the safety of roadside curbs was conducted by Navin and Thomson at the University of British Columbia in 1997 (13). They developed the following empir- ical relationships to estimate the ability of a dry concrete curb to safely redirect a vehicle based on the findings produced in previous research: 6 where h is the height of the curb required to redirect the impact- ing vehicle, r is the radius of the tire in millimeters, Vr is the speed at redirection, θ is the impact angle, μN is the coefficient of friction of smooth rubber on test surface, and μCD is the coefficient of friction of smooth rubber on dry concrete. Note that the required height of the curb increases as the radius of the tires increases, the velocity of the vehicle increases, the angle of impact increases, or the friction coefficient increases. Vehicle Dynamics Codes The first computer simulation program used for the analy- sis of vehicle-curb impacts was the Cornell Aeronautical Lab- oratory Single Vehicle Accident program (CALSVA), devel- oped in the 1960s (14). It was used in the early 1970s by Wayne State University and the Highway Safety Research Institute (HSRI) at the University of Michigan to determine the redirection capability of various curb configurations (15). The CALSVA program, developed by Cornell Aeronautical Laboratory, was only capable of simulating a limited range of impact scenarios because of the simplicity of the program; however, it did serve as a precursor to more advanced com- puter simulation codes. The second generation version of CALSVA was the High- way Vehicle-Object Simulation Model (HVOSM) (16). This program has been used extensively in conjunction with full- scale crash testing to study vehicle dynamics during impact with curbs. A comprehensive review of these studies will be presented in subsequent sections of this chapter. The vehicle dynamics code VDANL (Vehicle Dynamics Analysis, Non Linear) was developed in the 1980s by the NHTSA and Systems Technology, Inc. (STI). It is a com- prehensive vehicle dynamics simulation program that runs on a PC in a Windows environment. It was designed for the analysis of passenger cars, light trucks, articulated vehicles, and multipurpose vehicles, and it has been upgraded over the years to expand and improve its capabilities. It now permits analysis of driver-induced maneuvering up through limit per- formance conditions defined by tire saturation characteris- tics, as well as driver feedback control features. VDANL was chosen by the FHWA for use in the Interactive Highway Safety Design Model (IHSDM) (17). The IHSDM h r Vr N CD = ⎛⎝ ⎞⎠⎡ ⎣ ⎢⎢⎢⎢ ⎤ ⎦ ⎥⎥⎥⎥ sin . θ μμ 50 1 3 5 Figure 3. Performance characteristics of the Trief and Elsholz curbs (9).

program is used to assess new roadway designs by using a driver performance model to simulate the vehicle/driver response when traversing the proposed roadway configuration. The Driver Performance Model in IHSDM estimates drivers’ speeds and path choices along a roadway, and this informa- tion is provided as input to VDANL, which estimates vehicle kinematics such as lateral acceleration, friction demand, and rolling moment. The information from VDANL is used to identify conditions that could result in loss of vehicle control (i.e., skidding or rollover). Full-Scale Crash Testing Although advancements in computer simulation programs have made it possible to accurately reproduce and predict complex impact events, full-scale testing is still essential in evaluating the safety performance of curbs and other road- side appurtenances. To evaluate the performance of roadside safety barriers, impact conditions must meet the standard testing procedures accepted by the FHWA. The first proce- dures document was published by the Highway Research Board in 1962 (18). The later revisions of the procedures were made by the National Cooperative Highway Research Program. The latest revisions of the testing procedures were published in NCHRP Report 350 in 1993 (19). From 1981 to 1992 crash tests were conducted according to the test requirements specified in NCHRP Report 230 (20). The test conditions required for evaluation of guardrail in NCHRP Report 230 involved a 2000-kg sedan impacting the guardrail at a speed of 100 km/h and an angle of 25 degrees. The most important change in NCHRP Report 350 was that the large passenger sedan had virtually disappeared from the vehicle population, and new vehicle types, such as mini- vans, SUVs, and pickup trucks, had emerged in their place. Since the first testing procedures specified in Highway Research Circular 482 up until NCHRP Report 350, the large car sedan (i.e., a 2040-kg car) had served as the crash test vehicle representing the fleet of large passenger vehicles. NCHRP Report 350 replaced the large car with a 2000-kg pickup truck. The challenges that the pickup truck introduced to the crash testing procedures were due to its high, more for- ward center of gravity making it much more unstable during impacts than its predecessor, the large sedan. The performance of a curb/guardrail combination are eval- uated using test conditions specified in NCHRP Report 350 for evaluating the crashworthiness of the length of need (LON) section of a guardrail. There are currently two tests that are required in Report 350 to evaluate guardrail systems for use along high-speed roadways: (1) Test 3-11, which involves a 2000P pickup truck (e.g., Chevrolet 2500) impacting the guardrail at a speed of 100 km/h and an impact angle of 25 degrees, and 7 (2) Test 3-10, which involves an 820C (e.g., Geo Metro) impacting the guardrail at a speed of 100 km/h and an impact angle of 20 degrees. A guardrail system that meets the evaluation criteria for Tests 3-10 and 3-11 in NCHRP Report 350 is generally considered acceptable for use on all TL-3 roadways within the United States. EFFECT OF CURBS ON VEHICLE STABILITY Dunlap, 1973 (11, 21) The objective of Dunlap’s research was to determine how far in front of the barrier the curb should be placed to achieve the best redirection performance from the curb–traffic barrier system. Dunlap examined all the test data available in the early 1970s and found that the results were difficult to gen- eralize. While there were cases of vehicles vaulting over a guardrail or bridge railing when a curb was used in front of the guardrail, in many cases the guardrail itself had structural problems so it was difficult to assess the contribution of the curb to the failure. Dunlap performed computer simulations of a variety of curb and barrier combinations using HVOSM to determine the risk of overriding the barrier. Dunlap’s analy- sis indicated that for the six curb and barrier combinations studied, vaulting was not expected to be a problem. This analy- sis, however, has several serious limitations not least of which is the validity for barrier impact analysis of the HVOSM com- puter program that was being used at the time. Dunlap’s work does, however, illustrate two important points: (1) computer simulation is one possible method for assessing a variety of curb–barrier geometries and (2) the conventional wisdom that curbs should not be used in front of semirigid barriers warrants more careful investigation. Olsen et al., 1974 (22) Olsen and other researchers at Texas Transportation Insti- tute (TTI) conducted a study to investigate how various types of curbs affect vehicle response, such as redirection, trajectory, path, roll, pitch, and accelerations. Their study involved full- scale tests and simulations of vehicles traversing various types of curbs. Eighteen full-scale tests were conducted on types B and D curbs (see Table 1); nine full-scale tests were conducted on each curb type at speeds of 48, 72, and 97 km/h and at 5-, 12.5-, and 20-degree encroachment angles. The HVOSM computer program was used to simulate vehicle impact with three different curb types: AASHTO curb types B, D, and G. Although in the study, the curbs were referred to as C, E, and H curbs (which is consistent with the nomenclature of the AASHTO Blue Book), the AASHTO Green Book now refers to these curbs as B, D, and G, respectively. Nomenclature

