Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems (2012)

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.

53 As noted in Chapter 1, the panel directed the contractor to pursue analyses in the seven areas outlined in Table 1.1. This chapter will provide the details about the approaches used, the factors considered, and the results obtained for the following seven areas: • Cable barrier placement • Cable barrier deflection • End-anchoring and post-anchoring systems • Interconnection with other systems • Horizontal curvatures • Construction and maintenance tolerances • Installation and maintenance costs The intent of conducting the analyses in these areas was to derive science- and data-based insights on the influence of various factors with the subsequent translation of these findings into guide- lines that would facilitate the deployment of effective cable barrier systems for median and road- side applications. As one ventures into these analyses, it is necessary to be cognizant of the many factors associated with cable barriers that may need to be considered. There are many variations in cable barrier system designs, placement, and maintenance. These variations include the following: • Cable barrier system design – Number of cables – Cable positions (i.e., heights) – Cable placement on posts – Cable connectors – Post design – Post spacing – Post embedment including driven vs. placement in foundations – Anchorage features – Anchor spacing – Tensioning elements – Cable type – Tension levels – Transition options • Placement considerations – Median cross-section – Changes in median cross-section C h a p t e r 5 Analyses and Results

54 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems – Tangent and curved roadway sections – Obstacles in the median or roadside – Shoulder/slope design – Drainage requirements – Soil conditions – Seasonal effects – Transitions to other barriers – Installation operations • Maintenance considerations – Probability of nuisance hits – Anchor movement – Post foundation failures – Retensioning – Maintenance cycles – Ease of repair – Requirements for lane closure These factors can have different types and degrees of effects on cable barrier performance. The need exists to assess the implications of the factors to identify the critical ones and translate them into generic guidance. The following sections describe the analyses performed under this study to investigate the effects of some these factors and address the identified cable barrier issues. The results of these analyses were used to develop guidelines for the selection, use, and maintenance of cable barriers. 5.1 Cable Barrier Lateral Placement One of the most critical factors that affects the performance of cable barriers is the lateral placement relative to configuration (i.e., width, slope, shape) of the roadside or median. On flat terrain, almost all currently available/accepted cable barrier systems will perform adequately and can safely redirect most errant vehicles departing the roadway under nominal conditions. However, this is usually not the case when the cable barrier is placed on a sloped median/roadside. The sloped terrain affects the relative height at which the vehicle impacts the cable barrier, i.e., the vehicle could impact the barrier at a higher or lower vertical position compared to that on flat terrain. This phenomenon could lead to a vehicle not fully engaging the cables and consequently underriding or overriding the barrier. It is, therefore, critical to ensure that the barrier is placed at a location where it can capture and/or redirect the majority of vehicles successfully. To investigate the full effects of terrain profiles on cable barrier performance and to develop guidelines for a barrier’s optimum placement, a comprehensive analysis was performed. Vehicle dynamics simulations were conducted to compute the trajectories of vehicles as they traverse a median on a diagonal path. Two commercially available vehicle dynamics programs were used to conduct the simulations and generate data and animations reflecting the trajectories. For each vehicle type considered in these analyses, two points were defined to represent the primary interface (engagement) region on the front of the vehicle. These points are labeled 1 and 2 in Figure 5.1. A trace of these two points viewed from a position standing in the center of the median downstream from the point a vehicle leaves the roadway is shown as the dark lines in Figure 5.2. These same data points can be plotted on a diagram of the median cross-section (as shown in the lower part of Figure 5.2). It can be noted that in moving from left to right, after passing the breakpoint between the shoulder and the median onto a sloped surface, the vehicle will be airborne or at least have a low compression load on its suspension system. At some point the

analyses and results 55 vehicle will land (or return to a distribution of weight on all wheels), and the suspension will compress to absorb the dynamic load. As the vehicle continues its movement across the median there will be a rebound of the suspension as it dissipates energy. Thus, as the vehicle traverses the median, the height of its interface area will vary depending on the state of the vehicle’s suspension system and the slopes of the median. Effective lateral placement of the barrier involves finding the locations where the vehicle’s interface area matches the barrier’s cable heights. For median applications, finding these locations is complicated by the need to have an effective interface for impacts from either direction. The vehicle dynamics programs that were used in this study included HVE (Human Vehicle Environment, by the Engineering Dynamics Corporation) [51] and CarSim (by Mechanical Simulation Corporation) [52]. The programs were developed for use by engineers and safety researchers to study interactions among humans, vehicles, and their environment. They are high-level simulation tools aimed at creating three-dimensional models of vehicles and environ- ments and allow the study of their dynamic interaction under selected conditions. Physical and Point 2 Point 1 Figure 5.1. Critical interface points. Figure 5.2. Trajectory of interface points as the vehicle crosses the median.

56 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems mathematical models provide a detailed description of a motor vehicle that considers the influ- ence of weight, suspension system, and other vehicle factors. Available databases include a wide range of high-fidelity vehicle models that can be used in dynamic reconstructions and simula- tions. HVE and CarSim provide physical and visual environment models to simulate selected con- ditions. Weather attributes, road geometry, and pavement friction properties can be computed and their effects on vehicle dynamics analyzed. Driver’s action (e.g., throttle, brakes, steering, and gear selection) can also be simulated. The models have been thoroughly validated and are capable of predicting accurately a vehicle’s trajectory for different terrain profiles. NCAC used these programs in several cable barrier research studies, and the results were compared to full- scale crash tests. The predictions were a close match to the full-scale crash tests. Figure 5.3 shows predictions from HVE at the start of impact compared to two full-scale crash tests. The research considered a broad set of influencing factors as shown in Figure 5.4. This figure shows a typical divided highway where the median is the green area between the shoulders. The median can be of different widths and cross-sections. The cable median barrier is placed somewhere in the median and can be hit from either side. For the situation shown in Figure 5.4, a vehicle leaving the bottom roadway would have a “nearside” hit on the barrier. From the upper roadway, the vehicle would have a “farside” hit. A cable median barrier has to be located such that it functions effectively for both nearside and farside hits. Today’s vehicle fleet is a heterogeneous mix of vehicles with varying shapes and sizes. Figure 5.5 shows a sample assortment of vehicles from the fleet and the variations in the heights of their bumpers and primary structures. To address the effect of vehicle type on barrier performance, several vehicle models were used in the analyses to create an envelope of vehicle trajectories. Figure 5.3. Comparisons of HVE predictions and full-scale crash test results.

analyses and results 57 These vehicle models included a pickup truck (Chevrolet C2500) and a small car (Honda Civic) to represent the two NCHRP Report 350 required test vehicles. A large sedan (Crown Victoria) was also included in the analysis. This vehicle type has been found in previous studies to be critical for cable barrier performance due to its front profile (similar to a small car) and its mass (similar to a pickup truck). Additionally, a larger pickup truck and a sedan representing the new MASH 2270P and 1100C vehicles, respectively, were included in the vehicle dynamics analyses. Defining “effective interface conditions” for any cable barrier design and any median configuration can be accomplished in various ways. For this analysis, effective interface conditions were determined by the following: • Assessing relative positions of the vehicle to the barrier such that – To minimize the potential for override, the top cable should contact the vehicle above Point 1 (lower critical point in Figure 5.1). – To minimize the potential for avoid underride, lower cable should contact the vehicle below Point 2 (upper critical point in Figure 5.1). Vehicle Models: - 2000P - 820C - Mid-size Sedan Shoulder Shoulder Median Median Barrier Near Side Impact Far Side Impact Initial Speeds: 50, 70, 100 km/h Approach Angles: 5 to 25 deg 5 o 25 o - 2270P - 1100C Figure 5.4. Factors considered in cross-median events. Figure 5.5. Variation in vehicle front profile.

