Performance of Longitudinal Barriers on Curved, Superelevated Roadway Sections (2019)

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.

18 Literature Review and State DOT Survey 2.1 Introduction A detailed literature review was conducted to gather infor­ mation and synthesize relevant past efforts. The focus was on studies related to the design, performance evaluation, mainte­ nance, and application details of longitudinal barrier systems when placed on CSRS with emphasis on crash testing and computer simulations. TRID was used to identify domestic and international reference materials. The findings from the lit­ erature search revealed limited references related to this topic. Summaries of relevant knowledge from domestic and inter­ national reference materials are presented in this chapter. 2.2 Barrier Crashworthiness Research One of the first research efforts to determine the perfor­ mance of traffic barriers on curved alignments was described in Bridge Rail Retrofit for Curved Structures (Bronstad and Kimball 1986). Three bridge rail systems were installed on a curved, superelevated structure and crash tested. The railings tested were two 813­mm­ (32­in.­) tall NJ concrete barriers (one installed true vertical and the other perpendicular to the superelevated bridge deck) and a retrofitted tubular Thrie­ beam rail (i.e., double­sided Thrie­beam rail), installed per­ pendicular to the deck (Figure 2.1). Test vehicles included an 820­kg (1,800­lb) sub­compact car (Honda Civic), a 1,020­kg (2,250­lb) compact car (Vega), and a 9,070­kg (20,000­lb) school bus. Impact conditions for all tests were nominally 64 km/h (40 mph) with a 15° impact angle. The simulated bridge deck used in the testing was constructed with a 48.8­m (160­ft) outside radius on a 4.5% downgrade. The super elevation was 12% with no shoulder break. All three barriers contained and redirected the vehicles in the crash tests. There was not a significant difference in per­ formance between the two concrete barrier orientations (i.e., true vertical or perpendicular to the superelevated roadway surface), but vehicle climb was reduced in the perpendicular orientation tests. In all of the concrete barrier tests, the cars climbed at least 460 mm (18 in.) up the barrier. In one test of the vertically oriented barrier, the vehicle nearly climbed to the top of the barrier, even at the relatively low impact speed and angle. The tubular Thrie­beam retrofit design, installed perpendicular to the roadway surface, performed better than the safety shape in all the crashes. The second and most relevant reference was an FHWA report entitled Traffic Barriers on Curves, Curbs, and Slopes dated August 1993 (Stout et al. 1993). This study investigated performance of guardrails on curves, on slopes, and with curbs. The research was conducted in four main phases. In the first phase, previous references on related topics were reviewed. The second phase consisted of analyzing crash datasets to identify issues related to the study topic. Next, a series of crash tests were conducted on guardrails to assess their performance when installed on curves, on slopes, and with curbs. In the final phase, an attempt was made to use computer simulation using the Numerical Analysis of Roadside Design (NARD) program to assess the barrier performance under these conditions. The literature review revealed that the W­beam and Thrie­ beam barriers did not meet the NCHRP Report 230 testing requirements when installed on non­level terrain at the tested offset distances from the edge of the road, while the cable bar­ rier system did meet these requirements. Another finding indicated a similar observation that different lateral barrier offsets and heights could lead to vehicle override and under­ ride in sloped median installations. A third finding was that vehicle behavior is affected by highway features and roadside barriers. A final finding indicated that roadside slopes signifi­ cantly affect barrier performance. The data analysis revealed limited information due to the small size of some of the datasets available. Some of the key findings from the crash data analysis are listed as follows: • Pertaining to curved roadside sections, the accident data showed no evidence that guardrail performance is worse on curved road sections than on straight sections. C H A P T E R 2

19 • For installations on side slopes, the performance of barriers placed behind the hinge point was significantly worse than when placed before the hinge point. This was the case even though all barriers were installed in relatively gentle slopes (shallower than 4H:1V) as recommended by AASHTO guidelines for guardrail installations. In the third phase of the study, crash tests were conducted. Two tests involved impacts with an 820­kg (1,800­lb) small car and 2,450­kg (5,400­lb) pickup truck, both at 96.5 km/h (60 mph) and 20° impact angle into a standard W­beam guardrail on a 363­m (1,192­ft) radius curve with level ter­ rain. These tests indicated that the barrier would meet the crashworthiness requirements. Four additional tests were conducted with the 2,450­kg (5,400­lb) pickup at 96.5 km/h (60 mph) and 20° impact angle, but approaching the barrier on the diagonal of a 10% superelevation upslope. Four differ­ ent barrier and placement conditions were tested as noted in Table 2.1. In all cases for the standard guardrail at normal heights the outcome was negative. Only the high­performance Thrie­beam barrier met the requirements for these crash con­ ditions. The report did not cite specific issues with the vehicle­ to­barrier interface that might be a focal point for barrier redesign on curves. While these tests provided some use­ ful insights, they only considered a curve radius of 363 m (1,192 ft), superelevation slope of 10%, speed of 96.5 km/h (60 mph), impact angle of 20°, and a 2,450­kg (5,400­lb) pickup truck. There is a need to consider a broader set of (a) 32-in. NJ bridge rail—true vertical (b) 32-in. NJ bridge rail—normal to deck (c) Tubular Thrie-beam bridge rail—normal to deck Figure 2.1. Cross sections of tested bridge rails (Bronstad and Kimball 1986).