throughout this document will use the Green Book designa- tions. Twelve curb impacts were simulated on each curb type at impact speeds of 48, 72, and 97 km/h and at 5-, 12.5-, and 20-degree encroachment angles. Impacts at 121 km/h were also simulated at 5-, 10-, and 15-degree encroachment angles. The test vehicle used in the study was a 1963 Ford four-door sedan with heavy-duty suspension. The vehicle’s mass was 1,905 kg, and the center of gravity of the vehicle was 610 mm above ground. The test vehicle is shown in Figure 4. Olsen et al. found that AASHTO types B, D, and G curbs, which are sloping curbs 150 mm or less in height, provide no redirection for a large passenger vehicle, such as a 1900-kg sedan, traveling at speeds greater than 72 km/h at encroach- ment angles greater than 5 degrees. They also found that type B and D curbs can produce, under certain speed and encroach- ment angles, vehicle ramping high enough to allow the bumper height to equal or exceed the height of a typical guardrail, as illustrated in Figure 5. Such vehicle trajectories may result in a vehicle vaulting over the top of the rail or snagging on the tops of the posts and flipping over. Whether the vehicle penetrates behind the barrier or is redirected is, of course, influenced by other fac- 8 tors, including barrier configuration, lateral stiffness proper- ties of the barrier, and impact conditions, as well as vehicle characteristics, such as bumper shape and vehicle kinematic properties. The trajectory of the vehicle after mounting a curb must allow the vehicle to contact the guardrail, or other roadside device, at the appropriate height. Test number Approach speed (mph) Encroachment angle (degrees) Maximum bumper height during vehicle trajectory (inches) Curb Type D N-2a 30.4 5.1 24.1 N-3a 45.6 5.0 24.3 N-4 59.3 4.6 23.9 N-5 32 11.6 20.8 N-6 45.3 11.1 23.7 N-7 63.6 12.6 23.5 N-8 32.7 18.5 23.5 N-9 41.8 18.7 21.9 N-10 63.0 17.6 23.3 Curb Type B N-11a 34.2 4.9 26.2 N-12 44.7 5.1 24.8 N-13 34.2 11.2 23.8 N-14 43.5 12.8 23.1 N-15 32.1 17.4 22.1 N-16 43.0 18.4 23.5 N-17 66.5 5.1 24.3 N-18 62.2 12.3 21.4 N-19 61.5 18.6 23.0 a Vehicle redirected TABLE 1 Summary of full-scale test results from Olsen et al. (22) Figure 4. Vehicle used in Olsen et al.’s study (22).

Olsen et al. found that for 150-mm-high AASHTO B and D curbs an increase in either speed or impact angle resulted in greater lateral distances to the maximum rise point and higher vertical position of the vehicle at the maximum rise point. The encroachment angle had a more notable effect on the maximum rise point and position than did vehicle speed, when vehicle speed was greater than 100 km/h. The maximum rise height of the bumper, predicted from the simulations, was approximately 737 to 787 mm and occurred in the range of 2.44 to 3.0 m behind 150-mm-high curbs. The height of a typical W-beam guardrail is 686 mm, as shown in the sketch in Figure 5. The maximum rise height during impact with the type G curb was only slightly affected by vehi- cle speed and encroachment angle. The maximum vertical rise of the vehicle impacting the type G curb was less than 50 mm. Furthermore, the maximum rise height did not increase an appreciable amount for speeds greater than 48 km/h, indi- cating that the maximum rise height during impact with the type G curb is relatively independent of vehicle speed and impact angle. It was concluded that the maximum rise point was depen- dent on the combination of vehicle roll and pitch caused by striking the curb. When the wheel impacts the curb, the loads are distributed to the other three wheels, particularly the other front wheel. If the impacting wheel rises too quickly, then the vertical tire force will be sufficient to bottom out the suspen- sion, introducing shock loads. In addition, excessive pitch and roll angles are produced when the fully compressed sus- pension unloads. The effect that curb geometry has on damp- ing the roll angle during wheel impact obviously differs with the height and the steepness of the curb face. The pitch and roll angles produced by simulated collisions with type B and D curbs were as much as twice those produced by collisions with the type G curb. Curbs that are 150-mm high and set in front of a 685-mm W-beam guardrail at a 0.61-m lateral offset may result in the vehicle impacting the guardrail at a point below the lower edge of the rail, possibly causing snagging, as shown in Figure 5. During impacts with the 150-mm-high curbs, the bumper would dip down slightly and then began to rise as the vehicle crossed the curb. If the angle of impact is such that 9 the bumper is close to the guardrail before the wheel impacts the curb, then the dipping event would cause the bumper to impact the guardrail just below the W-beam rail. Note that the lower edge of the guardrail is 533 mm above the pavement surface due to the 150 mm elevation of the curb; whereas, the lower edge of the rail is only 381 mm above ground level in normal configuration. An initial dipping motion of the bumper was not evident during impact with the type G curb, and the bumper contacted the guardrail on the face of the W-beam in all impact cases. The simulation study by Olsen et al. also demonstrated that the stiffness of the vehicle’s suspension had little effect on vehicle trajectory. A summary of full-scale test results performed in Olsen et al.’s study is given in Table 1 and a summary of their HVOSM simulation results is given in Table 2. The HVOSM model had a disk wheel that was not detailed enough to accurately simulate wheel contact with a curb. The simulation results in NCHRP Report 150 predicted that full-size cars would be redirected by a 13-in.-high Type X curb in 60-mph impacts up to 12.5 degrees. However, in 60-mph crash tests, the test vehicles crossed the curb. The disparity between the test results and the HVOSM predic- tions was more apparent in the high-speed tests, particularly between the predicted roll and bumper rise and those values measured from the test data. Ross and Post, 1975 (23) Researchers at TTI conducted a study to evaluate auto- mobile behavior when traversing selected curb configura- tions and sloped medians and, also, to evaluate the potential for a vehicle to vault over roadside barriers placed in combi- nation with curbs or sloped medians. HVOSM was used to simulate vehicle impacts with 150-mm-high and 200-mm-high curbs, modified curbs, and slopes. The researchers also com- pared the effects of standard curb shapes to various retrofit alternatives, such as installing wedge-shaped asphalt plugs in front of the curbs and replacement of the curbs with slopes. It was concluded from the simulation results that traffic bar- riers should not be placed near curbs due to the probability of Lateral Offset from Curb Curb337 mm Figure 5. Possible trajectory of vehicle bumper relative to guardrail height.

vehicles vaulting or underriding the barrier. They also showed that problems with barriers on raised curb-medians or curb– guardrail configurations could be reduced in certain situations by sloping the median or the roadside to the top of the curb. Holloway et al., 1994 (24) Three types of sloping curbs commonly used by the Nebraska Department of Roads (NDOR) were investigated for safety performance through a combination of full-scale testing and computer simulation using HVOSM. The curb types investigated included a 100-mm lip curb (13 slope on curb face), a 150-mm lip curb (13 slope on curb face) and a 150-mm AASHTO type I curb. The AASHTO type I curb, shown in Figure 6, is the curb type most widely used by NDOR. The test matrix in the study included 23 full-scale 10 Curb Vehicle speed (mph) Impact angle (deg) Max roll angle (deg) Max pitch angle (deg) Max bumper height above curb (inches) Lateral distance to max rise point (ft) Bumper height above curb at 2-ft offset (inches) Type B (6-in.) 30 20 +8.8 2.9 22 5 12 45 20 -8.9 3.0 26 8 11 60 12.5 -13 2.0 27 7 13 60 20 -8 2.0 29 10 10 75 10 -15.5 2.0 30 6 13 75 15 -10.2 1.8 30 10 12 Type D (6-in.) 30 12.5 -9.5 2 21 4 13 30 20 -8 2.5 21 6 11 45 12.5 -11 2 23 5 12 45 20 -8 2.2 25 8 11 60 5 -11.2 2 23 3 17 60 12.5 -12 2 25 6 13 60 20 -9.5 2.5 31 10 11 75 5 -12 1.5 23 4 16 75 10 -13 2 25 6 13 75 15 -11 2 31 9 12 Type G (4-in.) 30 12.5 -5 1 18 5 13 30 20 -3 1 18 9 12 45 5 -7 1 20 3 15 45 20 -4 1 20 10 14 60 5 -7 1 20 4 15 60 12.5 -5 1 20 8 13 60 20 -3 1 20 10 13 TABLE 2 Summary of HVOSM simulation results from Olsen et al. (22) 6" 2% 1" 1/4" 6" 18" 8 -2/16 " R 2" 4" R Figure 6. AASHTO Type I Curb.