58 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems • Defining the impact conditions to be considered. One approach would be to follow NCHRP Report 350 or MASH requirements. In this analysis, a broader view was taken requiring that the conditions should reflect a broader range of approach angles, speeds, and vehicle types. • Associating interfaces with specific median configurations (i.e., width, shape, side slopes, and depth). Two critical points (Points 1 and 2) were used for each vehicle to assess its interface with the barrier when placed at different locations along the lateral direction of the median. These critical points were selected based on the geometry of the front structure of the vehicles and by examining full-scale crash tests of these vehicles impacting different cable barrier systems. (Please note that references to colors in the following text will be clear in the color figures which are available in the online version of the report which can be found by searching the TRB website for NCHRP Report 711.) Using the results of vehicle dynamics simulations, trace paths of Point 1 for vehicles crossing a median from both directions were plotted in Figure 5.6 with each individual trace representing a specific vehicle, speed, impact angle, and crossing direction. These curves are “normalized” to relate the relative heights of individual cables in the barrier, or the height of the effective interface area on the front of a vehicle to a horizontal plane. For any position across the median, the vertical height of the normalized plot to actual sloped surface is equivalent. Normalization is useful for comparing various types and features of medians and cable barriers. The array of lines represents the broader set of impact cases for the specified parameters. The heavy blue line represents the overall maximum heights for Point 1 for the set of impact cases associated with this particular median configuration. Similarly, plotting all cases for Point 2 yielded the array of lines in Figure 5.7 for the set of impact cases for a given median configuration. The heavy green line represents the overall minimum heights for Point 2. These plots were generated for different median profiles and can be used to define vehicle-to-cable- barrier engagement based on cable barrier lateral position and its cable heights. Comparing the resulting blue (minimum) override limit and green (maximum) underride limit lines for a given median provides a means of determining the interface effectiveness across all lateral positions for any given barrier design. The three yellow lines in Figure 5.8 represent the coverage of a particular barrier design (in this case a generic three-cable barrier). Where the blue Figure 5.6. Sample override limit curve.

analyses and results 59 line goes above the highest yellow line there is an opportunity for an override to occur. Where the green line falls below the lowest yellow line, the possibility of an underride exists. This approach is used to determine the potential effectiveness for varying cable barrier systems (e.g., number of cables, relative heights) across all possible lateral positions for a given median configuration. The following figures show more specific examples of how this metric can be applied. Figure 5.9 shows an example of the results for a specific median. The upper portion shows the normalized representation of the interface envelope, the minimum upper cable height curve, the maximum lower cable height curve, and the relative position isobars for a specific type of cable barrier (i.e., generic, low-tension, 3-cable system). The Barrier Interface Envelope is the gray shaded area that surrounds all of the trace bars for different vehicles traversing the median at varying angles and speeds from both directions. Figure 5.7. Sample underride limit curve. Override Limit Underride Limit Median Profile Figure 5.8. Sample override and underride limits plot.

60 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems The lower portion of Figure 5.9 shows the profile or cross-section of the median related to the upper graph. The green hatched portions indicate the lateral positions where this specific barrier will be effective. Since the median in this case is symmetric, the effectiveness regions are a mirror image on the opposite side. The red hatched area defines the lateral positions where the specific barrier has a cable arrangement that has a lower cable above the maximum lower cable height curve (green) and/or an upper cable below the minimum upper cable height curve (blue). Effective lateral placement occurs where both criteria are met. It can be noted that for this spe- cific barrier system, the red region corresponds to the lateral placement range where the maxi- mum lower cable curve height falls below the lowest cable in the system. This plot would indicate that for this 14.6 m (48 ft) wide median (measured from hinge point to hinge point of median) with 6H:1V side slopes, there is an area from about 0.3 to 1.8 m (1 ft to 6 ft) from the center of the median where placement of this generic cable barrier system is not recommended because of the risk of underriding. Numerous plots of this type for different median profiles were generated and were used in developing the guidelines. Some of these plots are included in Appendix C of the contractors’ final report. In this research, a number of different possible conditions for a median crossing were considered. The following conditions were assumed in the analyses: • Median has firm surface. Ploughing (or furrowing) into the surface by tires is negligible. Modeling the ploughing condition is beyond the capability of current computer simulation software. This condition has not been incorporated in all previous related research studies (using full-scale tests or computer simulations) and is considered beyond the scope of this project. • Vehicles are “tracking” as they enter the median (i.e., vehicle’s initial speed vector is in the same direction at its longitudinal axis). Even though it is possible to investigate non-tracking conditions using the vehicle dynamics programs, incorporating all non-tracking scenarios would render the number of simulations impractical. Different vehicle types, impact angles, and initial speeds are considered in this study, which should account for most vehicle/barrier interface situations. Figure 5.9. Sample plot generated using the results of vehicle dynamics analysis.

analyses and results 61 • Initial velocity occurs when the vehicle leaves the shoulder. Some deceleration is expected to occur (3–5 mph was noted in the research) for vehicles as they cross the median. • There are no driver inputs (e.g., steering, braking) that affect the vehicle trajectory. • No edge rounding was considered in this study. Based on previous investigations, edge rounding reduces the potential of overrides with no significant effect on the underrides. • A vehicle must have effective engagement with a minimum of one cable to be captured by the barrier. It is important to note that review of full-scale crash testing has indicated that for low-tension systems an engagement with one cable is sufficient to capture the small car but may not be suf- ficient to redirect mid-size and larger vehicles. Engagement with one cable was shown to be suf- ficient for the small vehicle and pickup truck for high-tension systems. This difference could be attributed to the fact that the connection between the cables and posts is stronger in high-tension systems than in low-tension systems. Additionally, the green hatched regions in the generated plots indicate barrier lateral placement where a minimum of one cable would engage the vehicle. This is a minimum condition that has to be satisfied but does not ensure a successful redirection or controlled stopping of the vehicle. In addition to this engagement condition, the barrier needs to have adequate strength to withstand the vehicle impact forces. This can be achieved through full-scale crash testing. At the time that this report was written, a matrix of full-scale crash tests that evaluate cable barrier systems when placed on sloped medians was being developed. These tests will ensure that the cable barrier design (e.g., the strength of connection between cables and posts, the post spacings, the cable vertical spacings, etc.) is adequate to capture and redirect the vehicle without leading to rollover, override, underride, or penetration in between the cables. The approach described in the previous sections allows for identifying optimum placement of cable barriers for specific median configurations. To develop generalized guidelines that can be used for a wider range of median geometries, the analysis was taken a step further. This was achieved by superimposing (i.e., overlaying) results from median profiles with varied widths. As an example, to evaluate the overall cable barrier performance on 4H:1V slope V-shaped medi- ans, the results from all simulations from this type of median with varied widths are combined. Figure 5.10 shows the lower cable maximum limits from 4H:1V slope V-shaped medians with varied median widths (5 to 17 m, 16 ft to 56 ft). The minimum of all curves is computed (shown in thick green line in the figure). Similarly, Figure 5.11, shows the upper cable minimum limit Zero Point at Center of Median Figure 5.10. Normalized underride limit plot for 4H:1V V-shaped medians, varied widths.