20 impact conditions consistent with updated crashworthi­ ness criteria. 2.3 Terrain Effect Studies A more recent study, not directly related but relevant to the topic of this research, was conducted at the Texas Transportation Institute (TTI) (Sheikh and Bligh 2006). Using FE simulations, this study investigated the safety performance of 813­mm­ (32­in.­) high F­shape concrete barriers when installed on sloped medians. The simulations included different median and barrier placement configurations. In the first configuration, the study focused on barriers installed at the center of symmetric V­shaped medians with side slopes of 6:1 or shallower (Fig­ ure 2.2). Horizontal curvatures were not considered in this study. To identify the most critical impact scenarios for this con­ figuration, simulations without a barrier were conducted to determine the trajectory of the corner of the front bumper relative to the ground as the vehicle traverses the median. This trajectory is shown in Figure 2.3. Two critical barrier place­ ments (i.e., median widths) were identified: (1) when the vehicle is at its highest point relative to the barrier and (2) when the vehicle is at its lowest point relative to the barrier. The first point was found to be 4 m (13.25 ft) from the edge of the road and the bumper point was about 150 mm (6 in.) higher than it would be when the barrier is installed on flat terrain. The second point was identified to be 7.2 m (23.5 ft) from the edge of the road, and the bumper was about 50 mm (2 in.) lower than the flat terrain condition. Simulations using a 2000P vehicle (represented by a Chev­ rolet C2500 pickup model) traveling at 100­km/h (62.2­mph) initial speed and 25° impact angle were conducted to assess the barrier performance in these two critical impact configu­ rations. The barrier in these simulations was assumed rigid because it was not expected that it would undergo significant deformation or damage during the impact. Both simulations showed that the barrier met all NCHRP Report 350 criteria (Ross et al. 1993). Two additional configurations were investigated where the barrier was placed on one side of the median. In the first config­ uration, the barrier was placed on the shoulder [Figure 2.4(a)]. In the second configuration, the barrier was placed at the edge of the shoulder as shown in Figure 2.4(b). In the latter case, one side of the median was regraded to accommodate the placement of the 0.61­m (2­ft) base of the barrier. A total median width of 9.1 m (30 ft) [2.2 m (40 ft) (with 1.22­m (4­ft) shoulders] was used in the second case. In both cases, the barrier’s vertical alignment was perpendicular to the road surface. The height of the bumper corner point relative to ground level as the vehicle crosses the median is shown in Figure 2.5. The bumper impact height for the first case was 200 mm (8 in.) lower than it would be on flat terrain and about 75 mm (3 in.) higher than it would be for flat terrain installations for the second case. Simulations with the 2000P vehicle traveling at a 100­km/h (62­mph) initial speed and an impact angle of 25° were conducted. In these simulations, the vehicle impacted the back side of the barrier after crossing the symmetric V­shaped 6:1 sloped median. The simulations Test Barrier Placement Outcome 1862-6-89 Standard W-beam guardrail w/ 1.83- m (6-ft) posts Beyond 3-m (10-ft) shoulder The vehicle was redirected on the traffic side of the barrier, but rolled over. 1862-9-90 Standard W-beam guardrail w/ 2.13- m (7-ft) posts Beyond 3-m (10-ft) shoulder The vehicle vaulted the rail and rolled over. The lateral torsion in the longer posts increased buckling. 1862-10-90 Thrie-beam guardrail Beyond 3-m (10-ft) shoulder The vehicle was redirected by this high-performance barrier. 1862-16-91 Standard W-beam guardrail w/ 1.83- m (6-ft) posts At edge of traveled way This option was intended to eliminate the possibility that the vehicle would become airborne at the break point of the superelevated section and shoulder, but the vehicle still vaulted and rolled. Table 2.1. Full-scale crash tests conducted on 10% superelevation. 2% roadway cross- slope 20:1 cross-slope 6:1 cross-slope 6-ft shoulder Figure 2.2. Barrier placed in the middle of a 6:1 or shallower sloped median (Sheikh and Bligh 2006).