tests: 13 tests on the 100-mm lip curb, 2 tests on the 150-mm lip curb, and 8 tests on the AASHTO type I curb. The three curbs tested were found to have little potential for causing a vehicle to lose control during tracking impacts, and, thus, the researchers concluded that the curbs would not pose a significant hazard to vehicles impacting in a tracking mode. Although the 100-mm curb performed better than the 150-mm curbs in all impact conditions, the safety benefit was not considered significant. It was also concluded that the per- formance of W-beam guardrails could be adversely affected when installed behind curbs and that, when curb–guardrail combinations are necessary, the curb should be placed behind the face of the guardrail to minimize the potential for vehicle ramping. The testing area was on a negative grade that may have had some effect on the vehicle kinematics during impact. Tests were conducted using two types of test vehicles: a small car with a mass of 817 kg (1984 Dodge Colt) and a large car with a mass of 2,043 kg (1986 Ford LTD). The center of gravity of the test vehicles were 533 mm and 572 mm for the 817-kg and 2043-kg vehicles, respectively. The impact speeds used in the full-scale tests were 64.4, 72.4, 80.5, and 88.5 km/h at encroachment angles of 5, 12.5, and 20 degrees. Vehicle decelerations were very low, indi- cating that there is little risk of occupant injury as a direct result of curb impact. The yaw rate and yaw angle were also very low, indicating that there was minimal redirection of the vehicles as they impacted and mounted the curbs. Thirteen full-scale tests were conducted on a 100-mm lip curb, and two full-scale tests were conducted on a 150-mm lip curb. For low-angle impacts on the 100-mm curbs with the 817-kg vehicle, the maximum roll and pitch angles increased as the impact velocity increased; values ranged from 5.6 to 9.0 degrees and 0.7 to 1.4 degrees for roll and pitch angles, respectively. For the moderate- and high-angle impact tests, the maximum roll angle increased as the impact speed increased, while the maximum pitch angle decreased with an increase in impact speed. The maximum roll angle in the tests was 9.3 degrees, and the maximum pitch angle was 2.6 degrees. Thus, the pitch and roll angles were considered to be relatively insignificant in terms of producing loss of vehicle control. It was also concluded in the study that there was only a slight potential for an 817-kg vehicle to underride a standard 686-mm W-beam guardrail when the 100-mm lip curb is placed in combination with the guardrail. The greatest poten- tial of the vehicle vaulting over the barrier would be when the barrier is located in a region 0.76 m to 2.74 m behind the curb. Similarly, for low-angle impacts with the 2043-kg vehi- cle impacting the 100-mm lip curb, the roll and pitch angle increased as the impact speed increased. The maximum roll and pitch angles were 7.2 degrees and 1.1 degrees, respec- tively, for the low-angle impacts. The maximum roll and pitch angles for the high-angle impacts were 7.2 degrees and 2.0 degrees, respectively. There were only two tests conducted on the 150-mm lip curb. In these two tests, a 2043-kg vehicle 11 impacted the curb at an encroachment angle of 20 degrees and at impact speeds of 72.4 and 86.9 km/h. The maximum roll and pitch angles were 7.8 degrees and 2.6 degrees, respec- tively. The tests indicated that there was a slight potential for the vehicle to underride a standard W-beam guardrail, if the guardrail was placed within 1.22 m of the curb; how- ever, the tests also indicated that there was very little potential for the vehicle to vault over the barrier. Tests conducted on the AASHTO type I curb resulted in maximum roll and pitch angles of 9.7 degrees and 3.1 degrees, respectively. Although the angular displacements of the vehi- cle during impact with this curb were somewhat higher than those produced in impacts with the lip curbs, the potential for loss of control of the vehicle was again considered very low. The driver of the vehicle in the study reported that the sus- pension system fully compressed and bottomed out against the suspension bumper stops during impact with the 150-mm curbs and a small jolt was felt. The trajectory of the vehicle during the tests indicated there was a potential for underride of a standard W-beam guardrail if the barrier is located within 1.22 m of the curb; however, there did not appear to be any significant risk of the vehicles vaulting over such a barrier. The HVOSM was also used to investigate alternate impact conditions. Simulation models of the 23 full-scale tests were developed, and the results were compared to the full-scale tests to validate the model. An additional 55 simulations were then performed. Thirty-one simulations were performed to supplement the original 23 impact scenarios, including 5 sim- ulations with the 100-mm lip curb, 16 simulations with the 150-mm lip curb, and 10 simulations with the 150-mm type I curb. Another 24 simulations were performed to evaluate the effects of curb impact with the curb placed on flat grade. The simulations with the lip curbs were performed with vehicle velocities of 72.4 and 88.5 km/h at encroachment angles of 5 and 20 degrees. The results of the simulations with the 100-mm lip curb showed no potential for either under- riding or vaulting a W-beam guardrail installed behind the curb. The results of the simulations with the 150-mm lip curb indicated that the small vehicle (817 kg) may underride a W-beam guardrail if the guardrail is placed within 1 m of the curb, and it is likely to vault over a guardrail placed 0.46 to 3.7 m behind the curb. The simulations with the large vehi- cle (2043 kg) indicated a slight potential for underriding a W-beam guardrail located within 1 m of the curb, and vault- ing of the guardrail was likely if the barrier was placed in a region of 0.61 to 3 m behind the curb. The simulations with the AASHTO type I curb indicated that impact with the curb could cause underride of a W-beam guardrail placed within 0.61 m of the curb. For small car impact, a potential for vaulting existed if the guardrail was placed 0.46 m to 3.0 m behind the curb. For large car impact, a potential for vaulting existed if the guardrail was placed 0.46 m to 3.7 m behind the curb. The additional 24 simulations were performed on all three curb types to investigate the effects of impact with the curbs