62 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems for the same medians with varied median widths (5 to 17 m, 16 ft to 56 ft) and the combined maximum curve (thick blue curve). Together, the minimum and maximum curves can be used to assess barrier performance on any 4:1 V-shaped median. The research considered a wide range of median profiles. The medians varied in shape, slope, and width. For each median profile, simulations with different vehicle types traversing the median at different speeds (from 50 to 100 km/h, 31 to 62 mph) and angles (from 5° to 25°) were performed, and the vehicle trajectory was determined. These trajectories were used to assess vehicle-to- barrier interface when the barrier is placed at different locations across the median. Figures 5.12 and 5.13 present summaries of the results for symmetric V-shaped medians. Figure 5.12 shows the underride limit plots for medians with slopes of 4H:1V, 6H:1V, 8H:1V, 10H:1V, and 12V:1H (with median widths from 5 to 17 m, 16 to 56 ft—not including shoulders). Cable heights of 533, 686, and 838 mm (21, 27, and 33 in.) are included in the plot to assess vehicle engagement Zero P oint at Edge of Median Figure 5.11. Normalized override limit plot for 4H:1V V-shaped medians, varied widths. Zero Point at Center of Median Critical under-ride region (-8 to 8 ft) Figure 5.12. Normalized underride limit plot for V-shaped medians 4H:1V to 12H:1V slope, varied width.

analyses and results 63 for systems where the top cable is at least 838 mm (33 in.) and the bottom cable is no higher than 533m (21 in.), which covers the majority of available and installed cable barriers. It can be noticed from the plots that there is potential for underride in the region between 0.3 and 2.4 m (1 ft and 8 ft) from the median center. The plot shows that even for shallow (12H:1V slope) medians there is potential for underride near the bottom of V-shaped medians. In the outer region, farther than 2.4 m (8 ft) from the median center, the results indicate that the vehicle is likely to engage at least one cable if the cable height is at 533 mm (21 in.) or lower. For the center region, between -0.3 and 0.3 m (-1 ft and 1 ft) from the center of the median, the plot shows that the barrier would engage one cable for 6H:1V and shallower slopes, while there is a potential for underride with the 4H:1V slope. Figure 5.13 shows the override limit plots for symmetric V-shaped medians with slopes of 4H:1V, 6H:1V, 8H:1V, 10H:1V, and 12H:1V. For 4H:1V sloped medians, the results indicate that there is potential for override in the region between 1.2 and 6.0 m (4 ft and 20 ft) from the edge of the median. In the region between 0.6 and 1.2 m (2 ft and 4 ft) and between 6.0 and 6.7 (20 ft and 22 ft) from the edge of the median, the vehicle would likely engage only the top cable. In the region between 0 and .6 m (0 and 2 ft), the vehicle would engage two cables. Similarly, in the region farther than 6.7 m (22 ft) from edge of the median, two cables would be engaged. For medians with slopes flatter than 6H:1V, the plot shows that there is no potential for override, i.e., the vehicle would engage a minimum of one cable if placed anywhere in the median. Similar plots were generated for flat-bottom medians. The plots; shown in Figures 5.14 through 5.16, include flat-bottom medians with slopes of 4H:1V, 6H:1V, and 8H:1V. For each slope, three depths, 0.6, 1.2, and 1.8 m (2 ft, 4 ft, and 6 ft), were analyzed. A third parameter that was varied is the width of the flat-bottom section, which was varied from 1.2 to 12.2 m (4 ft to 40 ft). The trajectories from all cases (i.e., different flat-bottom median profiles, vehicle type, speed, and angle) were included when generating the underride and override limits. Figure 5.14 shows the underride limit for the three slopes analyzed. The plot shows that for the 4H:1V slope profiles, there is potential for underride in the region between -1.8 and 3 m (-6 ft and 10 ft) from the flat-bottom breakpoint. In the outer region, farther than -1.8 m (-6 ft) in the up-sloped region and farther than 3 m (10 ft) in the flat region, the results indicate that the vehicle is likely to engage at least one cable if the bottom cable height is at 533 mm (21 in.) or lower. For 6H:1V Over-ride all cables 1 cable engage d 2 ca ble s engaged Cable e nga gem ents for 4H :1V sloped me dia ns ( over-ride lim it) Zero P oint at Edge of Median Figure 5.13. Normalized override limit plot for V-shaped medians 4H:1V to 12H:1V slope, varied width.

64 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems sloped profiles, the underride region is smaller (-1.2 m to 1.5 m [-4 ft to 5 ft] for 6H:1V and -1.2 m to 1.2 m [-4 ft to +4 ft] for 8H:1V). Figure 5.15 shows a plot similar to Figure 5.14 except that only medians with the flat-bottom section wider than 2.4 m (8 ft) are included. If the flat-bottom section is wider than 2.4 m (8 ft), the plot shows that the barrier will engage the vehicle if it is placed at the flat bottom breakpoint for medians with 6H:1V slopes and shallower. Figure 5.16 shows the override limit plots for symmetric flat-bottom medians with slopes of 4H:1V, 6H:1V, 8H:1V. The results are very similar to the V-shape profiles. For 4H:1V sloped medians, the region between 1.2 to 6.0 m (4 ft and 20 ft) from the edge of the median shows potential for override. In the regions 0.6 to 1.2 m (2 ft to 4 ft) and 6.0 to 6.7 m (20 ft to 22 ft) from the edge of the median, the vehicle would likely engage the top cable only. In the region between 0 to 0.6 m (0 to 2 ft), the vehicle would engage two cables. Similarly, in the region farther than Zero at Flat- Bottom Break Point Figure 5.14. Underride limit plot for flat-bottom shaped medians 1.2 to 12.2 m (4 ft to 40 ft) flat-bottom width. Zero at Flat- Bottom Break Point Figure 5.15. Underride limit plot for flat-bottom shaped medians 2.4 to 12.2 m (4 ft to 40 ft) flat-bottom width.

analyses and results 65 6.7 (22 ft) from the edge of the median, two cables would be engaged. For medians with slopes flatter than 6H:1V, the plot shows that there is no potential for override, i.e., the vehicle would engage a minimum of one cable if placed anywhere in the median. Vehicle dynamics analyses were conducted to investigate vehicle-to-barrier interactions on non-symmetric medians. Figure 5.17 depicts a typical override and underride plot from these analyses. The plot shows that the side with the shallower slope is less susceptible to overrides. The underride region is similar on both sides of the median. Comparing this plot to symmetric medians, similar observations can be made for the critical override and underride regions. Zero at Edge of Median Figure 5.16. Override limit plot for flat-bottom shaped medians. Figure 5.17. Sample underride and override limit plot for non-symmetric medians 4H:1V and 12H:1V slope—14.6 m (48 ft) median width.

66 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Zero at Edge of Median Figure 5.18. Override limit plots for 4H:1V slope medians at varied superelevations. To investigate the effects of superelevation on vehicle to barrier interface, vehicle dynamics simulations with varied superelevations (0 percent, 2 percent, 4 percent, 6 percent, and 8 percent) were conducted. Two symmetric V-shaped medians, with 4H:1V and 6H:1V side slopes, were used in the analysis. The width of both medians was 17 m (56 ft), from break point to break point (similar results are expected for other median types and widths). Simulations with varied vehicle types, initial speeds, and approach angles were performed to investigate the superelevation effects on the vehicle trajectories. The simulation results showed that higher superelevation increases the chance of overrides. This is especially critical for medians with steep side slopes (steeper than 6H:1V). Figures 5.18 and 5.19 show the normalized override limit plots for the 4H:1V and 6H:1V sloped medians, respectively. It can be noticed from the figures that the vehicle height relative to ground level increased with increased superelevation. The simulation results showed that the superelevation had negligible effects on the underride limit plots. Zero at Edge of Median Figure 5.19. Override limit plots for 6H:1V slope medians at varied superelevations.