21 Case 1 Barrier at 13.25 ft Case 2 Barrier at 23.5 ft 34.00 32.00 30.00 28.00 26.00 24.00 0.00 5.00 10.00 15.00 20.00 25.00 N o m in al b u m p er h ei g h t (i n ) Lateral position of the bumper (ft) Figure 2.3. Bumper height relative to the ground as vehicle crosses the median (Sheikh and Bligh 2006). (a) Barrier placed on shoulder (b) Barrier placed at edge of shoulder Figure 2.4. Barrier placed at the edge of a 6H:1V or shallower sloped median (Sheikh and Bligh 2006).

22 showed that the barrier met all NCHRP Report 350 criteria in both cases. It is noted from these simulations that the 813­mm (320­in.) F­shape concrete barrier performed adequately, even when the height of the vehicle relative to the barrier was 150 mm (6 in.) higher (Figure 2.3) and 200 mm (8 in.) lower (Figure 2.5) than that of the flat terrain case. Two full­scale crash tests were conducted to validate the simulation results. The tests were set up in similar configura­ tions to the first two simulation cases (representing barrier placement at the center of V­shaped symmetric medians). Both tests met the NCHRP Report 350 criteria. Based on the simulations and tests, guidelines for the use of concrete barriers on sloped medians were recommended for the Texas DOT (TxDOT) as follows: “The TxDOT cast­in­place permanent F­shape barrier and the precast free­standing F­shape barrier are considered suitable for placement on roadside and median fore­slopes of 6H:1V or less. Additionally, these barriers are suitable to be placed at any lateral offset of the barrier from the roadway edge and for any width of depressed V­ditch median as long as the barrier is placed at its center. Similar or better per­ formance would be expected for placements on more gentle (e.g., 8H:1V) slopes.” Testing was not conducted for the cases where the barrier was installed on the shoulder and no recom­ mendations for these cases were included. In another study conducted at TTI, the effects of barrier vertical orientation (inclination) on the performance of safety shape bridge rail parapets were investigated (Sheikh and Alberson 2005). Simulations with F­shape parapets installed on five different roadway cross slopes were con­ ducted and compared to study the effects of barrier vertical orientation. The cross slopes studied are shown in Table 2.2. For all cross section profiles, the parapet was modeled plumb to the earth. The simulations were conducted using a 2000P vehicle model impacting the barrier at 100 km/h (62.2 mph) and a 25° angle. Using the simulation results, the effects of barrier orien­ tation on the vehicle roll and pitch angles and vehicle center of gravity (CG) vertical displacement (representing vehicle climb) were assessed. Figure 2.6 shows plots of these mea­ sures for different roadway cross slopes. The simulations showed that the vehicle roll, pitch, and vertical displacement increase with increased cross slope angles (i.e., increased inclination). N o m in al b u m p er h ei g h t (i n .) Lateral position of the bumper (ft) Case 3 Case 4 Figure 2.5. Bumper height relative to the ground as vehicle crosses the median (Sheikh and Bligh 2006). Table 2.2. Cross slopes used in the study (Sheikh and Alberson 2005).