placed on flat grade. Impact conditions included vehicle speeds of 72.4 and 88.5 km/h and encroachment angles of 5 and 20 degrees. The results of these simulations showed only minor differences in angular displacements of the vehi- cle, compared with the simulations with the curb placed on a negative grade (i.e., the test area was on a negative grade). Nontracking impacts of vehicles with the three curb types were also investigated using computer simulation; however, no test data were available for validating the results. Impact conditions used in the study were based on Appendix G of NCHRP Report 350 and from accident data analysis stud- ies. All simulations were performed with vehicle speed of 80.5 km/h and impact angle of 20 degrees. Three initial posi- tions of the vehicle were investigated: (1) 150 degree yaw angle with 50 deg/sec yaw rate, (2) negative 30 degree yaw angle with a negative 25 deg/sec yaw rate, and (3) 180 degree yaw angle with 50 deg/sec yaw rate. They found that these curbs may be traversable over a wide range of vehicle orien- tations and impact conditions, and the curbs pose little threat of vehicle rollovers during impact. EFFECT OF CURBS INSTALLED IN CONJUNCTION WITH GUARDRAILS Buth et al., 1984 (25) During the 1980s, the FHWA sponsored the testing of numerous bridge railings, some of which included curbs. In particular, TTI tested a New Hampshire bridge rail system with a curb protruding in front of the barrier face, and a Col- orado Type 5 bridge rail system with a curb flush with the face of the barrier. In both tests, the front impact-side wheel was damaged during impact with the curb, and the wheel wedged between the curb and the bottom rail of the traffic barrier. The performance of both bridge railings was considered unsat- isfactory, but it should also be noted that both railings had other poorly designed features that may have contributed to the poor performance. Bryden and Phillips, 1985 (26) Bryden and Phillips performed 12 full-scale crash tests for the New York Department of Transportation to evaluate the performance of a thrie-beam bridge-rail system. Two tests were conducted with a 150-mm curb placed flush with the face of the thrie-beam rail. The tests involved a 2043-kg Dodge sta- tion wagon impacting the system at approximately 100 km/h at an impact angle of 26 degrees. The vehicle remained stable and was smoothly redirected in both tests. FHWA Memorandum, February 28, 1992 (27 ) The results of a series of crash tests conducted by ENSCO, Inc., were reported in an FHWA Memorandum distributed 12 on February 28, 1992. The tests involved various types and sizes of vehicles impacting W-beam guardrails with curbs placed behind the face of the W-beam rail element. In the cases involving curbs 150 mm high or higher, it was found that the vehicle would vault over the guardrail, if the guard- rail deflected enough for the wheels to mount the curb. In crash tests in which the 100-mm AASHTO Type G curb was placed behind the face of the W-beam, the vehicle became airborne when guardrail deflection permitted the wheels to mount the curb; however, the vehicle did not vault the rail. The best alternative for reducing the safety hazards associated with guardrail-curb systems is to stiffen the guardrail. Stiff- ening the guardrail reduces guardrail deflection and reduces the potential of the vehicle contacting the curb. In tests where the guardrail was sufficiently stiff, the tires of the vehicle did not contact the curb, and the vehicle was redirected in a much more stable manner. Below is a summary of the ENSCO tests. Test Number 1862-1-88 A 2452-kg pickup truck impacted a G4(1S) guardrail sys- tem with a 203-mm-high concrete curb (AASHTO type A) installed behind the face of the W-beam. The impact speed was 100 km/h and the impact angle was 20 degrees. There was significant deflection of the guardrail, and the wheels of the vehicle contacted the curb. The vehicle vaulted over the guardrail. Test Number 1862-4-89 An 817-kg car impacted a G4(1S) guardrail system with a 150-mm-high asphalt dike. The impact speed was 100 km/h and the impact angle was 20 degrees. The wheels of the vehi- cle did not contact the curb during the crash event, and the vehicle was smoothly redirected. Test Number 1862-5-89 A 2043-kg sedan impacted a G4(1S) guardrail system with a 150-mm-high asphalt dike. The impact speed was 100 km/h and the impact angle was 25 degrees. There was significant deflection of the guardrail, and the wheels of the vehicle con- tacted the curb. The vehicle vaulted over the guardrail. Test Number 1862-12-90 A 2452-kg sedan impacted a G4(1S) guardrail system with a 100-mm-high concrete curb (AASHTO type G). The impact speed was 100 km/h and the impact angle was 25 degrees. The vehicle became airborne but did not vault the guardrail.

Test Number 1862-13-91 A 2043-kg sedan impacted a G4(1S) guardrail system stiffened with a W-beam bolted to the back of the steel posts. A 150-mm-high asphalt dike was placed behind the front face of the W-beam. The impact speed was 100 km/h and the impact angle was 25 degrees. The guardrail system was sufficiently stiff to prevent the wheels of the vehicle from impacting the curb. The vehicle was successfully redirected. Test Number 1862-14-91 A 2043-kg sedan impacted a G4(1S) guardrail system stiff- ened with a C6x8.2 hot-rolled channel rub rail. A 150-mm- high asphalt dike was placed behind the face of the W-beam. Again the guardrail system was sufficiently stiff to prevent the wheels of the vehicle from impacting the curb and the vehicle was successfully redirected. The vehicle speed change at redi- rection, however, was greater than the allowable (24 km/h) according to NCHRP Report 230; thus the system did not meet all required safety criteria. 13 Holloway and Rossen, 1994 (28) A study was conducted by Holloway and Rossen at Mid- west Roadside Safety Facility at the University of Nebraska– Lincoln that involved a full-scale crash test on Missouri’s 150-mm-high vertical curb placed behind the face of a strong- post W-beam guardrail (i.e., G4(1S)). Missouri’s 150-mm- high vertical curb is very similar to the AASHTO type B curb, except that the Missouri vertical curb is on a flat grade and has very little rounding on the top and bottom edges of the curb. The impact conditions for the test was in accordance with NCHRP Report 230 specifications; a 2043-kg test vehi- cle (1985 Ford LTD) impacted the system at 96 km/h at 25.1 degrees. The center of gravity of the test vehicle was 597 mm above ground. A summary of test M06C-1 is shown in Figure 7. During the test, the right front tire contacted the curb 20 mil- liseconds after initial contact with the guardrail and mounted the curb soon after. The maximum roll angle was negative 14 degrees (the roll angle was away from the system). The vehicle exited the rail at 706 milliseconds at a speed of 64 km/h and an angle of 6.2-degrees. Vehicle decelerations and tra- jectory were well within the recommended limits of NCHRP Figure 7. Summary of results for MwRSF Test M06C-1 (28).

Report 230. As a result of the test, the researchers concluded that the system performed satisfactorily and the Missouri Department of Transportation should continue to use the guardrail-curb system where warranted. Polivka et al., 1999 (29) A study was conducted by researchers at the Midwest Roadside Safety Facility (MwRSF) at the University of Nebraska-Lincoln to evaluate the effects of an AASHTO type G curb (i.e., 102 mm high and 203 mm wide) placed flush behind the face of a G4(1S) guardrail system. Test NEC-1 was conducted with impact conditions recommended by NCHRP Report 350 TL-3, which involves a 2000-kg pickup truck (1991 GMC 2500) impacting at a speed of 100 km/h at an impact angle of 25 degrees (19). Sequential photographs of the crash test are shown in Figure 8. The center of gravity of the test vehicle was 737 mm. The test installation was a standard 53.34-m-long G4(1S) guardrail system anchored on both the upstream and down stream ends of the system by an inline breakaway cable ter- minal with a strut between the two end posts. The guardrail ruptured at a splice connection, thus the test was a failure. There was little vertical displacement of the vehi- cle as it crossed the curb in the full-scale test, and there seemed to be very little potential for underride or vaulting of the bar- rier. The anchor posts split during the collision, as shown in Figure 9, and there was a loss of tension in the W-beam, which resulted in pocketing and rupture of the W-beam rail at a splice connection. The splice failure was attributed to contact and snagging of the post blockout against the W-beam rail splice. The post twisted as it was pushed back in the soil, causing the bottom corner of the blockout to push up against the cor- ner of the W-beam rail splice. This resulted in a tear in the W-beam at the lower downstream bolt location. It was sug- gested that the guardrail-curb combination could be signifi- cantly improved by increasing the capacity of the W-beam rail. 14 Bullard and Menges, 2000 (30) This study was conducted by researchers at the TTI and involved the evaluation of a 100-mm-high asphaltic curb, set out 25 mm from the face of the rail of a G4(2W) strong-post guardrail system, as shown in Figure 10. TTI test 404201-1 was conducted at the TTI on May 23, 2000, and involved a Chevrolet C2500 pickup impactingFigure 8. Sequential video frames from Test NEC-1 (29). Figure 9. Guardrail terminal damage during Test NEC-1 (29).