analyses and results 67 5.2 Cable Barrier Deflection The dynamic deflection of a cable barrier during impact is an important characteristic for many reasons. Compared to semi-rigid W-beam barriers and rigid concrete barriers, cable barriers have greater deflections, which is the reason that cable barriers typically are more forgiving to the impacting vehicle’s occupants. For the barrier to be safe, adequate space behind the barrier that is clear of hazards must be provided to accommodate the expected deflections. If deflections exceed the space provided, the errant vehicle could impact rigid objects behind the barrier, or worse yet, in median applications on divided highways, cross into opposing traffic. Various cable barrier systems have been crash-tested successfully and accepted for use on U.S. highways. High-tension cable systems, when compared to low-tension systems, have lower deflection during impacts and reduced maintenance costs. NCHRP Report 350 and MASH require that dynamic deflections observed in the crash tests be reported. These requirements do not standardize all of the test installation features for cable barrier systems. For example, cable barriers are typically tested at 90 to 180 m (300 to 600 ft) installation lengths with the cables anchored at both ends. However, field installations on U.S. highways are typically much longer with anchor-to-anchor distances more than 10 times longer than the ones used in crash tests. With longer anchor-to-anchor spacing, the deflection of the barrier could be significantly higher and could lead to higher likelihoods of barrier penetration. Barrier deflection is also affected by the spacing between the posts, the strength of the posts, the connection between the cable and posts, and the amount of tension in the cable. Understanding the influence of these features on deflection during impact is critical to making effective design decisions about the features of cable barrier systems. To evaluate the influence of critical installation parameters, such as end-anchor spacing, cable tension, post spacing, etc., on cable barrier deflection, crash scenarios with varied configurations have to be analyzed. To conduct such analyses using only full-scale crash tests would be very costly. Additionally, conducting full-scale crash tests with very long cable barrier installations (300 m [980 ft] and longer) is often not practical and beyond the capabilities of most test facilities. The approach used in this study is based on the coupling of full-scale crash testing with computer simulations using the LS-DYNA finite element program [53]. First, the cable barrier modeling approach was validated using previously conducted full-scale crash tests. Several full-scale crash tests with varied barrier design, barrier length, and post spacing were used in the validation process. Upon completing the validation of the modeling approach, a matrix of simulations with varied design parameters and installation configurations was run to investigate the effects on cable barrier deflection. The installation configuration variations included the following: • End-anchor spacings—Anchor spacings of 100 m (328 ft) (typical spacing used for NCHRP 350 crash tests), 200 m (656 ft), 300 m (984 ft), 500 m (1,640 ft), and 1,000 m (3,280 ft). • Initial cable tensions in the system before impact—Cable tensions, 15 kN (3.4 kips) and 24 kN (5.4 kips). These tensions approximately represent typical hot weather (38°C, 100°F) and average weather (10°C, 50°F) conditions, respectively, for high-tension cable barriers. • Post spacings—Post spacings of 1.6 m (5 ft), 3.2 m (10 ft), 4.8 m (15 ft) and 6.4 m (20 ft). Five systems were selected for the analyses: Brifen Wire Rope Safety Fence, Gibraltar Cable Barrier System, Nucor Steel Marion Cable Barrier System, Safence Cable Barrier System, and Trinity CASS Cable Barrier System. Information was collected for these five systems, including design drawings, crash test reports, crash test video clips, test data, and acceptance letters. The information for these five proprietary systems was obtained from the manufacturers. These systems are available in different configurations. One system from each cable barrier manufacturer was selected. To assist in the selection, the team contacted the manufacturers to get their feedback on which system should be selected. When selecting the systems, emphasis was placed on choosing

68 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems systems that are most commonly installed and have multiple full-scale crash test data available. This process ensured that the computer models could be validated fully with crash test data and that the analyzed systems would represent the majority of installed systems. NCAC models for standard NCHRP Report 350 test vehicles were used in the simulations. These included the Chevrolet C2500 pickup truck (2000P) model and the Geo Metro (820C) vehicle models. These vehicle models were originally validated and subsequently updated over years of application in many crash simulation efforts [54, 55, 56]. These models conformed to the test vehicles reflected in the available crash test data. Since maximum dynamic deflections were the metric of interest, impacts with the 2000P vehicle at 100 km/h (62 mph) and 25° impact angle were the focus. Similar impact location along the barrier was used for all simulations. The impact location was selected such that the maximum deflection would occur at the center of the barrier installation. Simulations with all of the above-stated installation variations were carried out, and the results are presented to show the effects on deflection of end-anchor spacings, post spacings, and initial cable tensions. Model Development and Validations Highly detailed computer models of cable barrier systems were created and used in this study. A sophisticated modeling approach was used in creating these models to ensure that they would accurately capture the barrier response during simulation of the crash. The approach used in modeling the different cable barrier systems is described in the following sections. To create the finite element models of the cable barrier systems, several key features were examined carefully, and appropriate modeling techniques were used to ensure that the models were accurate representations of the actual systems. First, explicit geometry of all components of the system was incorporated in the model including, for example, the cables, the posts, the sleeves, etc. This step was important to ensure that the correct mass, inertia, and stiffness of the different parts are reflected in the model. The soil and concrete were also modeled explicitly using solid elements. The shape of the post/sleeve was incorporated in the soil or concrete mesh to simulate the post/soil interactions. The cables for the models were created using beam elements with the cross-sectional and material properties of the specified cable. To replicate the cable-to- vehicle and cable-to-post interactions accurately, each beam was surrounded by shell elements with null material properties. The beams were connected to the null shell elements using nodal rigid body connections. The necessary initial stress was applied to the beam elements in the initialization phase of the simulation to simulate the pre-crash tension in the cables. For all systems, the connection between the cables and end anchor was considered rigid. To validate the modeling approach, computer simulations, setup in a similar configuration to the full-scale crash tests, were conducted and the results were compared to the tests. Full-scale crash tests, with varied barrier design, end-anchor spacing, and post spacing were used in the validations. Some of these validations are described in the following sections. Gibraltar Cable Barrier System The three-cable, high-tension median cable barrier from Gibraltar Cable Barrier Systems was accepted by the FHWA for use on U.S. highways [57, 58]. This cable barrier system con- sists of three 19 mm (¾ in.) diameter steel cables supported by steel 83 × 63.4 × 3.8 mm thick (3.25 × 2.5 × 0.15 in. thick) and 1,500 mm (4.9 ft) long C-posts. The bottom, middle, and top cable heights are set at 508 mm (20 in.), 762 mm (30 in.), and 990 mm (39 in.), respectively. The three cables are locked in place using an 11 mm (7⁄16 in.) diameter galvanized steel hairpin and lock plate that fits inside each post. The finite element model of the Gibraltar cable barrier system was created and validated with two full-scale crash tests. The full-scale crash tests were conducted by Karco Engineering, LLC (Test Report No. TR-P26021-01-B and TR-P26028-01-B). Figure 5.20

analyses and results 69 shows the details in the finite element model of the Gibraltar cable barrier system. The geometry and design details of all components were obtained from FHWA acceptance letters. The total installation length for both tests was 93 m (305 ft), and the cables were tensioned to 25 kN (5.7 kips). For the first test, the line posts were set on 3 m (10 ft) centers and for the second test, 9.1 m (30 ft) centers. For both tests, a 2000P C2500 vehicle was used, the impact angle was 25°, and impact speed was 100 km/h (62 mph). Figure 5.21 shows side-by-side comparisons of sequential images from the first crash test and its simulation. In the simulation, the two top cables engaged and redirected the vehicle with a maximum dynamic deflection of 1.9 m (6.4 ft), and the maximum dynamic deflection for the crash test was 2.0 m (6.8 ft). Figure 5.22 shows side-by-side comparisons of sequential images from Crash Test 2 and its simulation. In the simulation, the two top cables engaged and redirected the vehicle with a maximum dynamic deflection of 2.9 m (9.5 ft), and the maximum dynamic deflection for the crash test was 2.8 m (9.3 ft). Safence Cable Barrier System The Safence Wire Rope Barrier produced by Blue Systems was accepted as an NCHRP Report 350 TL3 traffic barrier in 2001 [59]. The original design consisted of four 19 mm (¾ in) diameter steel cables supported on 2.1 m (83 in) long elliptically shaped posts spaced on 2.5 m (8.2 ft) centers. In its current design [60], the cables are supported using Safence C-shaped posts embedded in concrete footings. Each post was stiffened at the ground line by adding a steel plate inside the C-post to increase its resistance to bending, and a steel hook was added to the top of each post to retain the cables within the post center slot for a longer time upon barrier impact. The finite element model of the Safence system was developed and validated using a VTI crash test (Test Report No. 56649). The design drawings and details were obtained from the FHWA acceptance letter [61] and manufacturer’s drawings. Figure 5.23 shows details of the finite element model of the Safence cable barrier system. The overall dynamics of the vehicle in the finite element simulation were similar to those reported in the crash test. The maximum dynamic deflection for the simulation was 3.6 m Figure 5.20. Finite element model of the Gibraltar cable barrier system.