23 2.4 General Curve Safety Guidance Another reference related to the topic of this research is NCHRP Report 500: Guidance for Implementation of the AASHTO Strategic Highway Safety Plan, Volume 7: A Guide for Reducing Collisions on Horizontal Curves (Torbic et al. 2004). This report includes guidelines developed to reduce fatal and serious injuries on curved roads. The guidelines are aimed at reducing the likelihood of a vehicle leaving the road and minimizing the adverse consequences of run­off­ road situations at horizontal curves. The guidelines are listed below. The first 15 guidelines are aimed at reducing the num­ ber of vehicles leaving the road, while the last 5 are intended to reduce the severity of crashes on curved roads. The last two guidelines specifically address roadside safety hardware; limited information was available in the report on their implementation: 1. Provide advance warning of unexpected changes in hori­ zontal alignment 2. Enhance delineation along the curve 3. Provide adequate sight distance 4. Install shoulder rumble strips 5. Install centerline rumble strips 6. Prevent edge drop­offs 7. Provide skid­resistant pavement surfaces 8. Provide grooved pavement 9. Provide lighting of the curve 10. Provide dynamic curve warning system 11. Widen the roadway 12. Improve or restore superelevation 13. Modify horizontal alignment 14. Install automated anti­icing systems 15. Prohibit/restrict trucks with very long semitrailers on roads with horizontal curves that cannot accommodate truck off tracking 16. Design safer slopes and ditches to prevent rollovers 17. Remove/relocate objects in hazardous locations 18. Delineate roadside objects 19. Add or improve roadside hardware 20. Improve design and application of barrier and attenuation systems Figure 2.6. Effects of roadway cross slopes on vehicle roll, pitch, and displacement (Sheikh and Alberson 2005). (a) Roll angle versus inclination (b) Pitch angle versus inclination (c) CG vertical displacement versus inclination

24 2.5 Road Design Guidelines A few guidelines were obtained from respondents of the state DOT survey. An initial review showed that these guide­ lines are similar to the Green Book recommendations sum­ marized in the following subsection. Guidelines pertaining to curved road sections address only geometric aspects of the road. No specific information was found related to lon­ gitudinal barrier installations. Few differences were observed between the state DOT survey responses and Green Book guidelines for maximum superelevation rates, side friction factors, superelevation design tables, and so forth. The fundamentals for barrier design and deployment on U.S. highways is provided in the Green Book and the Roadside Design Guide. These documents were reviewed to understand the prevailing rationale and determine the conditions and parameters cited for when barriers are needed, the recom­ mended types, and where and how they are to be deployed. The focus was on longitudinal barrier installations on high­ speed CSRS. The relevant elements from these documents are summarized below. Because states can establish their own standards and practices, a state DOT survey was also conducted. 2.5.1 Green Book To establish a safe and comfortable driving environment on curved road sections, guidelines for the selection of road curvatures and superelevations, given a selected design speed, are provided in the Green Book. The guidelines that are rel­ evant to this research include design speed; maximum super­ elevation rate; side friction factor; minimum curve radius; superelevation distribution methods and superelevation cal­ culation; and shoulders. These guidelines are used by most states when determining the curvatures and superelevation rates of curved roads. These guidelines were also used in this effort to create the road profiles for the computer simula­ tions. These guidelines are summarized in the following subsections. 2.5.1.1 Design Speed The Green Book defines design speed as “a selected speed used to determine the various geometric features of the roadway.” The design speed should be selected based on the topography, anticipated operating speed, adjacent land use, and functional classification of the highway. The design speed affects many aspects of the roadway geometric elements. It directly influences superelevation and curvatures as well as several other design parameters. The Green Book recommen­ dation for minimum design speed on high­speed roadways (freeways) is 80 km/h (50 mph). The use of design speeds of 100 km/h (60 mph) or higher is encouraged for urban freeways, because it can be achieved with minimal additional costs. A 110 km/h (70 mph) design speed is recommended for rural freeways and interchange locations consistent with higher design speeds. For mountainous terrain, a design speed of 80 to 100 km/h (50 to 60 mph) is recommended. The Green Book gives an approximation of the running speed as a function of design speed as shown in Table 2.3. Table 2.3. Average running speed versus design speed (AASHTO 2011a).