the curb-and-barrier system at 101.8 km/h at an angle of 25.2 degrees (i.e., NCHRP Report 350 Test 3-11). During the test, there was significant movement of the anchor system as the foundation of the anchor posts moved in excess of 70 mm. The test was successful; however, there was considerable damage to the guardrail system, as shown in Fig- ure 11. The extent of damage to the system was much greater than that of previous crash tests on the G4(2W) guardrail sys- tem without a curb present (31). From reviewing the film from the crash test and the test report, it is believed that the exces- sive damage to the system is due, in part, to the use of poor grade posts in the guardrail installation. Many of the posts split vertically during impact along preexisting splits passing through the bolt hole location in the posts, as shown in Fig- ure 12. A summary of Test 404201-1 is shown in Figure 13. Polivka et al., 2001 (32) This study involved the second phase of the curb-and- barrier impact investigation conducted by MwRSF, in which 15 the 102-mm AASHTO type G curb was installed in combi- nation with a strong-post guardrail system. Test NEC-2 was conducted with impact conditions recommended in NCHRP Report 350 TL-3. The test vehicle was a 2000-kg pickup truck (1994 GMC 2500) and the impact speed and angle were 100.3 km/h and 28.6 degrees, respectively. The center of grav- ity of the test vehicle was 667 mm. The test installation was a modified G4(1S) guardrail with routed wood blockouts. In order to reduce the potential for rup- ture of the rail, two layers of 12-gauge W-beam were nested over a 26.67-m section of the guardrail. This modification was incorporated based on the results of test NEC-1, con- ducted in the first phase of the study, in which a splice rup- ture occurred during impact. The total length of the guardrail was 53.34 m, including an inline breakaway cable terminal located at both ends of the system. The vehicle vaulted during impact and was airborne for much of the impact event. While the vehicle was airborne, it did get over the rail, as shown in Figure 14; however, the vehicle remained upright, came down on the front side of the guardrail, and satisfied all safety requirements of NCHRP Report 350. A summary of test NEC-2 is shown in Figure 15, which was taken from Polivka et al. EFFECTS OF CURB TRIP ON VEHICLE STABILITY DeLeys and Brinkman, 1987 (33) Computer simulation was used in a study to determine the dynamic response of small and large passenger cars travers- ing various sideslope, fill-embankment, and ditch configura- tions. Both tracking and nontracking departures from the roadway were investigated. A modified version of HVOSM was used in this research that improved the program’s appli- cation to rollover situations. The modifications to the program were made by McHenry Consultants, Inc. These modifications Figure 10. Guardrail–curb installation for TTI test 404201-1. Figure 11. Guardrail damage in TTI Test 404201-1.

16 included further development of the tire model and the addi- tion of a tire/deformable-soil interaction model to the program. A literature review and analysis of accident data recorded in the 1979–81 National Accident Sampling System (NASS) was performed; some of the principal findings from that review are quoted below: • Embankments, ditches, and culverts are the roadside ter- rain features cited as being most frequently involved in overturn accidents. However, detailed information on the geometry of the terrain and whether the rollover was caused by vaulting, or by the wheels hitting a small obsta- cle, or by the wheels digging into soft soil and tripping the vehicle is generally lacking in accident data files. • In most (50% to 80%) of the rollover accidents, the vehi- cles were skidding out of control at a large yaw angle prior to overturning. • About half of all accidental departures from the road- way occurred at path angles greater than 15 degrees, and the majority of the vehicles were estimated to have been traveling at speeds less than 64 to 80 km/h. Full-scale tests were performed with an instrumented 1979 VW Rabbit automobile to provide data for evaluating the validity of the modified computer program. The tests included spinout of the car on level turf, dragging the car over a sod field, traversals of fill-embankments, and traversals of the front slope of a wide ditch. Motion-resistance force data were collected in these tests. They were used for obtaining tire/ ground coefficients of friction for typical roadside terrain surfaces, as well as for validating the computer simulation models. The drag tests were performed by attaching two steel cables to the center of the front and rear wheels on the right side of the vehicle. A load cell was installed on each cable to measure the forces as the vehicle was pulled sideways over the ground surface at speeds of 16 to 24 km/h. The data from the tests indicated that the average coefficient of friction between the tires of the VW Rabbit and the sodded ground surface was typically about 0.5. The modified version of HVOSM provided reasonable accuracy of the simulations of the tests on the various road- side terrains. The authors do point out, however, that “the study did not thoroughly establish the extent to which the model accounts for all of the various real-world conditions that contribute to vehicle rollover” (33). Over 200 HVOSM simulations of vehicles traversing var- ious sideslopes, fill-embankments, and ditch configurations were used to determine how much these roadside conditions affect the rollover tendencies of vehicles. In addition to the VW Rabbit model (1093-kg vehicle) that was developed and validated with the full-scale tests, two other vehicles were modeled: one was a relatively light vehicle and the other a much heavier vehicle. The lighter vehicle had a mass of Figure 12. Posts split vertically during TTI Test 404201-1 along preexisting splits in posts.

17 Figure 13. Summary of results of TTI Test 404201-1 from Bullard and Menges.

18 Figure 14. NCHRP Report 350 Test 3-11 impact with modified G4(1S) guardrail with nested 12-gauge W-beams and a 102-mm curb under the rail (32). Figure 15. Summary of results of Test NEC-2 from Polivka et al. (32). 816 kg and was identical to the VW Rabbit model, except that the mass and moments of inertia were different. The heavier vehicle model had a mass of 2,018 kg, representing the larger class of passenger cars, and its physical character- istics were defined in HVOSM using available data typical for that vehicle type. The conclusions that the authors made from the study, that pertain to the use of HVOSM for predicting the dynamic response of vehicles traversing various types and shapes of terrain, are presented below: • The modified HVOSM has been demonstrated to be capable of predicting the response of vehicles operating on off-road terrains with reasonable accuracy. The development and incorporation of the deformable-soil model in HVOSM is considered an important improve- ment since it allows simulation for the effects of tire sinkage in soil which has been identified as one of the leading causes of rollover. However, evidence of the validity of the deformable-soil model is clearly still very limited. • The relatively few simulations that resulted in vehicle rollover in this study point to the dynamic nature of the rollover phenomenon, which is sensitive to the complex interactions of many factors whose effects are not inde- pendent. Adequate vehicle parametric data for the severe operating regime associated with the rollover response are generally lacking. Among the most important of these are definitive data for tire properties under the