70 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Figure 5.21. Comparison of sequential plots; Gibraltar Test 1.

analyses and results 71 Figure 5.22. Comparison of sequential plots; Gibraltar Test 2.

72 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems (11.7 ft) and for the crash test, 3.7 m (12.1 ft). Side-by-side comparisons of sequential images of the test and simulation are shown in Figure 5.24. Brifen Cable Barrier System Brifen Cable Barrier Systems incorporate a proprietary interweaving concept. The systems have 19 mm (¾ in) cables supported by S-shaped posts that are 100 mm × 55 mm × 4.55 mm thick (4 in. × 2 in. × 0.18 in.) and manufactured from ASTM A-36 steel. The top cable is set in a 101 mm deep × 22 mm (4 in. × 1 in.) wide slot cut into the top of each post and bottom cables (two in case of a 3-cable system and three in case of a 4-cable system) are interwoven between the posts. Figure 5.25 shows the details of the finite element (FE) model of the Brifen wire rope system. The FE model was validated using two full-scale crash tests (Test B-USA-C-2 and Test BCR-1). In the first test, the test vehicle was a 2000P C2500 pickup truck impacting a 3-cable system at a 25° angle and 100 km/h (62.1 mph). The test article, installed on flat terrain, had a total length of 278 m (912 ft) and post spacings of 3.2 m (10.5 ft). Figure 5.26 shows side-by-side com- parisons of sequential images from the test and simulation. In both cases, the top two cables engaged and redirected the vehicle. Maximum dynamic deflection observed in the test was 2.6 m (8.6 ft) and, in the simulation, 2.7 m (8.9 ft). The overall dynamics of the vehicle showed a good correlation with the test. Figure 5.27 shows the comparison of vehicle CG yaw between the test and simulation. In the second test, the vehicle was a 2000P C2500 pickup truck impacting a 3-cable system at a 25° angle at 100 km/h (62 mph). The test article, installed on flat terrain, had a total length of 111 m (365 ft) and post spacings of 3.2 m (10.5 ft). Figure 5.28 shows side-by-side comparisons of sequential images from the test and simulation. Vehicle behavior in the simulation showed Figure 5.23. Finite element model of the Safence cable barrier system.

analyses and results 73 Figure 5.24. Comparison of sequential plots; Safence cable barrier system.

74 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Figure 5.25. Finite element model of the Brifen wire rope system. Figure 5.26. Comparison of sequential plots; Brifen Test 1.

analyses and results 75 Figure 5.26. (Continued). Figure 5.27. Vehicle yaw comparison; Brifen Test 1.

76 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Figure 5.28. Comparison of sequential plots; Brifen Test 2.

analyses and results 77 a good correlation with the test. Maximum dynamic deflection observed in the test was 2.1 m (6.9 ft) and, in the simulation, 2.2 m (7.3 ft). Figure 5.29 shows the vehicle yaw comparison between the test and simulation. Trinity CASS Cable Barrier System The details of the finite element model of the Trinity CASS Cable Barrier System are shown in Figure 5.30. The system consisted of three 19 mm (3/4 in) steel cables supported on posts that were spaced 3 m (10 ft) apart. The posts were installed in concrete foundations that were 300 mm (1 ft) in diameter and 760 mm (2.5 ft) deep. The test vehicle was a 2000P C2500 pickup. The impact speed was 100 km/h (62 mph), and the impact angle was 24.2°. The total installation length was 102 m (335 ft). Figure 5.29. Vehicle yaw comparison; Brifen Test 2. Figure 5.30. Finite element model of the CASS cable barrier system.

78 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems In both the test and simulation, the top two cables engaged the vehicle and redirected it. The maximum dynamic deflection for the simulation was 2.5 m (8.1 ft) and, for the crash test, 2.4 m (7.9 ft). Comparisons of the limited sequential plots are shown in Figure 5.31. The vehicle tra- jectories in the crash test and simulation show good comparison. Nucor Cable Barrier System The details of the finite element model of the Nucor cable barrier system was created and used for validation (Figure 5.32). This system uses three 19 mm (0.75 in.) steel 3 × 7 cables. The cables are attached to 6 kg/m (4 lb/ft) U-channel steel posts using locking hook bolts, 6.4 mm (0.25 in.) in diameter. The three cables are placed at heights of 545 mm (21.5 in.), 650 mm (25.6 in.), and 750 mm (29.5 in.). Two of the cables (top and bottom) are placed on one side of the posts while the other cable (middle) is placed on the opposite side. The posts are spaced at 3.8 m (12.5 ft) and anchored using 100 mm (4 in.) diameter 12-gauge steel pipe sockets embedded in a 300 mm (12 in.) diameter by 760 mm deep reinforced concrete footing and embedded in soil with a trapezoidal soil plate. The barrier length was 101.4 m (333 ft). In both the test and simulation, the top two cables engaged the vehicle and redirected it. The maximum dynamic deflection for the simulation was 1.9 m (6.2 ft) and, for the crash test, was 1.8 m (5.9 ft). Comparisons of sequential plots from the simulation and test are shown in Figure 5.33. Simulation Results After completing the development of finite element models of the available cable barrier systems and validating them by comparison to previously conducted full-scale tests, computer models reflecting various post spacings, end-anchor spacings, and cable initial tensions were created. In all cases, the barrier was set up on flat, level terrain and impacted with the 2000P (Chevrolet C2500) pickup truck at 100 km/h (62 mph) initial speed and 25° impact angle. The posts in all systems were placed in sockets embedded in concrete foundations. The systems used in these simulations do not have the exact design as the ones used for the validations. The designs were chosen based on consultations with the cable barrier manufacturers to select the systems that are most commonly installed and most crash-tested. It is important to note that not all systems selected for the analysis use the same number of cables. Cable Initial Tension Effects Figure 5.34 shows barrier deflections for two different systems at two initial tension levels (15kN and 24 kN, 3.4 and 5.4 kips) and varied barrier lengths (100 to 1,000 m, 328 to 3,280 ft). The simulations showed that lower initial tension leads to increased barrier deflection. However, the magnitude of the increase in deflection is small compared to the actual deflection. The simu- lations showed that the maximum tension reached in the cables at the end anchors to be four to five times higher than the initial tension. A reduction in initial tension from 24 kN to 15 kN (5.4 to 3.4 kips), 38 percent, would therefore have less of an effect on the significantly higher maximum tension and, consequently, small effects on barrier deflection. Full-scale crash tests showed that barrier deflections from generic low-tension cable barrier systems are significantly higher (almost twice) than those observed in the high-tension systems. The reason for this difference in deflection between low-tension and high-tension systems seen in the crash tests is attributed more to the cable/post connections than the initial tension. Most high-tension systems have a significantly stronger cable-to-post connection (by weaving the cables, using a splice at the center of the post, etc.) than the low-tension systems (which typically use open

analyses and results 79 Figure 5.31. Comparison of sequential plots from CASS simulation.