25 2.5.1.2 Maximum Superelevation Rate Vehicles traveling on curved roads are subjected to a lat­ eral force known as centrifugal force. This lateral force, which pushes the vehicle outward from the curved road center point, increases as the vehicle speed increases or the radius of the curve decreases. Superelevation is the sloping (banking) of the road to oppose this lateral centrifugal force. For high­speed roadways, the Green Book recommended range for maximum superelevation rate is 6% to 12%. This range is reduced to 6% to 8% in regions where snow or ice is of concern, because vehicles traveling at low speeds in snowy or icy conditions tend to slide on roads with high superelevations. A superelevation rate of 6% to 8% is also recommended for viaducts; the lower 6% superelevation rate is recommended when freezing and thawing conditions are likely. 2.5.1.3 Side Friction Factor The side friction factor is defined in the Green Book as the “vehicle need for side friction.” When a vehicle is traveling on a curved road, the lateral centrifugal force is resisted by a combination of the superelevation and the friction between the tires and the road surface. For a given vehicle speed and curvature radius, an increase in superelevation would lead to lower lateral friction force (i.e., a larger portion of the cen­ trifugal force is resisted by the superelevation). The side fric­ tion factor is the ratio between this lateral friction force and the weight of the vehicle (with a small conservative simpli­ fication). The side friction factor f is expressed in the Green Book as follows: 15 0.01 2 f V R e= − where V is the vehicle speed (mph), R is curve radius (ft), and e is the superelevation rate (%). The side friction factor depends on many variables such as the speed of the vehicle; weight; braking and accelerating; suspension; and tire design and condition. It decreases when the speed of the vehicle increases, and during braking or acceleration. The maximum value of the side friction factor is reached when the vehicle starts to skid. To avoid skidding, the maximum side friction factor used in highway design is much less than the actual value (i.e., roads are designed based on a portion of the maximum side friction available to ensure safety and comfort of the driver). Based on several research studies, the Green Book defines a curve for maximum side friction factor versus design speed. This curve is shown in Figure 2.7. Figure 2.7. Side friction factor assumed for road design (AASHTO 2011a).

26 2.5.1.4 Minimum Curvature Radius The minimum curvature radius defines the sharpest curva­ ture for a given design speed, maximum superelevation, and maximum side friction factor. It can be expressed as follows: 15 0.01 max max min 2 R V e f D ( )= + where VD is the design speed (mph), fmax is maximum side friction factor, and emax is the superelevation rate (%). 2.5.1.5 Superelevation Distribution Methods and Superelevation Calculations For a given design speed and road curvature, several com­ binations of superelevation and side friction can be used to resist the lateral centrifugal force. The Green Book lists five different methods for the distribution of the superelevation and lateral friction forces. For high­speed roadways, the last method (Method 5) is recommended. In this method, the superelevation and side friction have a curvilinear relation­ ship with respect to the inverse of the curvature radius. Using the maximum side friction factor shown in Figure 2.7 and this method, diagrams of superelevation in relation to curvature radius and design speed for different maximum supereleva­ tions are generated. A sample diagram is shown in Figure 2.8. Using the same method, superelevation tables were generated. A sample superelevation design table is included in Table 2.4. These diagrams and tables are used to determine road curva­ tures and superelevation rates. 2.5.1.6 Shoulders The Green Book recommends that shoulders in heavily traveled high­speed highways be at least 3 m (10 ft) wide, with a 3.66­m (12­ft) width preferable. On four­lane freeways, the recommended shoulder width on the left side of the road is 1.22 m to 2.44 m (4 ft to 8 ft) and on the right side at least 3 m (10 ft). Asphalt and concrete shoulders should be sloped from 2% to 6%. Gravel or crushed rock shoulder slopes should be from 4% to 6% and turf shoulders should be 6% to 8%. 2.5.2 Roadside Design Guide A review of the Roadside Design Guide revealed no specific recommendations for longitudinal barrier installations on curved and superelevated road sections (i.e., barrier instal­ lations on curves follow the same guidelines as on straight roads). A few recommendations related to the research topic are listed as follows: • A barrier should not be installed on a slope steeper than 6H:1V unless it has been tested and found to meet the NCHRP Report 350 or MASH evaluation criteria. Figure 2.8. Sample design superelevation diagram, for emax = 6% (AASHTO 2011a).