high tire load and large slip and camber angle conditions that prevail in most rollover events. • Ultimately, the vehicle rollover potential associated with roadside features is reflected by real-world accident expe- rience. From the literature review performed as part of the study, it is apparent that the existing accident data base lacks the comprehensive and detailed information neces- sary to define the conditions that lead to rollover for dif- ferent vehicle types. For example, data contained in acci- dent data files, such as NASS and FARS, usually provide little or no information regarding the geometrics of the accident site (e.g., steepness of slopes, embank- ment height and roundings), whether the vehicles were tripped by a surface irregularity or as a result of tire ruts in soft soil, where rollovers were initiated with respect to the terrain feature (sideslope, backslope, toe of embank- ment, etc.), vehicle trajectory, and so forth Cooperrider et al., 1990 (34) Researchers at Failure Analysis Associates, Inc. (FaAA) performed a study to investigate the mechanics of vehicle rollovers. It was their perception that the experimental and analytical methods that were being used at that time (late 1980s) did not accurately represent real-world vehicle roll- overs. Their investigation involved full-scale tests in which vehicles were tripped by three different trip mechanisms: sliding into a curb, sliding in soil, and being thrown from a dolly. They also developed a simple analytical technique to characterize the mechanics of these different trip modes based on a constant force method. 19 Eight full-scale tests were conducted using four different vehicle types to examine the rollover mechanics of vehicles tripped by a curb, rolled off a dolly, and tripped by tire-soil interaction. The test matrix and results from the study are presented in Table 3. For the curb impact tests, a 152-mm-square section of steel box tubing, rigidly affixed to the roadway, was used to repre- sent a curb. The vehicles were towed sideways and released just prior to contact with the curb. The friction between the tires and the road surface was reduced by applying soap film to the roadway. In order to more accurately represent the impact conditions of vehicles in real-world accidents, where an initial roll of the vehicle would be produced from the tire- ground interaction, a roll angle of 2.5 degrees was built into the test vehicles by extending the left suspension with wood blocks. Two of the five curb impact tests resulted in rollover. The three vehicles that did not rollover sustained excessive dam- age to their wheels or axles during impact. Failure or partial failure of these components may result in a reduction of load applied to the vehicle, which reduces the potential for roll- over. The tripping force must be applied for sufficient dura- tion to cause rollover. For the vehicles that did roll over, the average maximum decelerations at the center of gravity was 12.4 Gs, compared with maximum decelerations of 1.62 Gs and 1.3 Gs in the soil trip tests and dolly tests, respectively. The curb trip tests resulted in peak angular velocities of 260 deg/sec and 300 deg/sec. The peak angular velocities in the soil trip tests were similar with values of 230 deg/sec and 390 deg/sec. The peak angular velocity of the vehicle in the dolly test was 460 deg/sec, which was much higher than the Test no. Vehicle model Trip method Test speed (km/h) Results 1 1981 Dodge Challenger Curb 48.1 no rollover 2 1981 Dodge Challenger Curb 47.6 rollover 3 1979 Datsun B210 Curb 47.2 rollover 4 1972 Chevrolet C20 Van Curb 47.6 no rollover 5 1981 Chevrolet Impala Curb 48.6 no rollover 6 1981 Dodge Challenger Dolly 48.6 rollover 7 1981 Dodge Challenger Soil 54.2 rollover 8 1979 Datsun B210 Soil 43.5 rollover TABLE 3 Test matrix for Cooperrider et al. study (34)

curb-tripped and the soil-tripped vehicles. The higher roll rate of the dolly-rolled vehicle was attributed to the 48-degree initial roll angle of the dolly when it contacted the ground. This caused a greater moment arm from the point of impact to the center of gravity of the vehicle. The analytical model developed in the study was based on the assumption that a constant tripping force acts on the vehi- cle during the rollover initiation phase. Although the model did not account for the effects of tire and suspension system compliance, the results compared well with the test data. It was found that the kinematics of the tripped vehicle varied significantly, depending on the tripping mechanism (i.e., curb, soil and dolly). Curb impacts produced very high decelerations, usually in excess of 10 Gs. Some curb-tripped vehicles, however, did not rollover because critical structural components (e.g., the wheel assembly) failed during impact, providing an alternate path for the unbalanced forces. When components of a vehicle collapse or break during these types of impact, the duration force may not be sufficient to initiate a rollover. Allen et al., 1991 (35) Researchers at STI conducted a study to determine the directional and rollover stability of a wide range of vehi- cles using the computer simulation program VDANL. They showed that rollover stability and directional stability are related to center of gravity location and track width, as well as the other characteristics that influence these variables under hard maneuvering conditions. Vehicle dynamics and tire-ground interaction under such conditions are nonlinear and can be quite complex; therefore, computer simulation is essential in analyzing stability problems. Forty-one vehicles were used in the study for parameter and field testing. Spinout occurs when rear tire adhesion lim- its are exceeded while the front tires still have side force capacity available. Computer simulation results were vali- dated with the field test results, and it was found that in many cases the dynamic behavior of the vehicle was largely depen- dent upon the tire model and tire-ground interaction. Thus, detailed information about the tire properties and friction coefficients are necessary for valid model development. One conclusion from their study was that load transfer dis- tribution among the tires should be near to, or greater than, the vehicle weight distribution, although there are several other factors that influence limit performance maneuvering. As the center of gravity of a vehicle is raised or the track width is narrowed, wheel lift off becomes more likely and balancing load transfer distribution becomes a critical issue. The computer simulation program, VDANL, was validated for both stable and unstable vehicle maneuvering conditions and was considered to be a practical and effective means of analyzing vehicle stability problems. 20 Allen et al., 1997 (36) Researchers at STI and JPC Engineering further improved the Slip Tire Model (STIREMOD) for use in the vehicle dynamics computer simulation program, VDANL. STIRE- MOD was expanded to include the full-range of operating con- ditions for both on- and off-road surfaces, including unlevel terrain, changing surface conditions, and tires plowing through soil. They discussed in some detail the input parameters for the model and the means for establishing typical model parameters. The model would be useful for the analysis of vehicle encroachments onto the road shoulders and sideslopes. The model could also be used for analyzing vehi- cle tire interaction with curbs, where the curb would be modeled as an abrupt change in surface shape and surface properties (e.g., asphalt pavement to a concrete curb). Allen et al., 2000 (37 ) Allen and other researchers at STI wrote a paper summa- rizing the development and application of the vehicle dynam- ics computer simulation model, VDANL. The subsystem models of VDANL are described (e.g., tires/wheels, brakes, steering, power train, roadway inputs, driver model, steering control, and speed control). Discontinuities in the roadway, such as potholes, speed bumps, and curbs, can be modeled in VDANL with additional inputs to the surface profile. VDANL models the inertial component of the vehicle as a six-degree-of-freedom sprung mass connected by springs and dampers to the axles, which are supported by pneumatic tires. According to Allen et al., “Communications services have also been added to VDANL so that it can provide commands for display image generators, feel and motion systems, sound cuing, and miscellaneous controls and displays”(37). The program runs in real time on Pentium-class computers run- ning Windows 95/98/NT network. A specialized version of the software was developed for the FHWA as part of the IHSDM, which allows new road- way designs to be assessed using a driver model. Two case studies were presented in their study using VDANL–IHSDM to determine (1) if a truck-climbing lane was necessary for a proposed roadway alignment and (2) if a loaded tractor- trailer would be able to maintain a specified speed traveling downgrade on the roadway without losing control. SYNTHESIS OF LITERATURE REVIEW Both sloping and vertical curbs are regularly used in urban areas along low-speed roadways for drainage purposes, walk- way edge support, pavement edge delineation, to discourage vehicles from leaving the roadway, and to provide limited redirection of encroaching vehicles. Vertical curbs have a vertical or nearly vertical face and are recommended for use only on low-speed roads. Sloping curbs have a sloping face