80 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Figure 5.32. Finite element model of the Nucor cable barrier system. Figure 5.33. Comparison of sequential plots from Nucor simulation.

analyses and results 81 Figure 5.33. (Continued). 5 6 7 8 9 10 11 0 500 1000 1500 2000 2500 3000 3500 4000 1.5 2.0 2.5 3.0 3.5 0 200 400 600 800 1000 1200 B ar ri er D ef le ct io n (f t) End-anchor Spacing (ft) B ar ri er D ef le ct io n (m ) End-anchor Spacing (m) Weaved System - 24 kN (5400 lb) Tension - 3.2 m (10.5 ft ) Post Spacing Weaved System - 15 kN (3400 lb) Tension - 3.2 m (10.5 ft ) Post Spacing Parallel System - 24 kN (5400 lb) Tension - 3 m (9.8 ft ) Post Spacing Parallel System - 15 kN (3400 lb) Tension - 3 m (9.8 ft ) Post Spacing Figure 5.34. Effect of initial cable tension on barrier deflection.

82 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems hooks to hold the cables to the posts). These stronger cable/post connections delay the release of the cable and lead to reduction in barrier deflection. End-Anchor Spacing and Post Spacing Effects To investigate the effects of end-anchor spacing and post spacing on barrier deflection, the simulations with different cable barrier lengths and post spacings were compared. Figures 5.35 through 5.38 show computer-predicted deflections of five cable barrier systems. The figures show the deflections at different end-anchor spacings (100 m to 1,000 m, 328 to 3,280 ft) and post spacings (1.6 to 6.4 m, 5 to 20 ft). The simulation results show that for all systems the deflection increases as the spacing between the end anchors is increased. The results also show that the effect of end-anchor spacing is different for different cable barrier systems. The difference is mainly attributed to the effect of the cable/post interaction. Systems that restrict the longitudinal sliding of the cables relative to the posts (by engaging the posts or other means) lead to a smaller deflection increase when the end-anchor spacing is increased. The simulations show that the ratio between the increase in barrier deflection and the increase in anchor spacing was less between the 300 m (980 ft) and 500 m (1,640 ft) anchor spacings and even less between the 500 m (1,640 ft) and 1,000 m (3,280 ft) anchor spacings. For all systems, the simulations show that barrier deflection increases as the post spacing increases. The rate of increase in deflection decreases as the post spacing increases. There was about 30 to 50 percent increase in deflection from 1.6 m (5 ft) post spacing to 3.2 m (10 ft) post spacing, about 10 to 28 percent increase in deflection from 3.2 m (10 ft) post spacing to 4.8 m (15 ft) post spacing, and about 6 to 14 percent increase in deflection from 4.8 m (15 ft) post spacing to 6.4 m (20 ft) post spacing. 3 5 4 6 7 8 9 10 11 12 13 0 500 1000 1500 2000 2500 3000 3500 4000 1.5 2.0 1.0 2.5 3.0 4.0 3.5 0 200 400 600 800 1000 1200 B ar rie r D ef le ct io n (ft ) End-anchor Spacing (ft) B ar rie r D ef le ct io n (m ) End-anchor Spacing (m) 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 4.8 m (15 ft) Post Spacing 6.4 m (20 ft) Post Spacing Figure 5.35. Deflection plots for Gibraltar cable barrier system.

analyses and results 83 3 5 4 6 7 8 9 10 11 12 13 0 500 1000 1500 2000 2500 3000 3500 4000 1.5 2.0 1.0 2.5 3.0 4.0 3.5 0 200 400 600 800 1000 1200 B ar rie r D ef le ct io n (ft ) End-anchor Spacing (ft) B ar rie r D ef le ct io n (m ) End-anchor Spacing (m) 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 4.8 m (15 ft) Post Spacing 6.4 m (20 ft) Post Spacing Figure 5.36. Deflection plots for Safence cable barrier system. 3 5 4 6 7 8 9 10 11 12 13 0 500 1000 1500 2000 2500 3000 3500 4000 1.5 2.0 1.0 2.5 3.0 4.0 3.5 0 200 400 600 800 1000 1200 B ar rie r D ef le ct io n (ft ) End-anchor Spacing (ft) B ar rie r D ef le ct io n (m ) End-anchor Spacing (m) 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 4.8 m (15 ft) Post Spacing 6.4 m (20 ft) Post Spacing Figure 5.37. Deflection plots for Brifen cable barrier system.

84 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems 3 5 4 6 7 8 9 10 11 12 13 0 500 1000 1500 2000 2500 3000 3500 4000 1.5 2.0 1.0 2.5 3.0 4.0 3.5 0 200 400 600 800 1000 1200 B ar rie r D ef le ct io n (ft ) End-anchor Spacing (ft) B ar rie r D ef le ct io n (m ) End-anchor Spacing (m) 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 4.8 m (15 ft) Post Spacing 6.4 m (20 ft) Post Spacing Figure 5.38. Deflection plots for CASS cable barrier system. 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 4.8 m (15 ft) Post Spacing 6.4 m (20 ft) Post Spacing 3 4 5 6 7 8 9 10 11 12 13 0 500 1000 1500 2000 2500 3000 3500 4000 1.0 1.5 2.0 2.5 3.0 4.0 3.5 0 200 400 600 800 1000 1200 B ar rie r D ef le ct io n (ft ) End-anchor Spacing (ft) B ar rie r D ef le ct io n (m ) End-anchor Spacing (m) Figure 5.39. Deflection plots for Nucor cable barrier system.

analyses and results 85 5.3 End-Anchoring and Post-Anchoring Systems Adequate anchoring of the cables is critical to ensure satisfactory barrier performance dur- ing impact. Anchor movements lead to lower tension in the cables, which results in larger deflection of the system. The movement could also lead to sagging of the cables, which affects cable heights and consequently affects the barrier’s ability to engage the vehicle. Anchor movements are often attributed to weak soil conditions and the substandard-sized anchors. Weather conditions in certain regions of the country can also lead to anchor movement. Tem- perature decreases lead to higher cable tensions, which, in turn, apply higher forces on the anchors. Anchor pull-outs have been observed in several states, and many state DOTs consider this one of the most critical cable barrier issues. Figure 5.40 shows cases where the anchor has pulled out of the ground. A recent study conducted at the MwRSF investigated end-anchor movement due to dynamic impact and temperature variation loads [62]. The maximum dynamic impact load used in the analyses was obtained from a MASH Test 3-11 full-scale crash test where load cells were attached to the cables at the end-terminals. The maximum dynamic impact load was found to be 137 kN (31 kips). The maximum thermal load was calculated based on a temperature change of 130°F (from 110 to -20°F). The maximum load due to this change in temperature was computed to be 125 kN (28 kips). To develop recommendations for end-anchor sizes, the MwRSF researchers used the LPILE software with various sizes and soil strengths [63]. A load of 177 kN (40 kips) was used in the analysis. Figure 5.41 shows the predicted end-anchor movements for different anchor depths, Figure 5.40. End-anchor pull-outs. NCHRP 350 Soil Stiff Clay Sand Figure 5.41. Pile deflection for different soil types and end-anchor sizes.