27 • Only flexible and semi­rigid barriers should be installed on slopes steeper than 10H:1V. • A barrier should be placed as far as possible from the trav­ eled way as practical without hindering its proper operation and performance. Barrier offset distances (Shy­Line) range from 1.22 m (4 ft) for a 50­km/h (30­mph) design speed to 3.66 m (12 ft) for a 130­km/h (80­mph) design speed. 2.6 State DOT Survey Results A state DOT survey was conducted to identify common barriers used on CSRS and to gather information pertaining to specific state standards, guidelines, and practices for the design, installation, and construction for such situations. The survey instrument was designed such that the questions were kept to a minimum but covered needed information and identified where the states had their own standards, guidance, or practices. It deemed more efficient to pursue the details of standards, guidance, or practices only with the states that had them. Thus, there were only eight questions included in the survey. The questionnaire is included in Appendix A. The survey sought the following information: • Types of longitudinal barriers currently used or likely to be used in the future on high­speed CSRS. • The existence of specific criteria for which high­speed CSRS need barriers. • Existence of specific criteria for the type of barrier to be used. • Availability of data for crashes in such situations. • Availability of in­service barrier performance assessments and safety concerns. • References for the specific standards, guidance, and practices. Of the 50 states surveyed, 33 responded for a 68% response rate. The responses received are summarized as follows: Question 1. The participants were asked to provide infor­ mation on the types of longitudinal barriers that are currently in place or are being installed on high­speed CSRS in their state. They were also asked to rank these barriers based on their usage (most to least commonly used). The survey form Table 2.4. Sample design superelevation table, emax = 6% (AASHTO 2011a).

28 focused on W­beam, Thrie­beam, and concrete barriers of various heights. The information received was grouped and ranked as shown in Table 2.5. The rankings obtained from the participating states were used to establish a global ranking for all states. A first ranking factor was computed by summing all rankings listed by the states for each barrier and dividing it by the total number of states that use that particular bar­ rier. This factor is shown in the third column of Table 2.5 for “previously installed barriers” and the seventh column for “currently being installed barriers.” The smaller this ranking factor, the higher the barrier usage. Another ranking factor used is the total number of states that ranked the barrier as 1 (most commonly used). This factor is shown in columns 4 and 8 for previously installed barriers and currently being installed barriers, respectively. Higher numbers for this fac­ tor indicate higher usage. Based on these two factors, a global ranking was determined (columns 5 and 9 in Table 2.5) to identify the most commonly used longitudinal barriers. The W­beam guardrail, with a height less than 31 in., was ranked as the most commonly used longitudinal barrier on high­speed CSRS. This was the case for both previously installed and currently being installed barriers. All participants indicated that it is used in their state and 26 participants ranked it as the most commonly installed. For previously installed bar­ riers, the concrete barrier with a height less than 32 in. ranked second. This barrier is used by 29 of the 33 participating states and was ranked first (most commonly used) by 2 states. For currently being installed barriers, the W­beam guardrail with a height of 31 in. or higher ranked second. Eleven of the 33 par­ ticipating states indicated that they are currently installing this barrier and 5 participants ranked it as the barrier most often currently being installed. Concrete barrier with a height greater than 32 in. was ranked third for both previously installed and currently being installed barriers. Concrete barrier with a height less than 32 in. was ranked fourth for currently being installed barriers. These results were consistent with expec­ tations; however, because higher barriers would increase the likelihood of capturing unstable vehicles, it might have been expected that there would have been more use of higher varia­ tions of W­beam and concrete barriers on CSRS. An unexpected response was in the usage of cable barrier systems on high­speed CSRS. Even though this system was not one of the barriers listed in Question 1, it was added by 20 of the 33 states as previously and currently being installed in their state. Two of the participants indicated that cable barrier was ranked first for previously installed barriers in their state for CSRS, one mentioned it is for roadside appli­ cation only. One of the participants indicated that cable bar­ rier is currently being installed in their state and is ranked first among the barriers. This response may accurately reflect that cable barriers are increasingly being used in median and roadside applications, which may include CSRS. The focus of this research has been on the outer, roadside applications of barriers. These responses suggest that future analysis should focus on the median barriers in such situations. Question 2. The participants were asked to provide infor­ mation on the type of barriers that are likely to be the stan­ dard applications in high­speed CSRS in future installations. The information from the states was analyzed and grouped as Previously Installed Barriers Currently Being Installed Barriers # of States Using Barriera Ranking Factor 1b Ranking Factor 2c Global Ranking # of States Using Barriera Ranking Factor 1b Ranking Factor 2c Global Ranking W-beam barrier (<31-in. height) 31 1.35 26 1 26 1.77 21 1 Concrete barrier (<32-in. height) 27 2.48 2 2 22 2.82 1 4 Concrete barrier (>32-in. height) 24 3.08 0 3 25 2.80 0 3 Cable barrier system 20 3.70 2 4 20 3.30 1 5 Thrie-beam barrier 15 4.2 1 5 15 4.13 0 6 W-beam barrier (>31-in. height) 9 4.33 0 6 11 2.45 5 2 a Number of participating states (33 total) that indicated barrier used in high-speed CSRS. b Rating factor based on states’ barrier rankings (sum of rankings divided by number of states using barrier). c Rating factor based on number of states that ranked barrier as 1 (most commonly used). Note: The darker shaded areas indicate the most commonly used longitudinal barrier on CSRS; the lighter shaded areas indicate the second most commonly used barrier. Table 2.5. Longitudinal barriers usage on high-speed CSRS.