and are configured such that a vehicle can ride up and over the curb, in order to reduce the likelihood of causing tire blowout or suspension damage. Sloping curbs are used pri- marily for drainage purposes, but are also used on median islands and along shoulders of high-speed roadways for delin- eation and other reasons. Curbs along low-speed roadways are not likely to result in serious injuries and are commonly used in urban areas where speed limits are in the range of 40 to 48 km/h. Curbs along high-speed roadways have been discouraged by AASHTO for many years because of the potential hazard caused by high-speed impact with curbs (1). In the intermediate range of speed (between 60 and 80 km/h), however, there are no standards for the use of curbs. Highway engineers must, therefore, determine if a curb is warranted based on individ- ual roadway conditions and location. In urban areas, curbs are often considered acceptable; whereas in rural areas curbs are discouraged at intermediate speeds (1). There have been a limited number of studies performed to determine the effects of impact with curbs on the dynamic stability of vehicles and on the performance of barriers placed in combination with curbs. The studies have involved full- scale crash testing (22, 24–30, 32) and computer simulation using the HVOSM (21–23). A summary of full-scale crash tests involving curb–guardrail combinations is presented in Table 4. Although it has been found that sloping curbs do not significantly redirect a vehicle during tracking impact, they do affect the vertical trajectory of the vehicle. Thus, while the curb itself presents very little threat of harm when hit by a vehicle, when a vehicle impacts and mounts a curb, the 21 dynamics of the vehicle may cause the vehicle to impact a secondary object in such a manner that will cause the object to not function properly. A curb located in front of a guardrail may cause an impact- ing vehicle to strike the guardrail at a point higher or lower than normal. Under certain impact conditions, the curb can cause the vehicle to ramp high enough to vault over the bar- rier, or, in some cases, underride and snag on the barrier (22, 24, 25, 29, 32). Another example of possible adverse effects of a curb on the performance of a device is the placement of a curb in front of a breakaway pole. The breakaway features at the bases of the poles are designed to work when the pole is struck near the base. If a vehicle is airborne when it hits a breakaway pole, the impact point may be well above the base; thus the breakaway feature may not work as it is intended. In some studies, the lateral displacement of the vehicle at maximum rise height has been considered an important fac- tor for determining the potential for vehicle underriding or vaulting a barrier (22, 24, 38). Design parameters defined by AASHTO for curb impacts are shown in Figure 16. It was reported that underride and vaulting of a standard strong-post guardrail were possible when the barrier was placed within some critical range behind the curb, usually within 0.76 m for underride and between 0.01 and 3.66 m for vaulting. These data were obtained through measuring vehicle trajectory dur- ing impact with curbs. It was assumed for many years by design engineers that if the curb is placed behind the face of the W-beam that the curb-guardrail system would perform adequately in safely con- taining and redirecting an impacting vehicle. Previous crash Literature reference Testing agency Test no. Vehicle type Speed and angle Curb type Guardrail type Result Comment Bryden and Phillips (26) NYDOT Dodge Station Wagon (2041 kg) 100 km/hr 26 degrees 152-mm vertical curb Thrie-Beam Bridge Rail Passed smoothly redirected 1862-1-88 3/4-ton Pickup Truck (2449 kg) 100 km/hr 20 degrees 203-mm AASHTO A G4(1S) Failed vehicle vaulted over rail 1862-4-89 Small Car (820 kg) 100 km/hr 20 degrees 152-mm Asphalt Dike G4(1S) Passed smoothly redirected smoothly redirected 1862-5-89 Large Car Sedan (2041 kg) 100 km/hr 25 degrees 152-mm Asphalt Dike G4(1S) Failed vehicle vaulted over rail 1862-12- 90 Large Car Sedan (2449 kg) 100 km/hr 25 degrees 100-mm AASHTO G G4(1S) Passed vehicle was airborne but did not vault 1862-13- 91 Large Car Sedan (2041 kg) 100 km/hr 25 degrees 152-mm Asphalt Dike G4(1S) stiffened with W-beam Passed FHWA Memorandum Feb 1992 (27) ENSCO 1862-14- 91 Large Car Sedan (2041 kg) 100 km/hr 25 degrees 152-mm Asphalt Dike G4(1S) stiffened with rub rail Failed vehicle speed change at redirection was too high Holloway & Rossen (28) MwRSF M06C-1 1985 Ford LTD (2041 kg) 96.1 km/hr 25.1 degrees 152-mm vertical curb G4(1S) Passed smoothly redirected Polivka et al. (29) MwRSF NEC-1 1991 GMC 3/4-ton Pickup (2,000 kg) 103.2 km/hr 24.5 degrees 102-mm AASHTO G G4(1S)-mod with wood blockout Failed excessive anchor movement / guardrail ruptured Bullard and Menges (30) TTI 404201-1 1995 Chevrolet 3/4-ton Pickup (2000 kg) 101.8 km/hr 25.2 degrees 100-mm CDOT curb G4(2W) Passed significant guardrail damage and anchor movement Polivka et al. (32) MwRSF NEC-2 1994 GMC 3/4-ton Pickup (2,000 kg) 100.3 km/hr 28.6 degrees 102-mm AASHTO G G4(1S)-mod with wood blockout nested W-beam Passed vehicle experienced extreme trajectory but did not vault over rail TABLE 4 Summary of full-scale crash tests of curb–guardrail combinations with curb located behind face of guardrail

tests, involving large sedans and pickup trucks impacting var- ious curb-guardrail combinations, have provided researchers with mixed results regarding the performance of such sys- tems (24, 25, 27–30, 32). In full-scale crash tests performed by ENSCO with full- size cars, it was shown that vaulting is possible even when the curb is located flush with the face of a W-beam guardrail. If guardrail deflections during impact are sufficient to allow the wheel of the vehicle to contact and mount the curb, the vehicle may vault over the barrier (28). Even though the vehicle contacts the barrier prior to reaching the critical tra- jectory height that would signify override, the vehicle will continue to rise while it is in contact with the barrier and may result in vaulting during redirection. Crash tests with pickup trucks performed at Midwest Roadside Safety Facility, on the other hand, have demonstrated that similar curb/W-beam guardrail combinations do not degrade the performance of the barrier systems (23, 28). 22 Some curb types are more likely to cause vaulting of a vehicle than others. The FHWA memorandum in February 1992 (27) reported that, in the case of curbs 150 mm high or higher, if a guardrail deflects enough for the wheels to mount the curb, the vehicle could vault over the guardrail. It was also reported in the FHWA memorandum that crash tests involving the AASHTO Type G curb (a 100-mm curb height with slanted face) placed behind the face of the W-beam resulted in the vehicle becoming airborne when guardrail deflection permitted the wheels to mount the curb; however, the vehicle did not vault the guardrail. A similar conclusion was found in other studies, which showed that vehicle impact with low curbs would result in very little change in the ver- tical trajectory of the vehicle (50-mm maximum), regardless of the vehicle’s speed and angle of impact (22, 24). A W-beam guardrail is sufficiently stiff that the lateral deflections of the barrier are minimal during impact with a small car; thus for curb–guardrail combinations in which the Figure 16. Design parameters for curb impacts as defined by AASHTO (38).