86 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Table 5.1. Recommended end-anchor sizes from MwRSF study [62]. Force 12 o Figure 5.42. End-anchor computer model setup. anchor diameters, and soil strengths. Based on the assumption that a 50 mm (2 in.) end-anchor movement would lead to complete tension loss in the cables, a recommendation for anchor sizes was developed. Table 5.1 lists the recommended anchor sizes for the different soil conditions. The research indicated these recommendations are very conservative and should not be utilized to indicate the inadequacy of existing foundation designs. A similar analysis was conducted under this study using the LS-DYNA finite element pro- gram. Computer models of cable barrier end anchors with different sizes and geometry were created and subjected to a dynamic load of 140 kN (the 140 kN magnitude was extracted from the simulations performed under Task 5.2 of this study and previously described in Section 5.2). The finite element model setup is shown in Figure 5.42. The soil used in the analysis was based on NCHRP Report 350 strong soil. The soil model was calibrated based on pendulum tests and was used in previous studies and found to give good predictions of the NCHRP Report 350 soil response. Efforts were made under this study to investigate soils with different strengths but, due to lack of test data to calibrate these soil models, the analysis could not be performed. Consequently, the results from this study should be considered as less conservative than the MwRSF results and should be regarded as the minimum size for adequate cable barrier anchoring. The results from the analysis are shown in Table 5.2. (Readers are reminded that color ver- sions of figures and tables are available in the online version of the report which can be found by searching the TRB website for NCHRP Report 711.) End-anchor movement of more than 50 mm (2 in.), shaded in red in the table, are considered inadequate. Movements of 25 to 50 mm (1 to 2 in.), shaded in orange, are considered marginal. Movement less than 25 mm (1 in.), shaded in green, are considered acceptable. It is important to emphasize here that these results are for strong soil and the results do not account for weaker and saturated soils. Since soil types and conditions vary significantly for different site locations, end-anchor size should be determined on a case-by-case basis. Soil analysis, using similar approaches to the two methods presented above, should be conducted based on the soil data and climate information. Likewise, many designs use posts set in concrete foundations to facilitate removal and replacement of damaged posts. These foundations too must be sized properly based on existing soil and climate conditions so they are not damaged or pulled out of the ground in a crash.

analyses and results 87 5.4 Interconnection with Other Systems There have been accepted systems for the interconnection of cable barriers with strong post guardrail systems. These designs are reflected in FHWA acceptance letters B-147 and B-147A. The letters list several cable-to-W-beam transition designs as meeting NCHRP Report 350 test conditions at Test Level 3. The designs are accepted based on one full-scale crash test. The test was conducted using a South Dakota design wherein a U.S. generic 3-cable barrier was carried over and under a W-beam guardrail and anchored independently behind the metal beam rail. Subsequent to this testing, several of the manufacturers of proprietary cable systems proposed similar transitions, with the basic difference being the cables were attached directly to the W-beam rail element, thus eliminating the need for a separate anchor for the cables. Because the low-tension design performed adequately, there was no reason to suspect that the high-tension proprietary designs would not function as well (or better). Thus, these designs were accepted without full-scale testing. Figure 5.43 depicts a few of these cable barrier to W-beam guardrail transition designs. In the cable barrier to W-beam transition test, the end terminal of the W-beam guardrail was flared 1.22 m (4 ft) behind the cables. Figure 5.44 shows the vehicle behavior during the impact. As seen in the figure, the vehicle exhibited significant roll due to the impact with the end terminal. The vehicle did not roll over in the test and the transition met the NCHRP Report 350 requirements. Based on the results from this test, 1.22 m (4 ft) should be considered as the minimum flare of the end of the guardrail behind the barrier to avoid vehicle rollover. Another critical issue with cable barrier transitions is the force between the cable barrier and the system it is connected to (often a W-beam guardrail). It is important to ensure that the cable barrier static tension forces (due to temperature variations) and the impact forces do not lead to pull-out the W-beam barrier from its anchors or failure of the connections between the cables and the W-beam rail. This is especially critical for high-tension cable systems. The W-beam barrier must be long enough and adequately anchored at its downstream end to resist the tension in the cables. Under this study, simulations were conducted to identify the minimum length needed for strong post W-beam systems when connected to cable barriers. Sections of a G41S W-beam guardrail system were subjected to a longitudinal load of 140 kN (31 kips). This loading was obtained from the simulations performed under Task 5.2 of this study and previously described in Section 5.2. The section length was varied until a movement of the end-post (last post) of the W-beam barrier was less than 25 mm (1 in.). The simulation setup is shown in Figure 5.45. The end-post movement at different G41S section lengths is listed in Table 5.3. The minimum section length was found to be 22.9 m (75 ft). Anchor Movement In mm (in.) Anchor Diameter in m (ft) 0.3 (1) 0.6 (2) 0.9 (3) 1.2 (4) Anchor Depth in m (ft) 0.6 (2) >55 (2) >55 (2) >55 (2) >55 (2) 0.9 (3) >55 (2) 49 (1.93) 20 (0.79) 14 (0.55) 1.2 (4) 29 (1.14) 24 (0.94) 11 (0.43) 9 (0.31) 1.5 (5) 19 (0.75) 12 (0.47) 7 (0.28) 6 (0.24) 1.8 (6) 16 (0.63) 7 (0.28) 5 (0.20) 5 (0.20) 2.1 (7) 13 (0.51) 6 (0.24) 5 (0.20) 4 (0.16) 2.4 (8) 11 (0.47) 5 (0.20) 4 (0.16) 4 (0.16) Table 5.2. End-anchor movement for different foundation sizes.

(a) Generic (b) Brifen (c) CASS (d) Gibraltar Figure 5.43. Sample cable to W-beam barrier transitions [64]. Figure 5.44. Sequential images from cable barrier to W-beam transition. 88 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems

analyses and results 89 Similarly, the connection between each cable and W-beam rail should be designed such that it can withstand 90 kN (20 kips), the maximum load on the cables observed in the simulations under Section 5.2 of this study. In some installations, the transition was accomplished by placing the cable barrier behind the W-beam guardrail. Figure 5.46 shows a post-crash picture of one of these installations. The picture depicts a case where the vehicle hit the cable system and went between the two barriers. After rebounding from the cable barrier, the vehicle hit the back of the W-beam guardrail. It is therefore recommended that this type of transition should not be used until further analyses and/or testing is conducted to ensure adequate performance is achieved [64]. 5.5 Horizontal Curvature Limited research has been conducted on the performance of cable barrier systems on horizontal curves. Impacts on the convex side of a curved cable barrier will result in larger deflections due to the slackening of the cables that occurs when posts are removed and the cables follow the alignment of a chord rather than the arc. To investigate this effect for high-tension cable barrier systems, finite element simulations were performed. The models in this study were based on the high-tension, 4-cable CASS system used in the analysis described in Section 5.2. Computer models with varied curve radius and the post spacing were used in the analysis. Curve radii of 150 m (500 ft) (12° curvature), 300 m (1,000 ft) (6° curvature), 450 m (1,500 ft) (4° curvature), and straight alignment were incorporated in the models. For each curve radius, barriers with post spacings of 1.6 m (5 ft), 3.2 m (10 ft), and 6.4 m (20 ft) were created. A total of 12 simulations were performed for the four different horizontal curvatures and three different post spacings. A barrier length of 200 m (650 ft) was used, which provides sufficient length for the barrier to redirect the vehicle. The cable heights were set at 530 mm (21 in.), 640 mm (25.2 in.), 750 mm (29.5 in.) and 968 mm (38 in.). In all simulations, a 2000P Chevrolet C2500 pickup truck impacted the barrier at 100 km/h and 25° angle to the tangent of the curve. In all cases, the barrier was set up on flat, level terrain. Figure 5.47 shows the plan views for the three curves with one of the post spacings, 6.4 m (20 ft). The maximum deflections of the barrier in the simulations were determined by measuring the distance from the impact side of the vehicle to the original alignment of the barrier. Figure 5.49 shows the location of the initial impact and the deflections for the 150 m (500 ft) (12° curvature), 300 m (1,000 ft) (6º curvature), 450 m (1,500 ft) (4º curvature) simulations for 0.4, 0.6, and 0.8 second time intervals. The deflection results are shown in Figure 5.48. Force Figure 5.45. Simulation setup for G41S section subjected to a longitudinal load. Section length in m (ft) 17.4 (57) 19.2 (63) 21.0 (69) 22.9 (75) 24.7 (81) End-post movement in mm (in.) 30 (1.2) 27 (1.06) 26 (1.02) 24 (0.94) 22 (0.87) Table 5.3. W-beam movement for different section lengths.