29 shown in Table 2.6. Concrete barriers were listed most often (by 25 of the 33 states). Participants also provided some additional details relative to the barriers’ cross sectional shape with single slope and the F­shape concrete barriers being listed by 10 and 7 participants, respectively. The NJ and double­faced concrete barriers were listed by three and two participants, respectively. The W­beam guardrail was listed by 15 of the 33 states. The cable barrier system was mentioned by 13 of the 33 states. The strong­post W­beam and the MGS were each mentioned by 8 of the 33 states. The Thrie­beam barrier was mentioned by 7 states. Three states listed the Box­beam barrier. Detailed descriptions of these barriers and web links to technical draw­ ings were included for most barriers listed by the respondents. Question 3. The participants were asked if their state has special criteria to determine whether a barrier is warranted on a high­speed CSRS. Eleven states indicated “yes” and 22 indicated “no,” suggesting that special criteria may only exist in a third of the states. Five states mentioned that they use the Roadside Design Guide for curve adjustment and for clear zone. Three other states indicated that they use their own state road design manual. Four states indicated that they are investigating increasing the clear zone and considering other alternatives based on the specific case. Three states men­ tioned that they evaluate each site separately and make an engineering judgment on appropriate treatment. One state noted it is considering using 42­in.­high concrete barriers in some applications and Thrie­beam guardrails in others. Question 4. The participants were asked if their state has special criteria for selecting the barrier type and test level for longitudinal barriers to be installed on a high­speed CSRS. Seven participants said “yes” and 26 answered “no.” The cri­ teria listed by the seven participants included the following: • Use NCHRP Report 350 or MASH (2 states), • Use higher test level than the usual TL­3­based crash his­ tory (3 states), • Use 42­in.­high concrete barrier (1 state), and • Use own state location and design manuals (4 states). While the use of prevailing crashworthiness requirements would seem to be the norm, the last three responses suggest that there are considerations for treating high­speed CSRS differently. Question 5. The participants were asked if their state has available data related to crashes involving longitudinal barriers installed on high­speed CSRS. Thirteen states answered “yes” and 20 answered “no.” Seven states—Arkansas, Delaware, Iowa, Kansas, Ohio, Oklahoma, and Pennsylvania—indicated that they should be able to pull crash data with some limited parameters from crash reports. North Carolina indicated that it could provide a crash dataset for sections of high­speed roadways with the distinction of curved versus straight road­ ways (without details of curvature specifications). Montana and New Jersey indicated that their database may contain some of the crash data on superelevated and curved roadways with guardrails, but the superelevations are not reported. Indiana has 825 collisions with guardrail face/end on Inter­ state highways in 2011, but each crash would then have to be investigated to see if it was on a high­speed curve. Washington State has a State Travel and Collision Data Office (STCDO) that handles crash data. Question 6. The participants were asked if their state is aware of in­service evaluations or accident investigations related to longitudinal barriers installed on high­speed CSRS. Two participants answered “yes” to this question. Illinois mentioned that the crash investigations are confidential, but may be shared if requested for research. North Carolina indi­ cated that many longitudinal barrier analyses have been com­ pleted, with before and after crash data evaluations where they installed barriers as a Spot Safety or Hazard Elimination project for roads that have curves. To date, they have evalu­ ated 26 guardrail projects at bridges and 31 guardrail projects for shoulder applications. Alaska mentioned that crash inves­ tigations are not widespread, and only site­specific evalua­ tion of installed rail is conducted. The responses indicate that detailed in­service evaluations or case studies for crashes on these types of road sections are not generally available. Question 7. The participants were asked if there are locations in their state where longitudinal barriers placed on a high­speed CSRS did not function as desired. Seven par­ ticipants answered “yes,” 22 said “no,” and 4 had no answer. One state had a case where an impacting car encroached into opposite travel lanes through a cable barrier installed in a median on a curve. Another state indicated that a W­beam median guardrail was replaced by a single slope 45­in. concrete Longitudinal Barrier Type # of States That Plan to Use the Barrier in Future Installations Concrete Barrier (All types) 25 • Single slope 10 • F-shape 7 • NJ 3 • Double-faced 2 W-beam guardrail 15 Cable barrier 13 Strong-post W-beam guardrail 8 MGS 8 Thrie-beam 7 Box-beam 3 Table 2.6. Barriers for future installations.