curb is placed underneath a strong-post W-beam guardrail, there is little chance of vehicle contact with the curb (24, 27). It has also been found that stiffening the guardrail system by installing a W-beam rail to the back of the posts or installing a rub-rail will enhance the safety performance of a curb– guardrail system (28). The installation of a rub-rail may pro- vide the most safety benefit, since it both stiffens the system to avoid vehicle-to-curb contact and shields the posts from potential wheel snag. There have been three tests performed on curb–guardrail systems under NCHRP Report 350 Test 3-11 impact condi- tions: MwRSF tests NEC-1, NEC-2 and TTI test 404201-1 (29, 30, 32). These tests involved 100-mm-high curbs placed in combination with strong-post guardrails. Both test NEC-1 and test TTI 404201-1 resulted in significant tensile forces in the W-beam rail and excessive movement of the anchor sys- tem. In test NEC-1, the two upstream anchor posts for the G4(1S) guardrail with wood blockouts ruptured causing the vehicle to pocket (29). This ultimately resulted in rupture of the W-beam rail element, and the vehicle penetrated the guard- rail. The poor performance of this system was not directly attributed to the effects of the curb, but rather to a loss of ten- sile capacity of the guardrail during impact when the anchor system failed. In TTI test 404201-1, the foundation of the anchor posts of the G4(2W) guardrail moved in excess of 70 mm at the ground line, and there was considerable damage to the guard- rail system; however the system did meet all safety require- ments of NCHRP Report 350 (30). Also, the extent of dam- age to the system in test TTI 404201-1 was much greater than that of previous crash tests on the G4(2W) guardrail system without a curb present (31). In test NEC-2, the G4(1S) guardrail with wood blockouts was modified and retested (32). The guardrail was modified by nesting 12-gauge W-beam rails along the length of the sys- tem. This test resulted in excessive vertical trajectory of the vehicle during impact, but the vehicle remained upright and successfully met all safety criteria of NCHRP Report 350. Vehicle tripping on curbs was addressed in a very limited number of studies. The studies that were identified in the lit- erature used a variety of techniques for analysis including analytical methods, computer simulation, full-scale crash test- ing, and accident data analysis (24, 33, 34). Vehicle tripping on curbs was addressed in Holloway et al. using HVOSM to simulate nontracking impacts of large passenger sedans (24). Based on the results of their simulations, they concluded that sloping curbs may not be a significant cause of vehicle roll- overs; however, it should be noted that the models used in their study were not validated for nontracking impacts. It was not reported whether or not friction between the tires and ground surface was included in the simulations. Fric- tion between the tires and ground will affect the initial roll angle and roll rate of the vehicle prior to impact, which may increase the vehicle’s tendency to rollover. 23 DeLeys and Brinkman used crash data analysis and com- puter simulation to investigate rollover tendencies of vehi- cles traversing various kinds of roadside terrain. They con- cluded that the data bases lacked the comprehensive and detailed information necessary to define conditions that lead to rollover. A modified version of HVOSM with improved application for rollover situations was used in their study (33). Full-scale tests were used to validate the computer models and, subsequently, over 200 simulations were con- ducted to investigate the rollover tendencies of vehicles tra- versing various sideslopes, fill embankments, and ditch con- figurations. They did not investigate vehicle-curb interaction; however, the models that were used in their study may have been applicable for such analysis. Cooperrider et al. carried out a series of full-scale crash tests to determine the potential for rollover of various vehicle types tripped by a curb, sliding in soil, and rolled off a dolly (34). A steel 152-mm-square tube section rigidly affixed to the roadway was used to represent a curb in their tests. In five of the eight tests that they conducted, the vehicles rolled over. In the cases where rollover did not occur, the wheel assembly failed during impact with the curb due to the high forces that were developed. The failure of the wheel assem- bly, consequently, removed the overturning force that was being applied to the vehicle. If the wheel assembly had not failed in those cases, it is possible that all the tests would have resulted in a rollover. The vehicle dynamics code, VDANL, has been used to study vehicle rollover as a function of unstable maneuvering conditions and also to investigate vehicle rollover because of impact with various vehicle tripping mechanisms such as curbs, soil, ditches, and so forth (35–37). The results of the computer models developed in those studies were validated with full-scale tests. VDANL was chosen by the FHWA to be incorporated into the IHSDM, which is used to assess new highway designs. SUMMARY While there has been some work performed on the safety effectiveness of curbs and the use of curbs in conjunction with traffic barriers, the literature review shows that there are many limitations, such as the age of the tests, the lack of sophistication in early computer models, and changing full- scale crash testing guidelines. The following are the major findings of the literature review: • Curbs should not be used in combination with W-beam guardrail systems on high-speed roadways due to the potential safety hazard of vaulting or underriding the barrier. In cases where design engineers include curbs along high-speed roadways for drainage reasons or to improve delineation, other methods should be sought to achieve those purposes.

• Neither the large and small cars crossing 150-mm-high or smaller curbs in a tracking manner are likely to result in loss of vehicle control or cause serious injuries. The response of the 2000-kg pickup truck crossing curbs, however, was not known. The large passenger car used in the previous crash testing procedures was replaced in the current testing procedures (NCHRP Report 350) with the 2000-kg pickup truck. The dynamic response of this particular vehicle type crossing over curbs (not in conjunction with a roadside safety barrier) has never been evaluated with either full-scale tests or computer simulation. • Most of the curb impacts that were found in the literature involved vehicles encroaching the curb in a tracking manner. It was concluded in every case that a vehicle encroaching onto a sloping curb in a tracking manner is not likely to cause the driver to lose control of the vehi- cle or cause the vehicle to become unstable unless a sec- ondary impact occurs. Another aspect of collisions with curbs involves an out-of-control vehicle impacting the curb in a nontracking position. In these situations, vehi- cle tripping may be highly probable during impact. • Errant vehicles leave the roadway in a variety of orien- tations; however, it is assumed that the majority of these vehicles encroach onto the roadside in a semicontrolled tracking manner. In such cases, the left or right front bumper would be the first point of contact with a road- side object in an impact event. The position of the bumper upon impact has, therefore, been a primary concern involving impacts with longitudinal traffic barriers, where it has been assumed that the position of the bumper dur- ing impact is a reasonable indicator of vehicle vaulting or underriding the barrier. Due to pitching of the mov- ing vehicle, the bumper height at impact may be higher or lower than the static position of the bumper. 24 • Nontracking impacts with curbs may result in vehicle instability and rollover, especially impacts involving vehicles with high centers of gravity. From the literature study it seems that the most likely methods for analyz- ing nontracking impacts will be vehicle dynamics codes, such as VDANL. There has been a great deal of advance- ment in computation power and in code development over the past few years that has enabled computer sim- ulation programs to become a very efficient means of analysis. Both tracking and nontracking impact on curbs may be investigated using vehicle dynamics codes, such as VDANL, and finite element analysis (FEA) using LS-DYNA. • A small number of tests have been performed in which a curb was placed behind the face of guardrail barriers. The idea was to locate the curb such that minimal inter- action between the vehicle and curb occurred. This worked well with lighter vehicles, such as the 820-kg small car, but did not prevent vehicle-curb interaction with the heavier vehicles, such as the 2000-kg pickup truck, unless the guardrail was retrofit in some man- ner to strengthen it and minimize guardrail deflection. To circumvent the problem, one option considered was to use a low-profile curb underneath the guardrail. This was expected to minimize the effects that the curb would have on vehicle trajectory when the wheels of the vehicle were able to contact the curb during impact; however, full-scale tests conducted by various organizations provided mixed results. In some cases the crash test was successful, while in others it was not. In cases where the test was a failure, it was not clear whether the failure was induced by vehicle-curb inter- action or if it was simply caused by inadequate barrier performance.

Next: Chapter 3 - Summary of State Surveys on Curbs and Curb Barrier Combinations »
Recommended Guidelines for Curb and Curb-Barrier Installations Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 537: Recommended Guidelines for Curb and Curb–Barrier Installations presents the findings of a research project to develop guidelines for the use of curbs and curb–guardrail combinations on high-speed roadways. The report includes recommendations concerning the location of curbs with respect to the guardrail for various operating speeds.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!