30 barrier in the median of a CSRS because of repetitive hits on the guardrail. One state had a segment of roadway that was the subject of an improvement project to permit the guard­ rail removal because motorists rebounded from the guardrail into the traveled way of an opposing lane, or else impacted the guardrail on the opposite shoulder. Another state is replacing a 32­in.­high concrete barrier rail with a 46­in.­high concrete barrier rail on a section of curved roadway in a mountainous location to reduce the likelihood of large trucks penetrating the barrier. Another state reported trucks that either dumped their loads on a curved overpass or tipped over the concrete parapet. One state indicated it had situations where vehicles penetrated through TL­3 cable barriers. Question 8. The participants were asked to list any avail­ able additional information related to the performance of longitudinal barriers placed on high­speed CSRS. A few states listed their own state guides (posted on their websites). Two states listed the Roadside Design Guide and the Green Book. 2.7 Summary A detailed literature review was conducted to gather infor­ mation and synthesize relevant past efforts. The focus was on studies related to the design, performance, maintenance, and application details of longitudinal barrier systems when placed on CSRS. TRID was used to identify domestic and international reference materials. The literature did not pro­ vide much insight about concerns related to the safety per­ formance of barriers placed on curves, much less on CSRS; however, the following insights were gained: • The nature of impacts on curved sections is not well known. Theoretically, without driver inputs, impacts would occur at shallower angles. The influences of gravitational forces on the impact angle on sloped surfaces have not been analyzed in depth. • There has been very limited testing of barriers on curved sections. The most significant studies undertaken for the FHWA occurred in 1986 and 1993. These efforts included analyses and tests related to curbs, superelevation, and bridge rails by Bronstad and Kimball (1986) and Stout et al. (1993). Later studies focused on barrier interface issues associated with slopes (Sheikh and Bligh 2006; Sheikh et al. 2008). • There have been successful efforts using simulation to understand the trajectories of vehicles on sloped surfaces. Recent applications of VDA software included efforts to determine effective placement of cable barriers on median slopes (Marzougui et al. 2012a). VDA tools had been applied earlier for vehicle performance studies and accident reconstruction. • Design guidance provided in the Green Book focuses on selecting curvatures and superelevation that will allow a vehicle to be driven around a curve at high speeds in com­ fort and the assurance that under wet conditions vehicles traveling at posted speeds would not be likely to lose control. • Guidance for the placement of longitudinal barriers is avail­ able in the Roadside Design Guide for the instances when control is lost, but there is little specific guidance offered for barriers on CSRS. The literature found and reviewed did not provide much insight about concerns related to the performance of barriers placed on CSRS. A state DOT survey was conducted to identify common barriers used on CSRS and to gather information pertaining to specific state standards, guidelines, and practices for the design, installation, and construction for such situations. Representatives from 33 state DOTs responded to the sur­ vey (a response rate of 67%). The responses provided use­ ful information relative to current state DOT standards and practices as follows: • A variety of longitudinal barriers are used for CSRS situa­ tions by the states. • State DOTs do not have specific criteria for longitudinal bar­ riers on CSRS. They tend to accept the NCHRP Report 350 or MASH crashworthiness requirements as sufficient. • No state reported knowledge of issues related to crashes on CSRS from in­service performance reviews or other studies. • State DOTs noted that they plan to use the same types of barriers for future CSRS barrier deployments. • The longitudinal barriers used varied by type and were about equally split between concrete safety shapes and W­ or Thrie­beam designs. • Some states specify a higher barrier for CSRS deployments where there is evidence of a crash problem. • Concrete barriers 42 in. high are sometimes specified. These findings suggest that most state DOTs have not per­ ceived the need for special barrier requirements for CSRS. This might be attributed to the fact that superelevation is more commonly used on high­speed roads that gener­ ally have better safety performance. The state DOT survey revealed that most states do not currently have special design, selection, or installation guidance for the installation of longitudinal barriers on CSRS.