90 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems Figure 5.46. Post-crash picture of overlapping cable barrier to W-beam guardrail transition. Figure 5.47. Plan views for the three convex curve simulations, 4-cable CASS system.

analyses and results 91 Figure 5.50 shows the percentage increase in deflection due to horizontal curvature for the three different post spacings. The base deflection is from the simulations using a straight alignment, i.e., zero degree curvature. The results were nearly identical for post spacings of 1.6 m (5 ft) and 3.2 m (10 ft), but going to a wider post spacing of 6.4 m (20 ft) shows a significant increase in deflection. For a curve with a 450 m radius (1,500 ft) (4° curvature), the increase in deflection is only about 10 percent for close post spacings but is 33 percent greater for post spacings of 6.4 m (20 ft). For this level of curvature, the maximum design speed is approximately 100 km/h (62 mph). For the sharpest curve simulated, 150 m radius (500 ft) (12° curvature), the deflection is 63 percent higher for a wide post spacing of 6.4 m (20 ft) and about 43 percent greater for the close post spacings. This level of curvature is associated with a maximum design speed of approximately 70 km/h (43 mph). These findings suggest that wide post spacings for cable barriers should not be used on horizontal curves where convex hits are possible and the curve radius is less than 400 m (1,300 ft) (degree of cur- vature greater than 4°). Even if adequate clear area is available, the greater deflection could adversely affect the barrier’s ability to capture and redirect impacting vehicles. Also, in median applications on sharp curves, placing barriers on each side of the median should be considered to reduce the likelihood of vehicles impacting the convex side of the barrier and intruding into on-coming traffic. 5.6 Installation Costs The selection of any roadway element should be made considering the associated life-cycle costs. Life-cycle costs include the costs of installation and costs associated with both routine periodic maintenance and occasional repair costs. The following sections describe the process and available data for life-cycle cost analyses for cable barrier deployments. 3 5 7 9 11 15 13 0 500 1000 1500 Straight Straight 1.5 2.0 1.0 2.5 3.0 5.0 4.0 4.5 3.5 0 100 200 300 400 500 B ar rie r D ef le ct io n (ft ) Horizontal Curvature Radius (ft) B ar rie r D ef le ct io n (m ) Horizontal Curvature Radius (m) 1.6 m (5 ft) Post Spacing 3.2 m (10 ft) Post Spacing 6.4 m (20 ft) Post Spacing Figure 5.48. Deflection plots for different horizontal curvatures for convex-side impacts.

92 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems 0.4 sec 0.6 sec 0.8 sec Figure 5.49. Sequential plots from convex-side impacts for different horizontal curvatures.

analyses and results 93 Figure 5.50. Influences of horizontal curvature and post spacing on deflection in convex-side impacts. Installation Costs Installation costs for cable barriers include the cost of the barrier and its end anchors as well as costs for modifications to the median or shoulder area. In some cases, the preparatory costs can exceed the cost of the barrier, particularly when major grading work is done and mow strips are constructed. This section of the report addresses costs for only the barrier and its end anchors. Installation costs for high-tension cable barriers vary widely from state to state. Some factors contributing to these variations could be associated with the length of the installation, cable barrier type, soil condition, and weather environment. Recent (2010) costs for TL3 systems in Texas average less than $5 per linear foot (LF, 0.3 m) but average costs in South Dakota in 2009 were over $15 per LF. An analysis of bid tabulations from Colorado for 2009 through 2011 showed that for 11 projects totaling 151,388 LF (29 mi, 46 km), the average and median winning bid for high-tension cable barriers was $11.69 per LF and $11.10 per LF, respectively. Among the 57 bids for these 11 projects, the unit bid price ranged from $6.10 to $27.00. Ignoring competitive factors, installation costs for high-tension barriers should depend on post spacing, anchor spacing, number of cables, soil conditions, and type of post foundation. The least expensive system would be a 3-cable system with driven posts, wide post spacing, and wide anchor spacing located in an area with very good soil conditions. The most expensive system would be a 4-cable system with posts in concrete foundations with short anchor spacing located in poor soil conditions. In general, lowering installation costs by increasing post and anchor spacing will be offset by decreased barrier performance. Barrier deflection during impact would likely be higher and the potential for barrier penetrations would possibly increase. Similarly, installation cost savings achieved by using driven posts instead of concrete foundations with inserts will be offset by increased impact repair costs. Life-cycle cost analysis provides a way to combine the effects of the tradeoffs mentioned above. For cable barrier systems, life-cycle costs include installation costs, maintenance costs, repair costs, and disposal costs at the end of the system’s useful life. The time value of money (i.e., discount rate) needs to be included in the analysis for it to be valid. Performance reductions and barrier failures also need to be included in the analysis. Data are not available to allow complete life-cycle analyses for the various cable barrier systems. Detailed cost and in-service performance data are needed to perform these analyses, and many

94 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems state DOTs have not collected this information or accumulated enough data to have meaningful averages. However, it is possible to provide information on some aspects of cable barrier costs that can be used to estimate life-cycle costs. Effects of Service Life and Discount Rate Service life of cable barriers is long, which reduces the impact of installation costs on total life-cycle costs. The longer a barrier is in use, the lower the average annual cost of installation will be since these costs are spread over more years. The discount rate determines how future costs are “discounted” so that the average annual costs reflect the time value of money and the opportunity costs associated with the initial investment. Higher discount rates increase the annual cost of installation costs. Service life for cable barriers can be very long if the roadway where they are installed is not modified in a way that requires the barrier to be moved or removed. For example, cable barriers in medians may have to be removed if the highway is widened by taking land from the median. The width of the modified median may be too narrow to accommodate cable barriers because of the relatively large dynamic deflections associated with flexible cable barriers. Shoulder improvements, pavement overlays, and alignment adjustments are other highway modifications that could shorten the service life of a cable barrier installation. Service lives of 5 to 50 years are used in the analysis to cover the expected range for cable barriers. The discount rate is primarily affected by interest rates, now and in the future. Interest rates in 2010 are at historic low levels because of the current economic situation. Since interest rates are effectively zero today, the only way they can move is up. During times of high inflation, such as the early 1980s, interest rates were in the high teens. Forecasting future interest rates is difficult, so analyses have been done on a wide range of interest rates from 0 to 14 percent, which should cover the likely range of future rates. As mentioned above, installation costs for cable barriers vary widely. To allow for the wide variation in costs, calculations have been done for a $1 unit installation cost per linear foot of barrier. Thus, to determine the equivalent average annual cost for an installation costing $10 per linear foot, the numbers in the Table 5.4 need to be multiplied by 10. Table 5.4 shows how significant the effect of service life and discount rate can be on the contribution of installation costs to total life-cycle costs. For a barrier that costs $1 per linear foot to install, the average annual cost for 1 mile of barrier can range from $106 for a service life of 50 years and a 0 percent discount rate to $1,538 for a 5-year service life and a discount rate of 14 percent. Annual cost ($ per mile) per $1.00 of installation cost per linear foot Discount Rate Service Life of Cable Barrier (years) 5 10 15 20 25 30 35 40 45 50 14% 1,538 1,012 860 797 768 754 747 743 741 740 12% 1,465 934 775 707 673 655 646 640 637 636 10% 1,393 859 694 620 582 560 547 540 535 533 8% 1,322 787 617 538 495 469 453 443 436 432 6% 1,253 717 544 460 413 384 364 351 342 335 4% 1,186 651 475 389 338 305 283 267 255 246 2% 1,120 588 411 323 270 236 211 193 179 168 0% 1,056 528 352 264 211 176 151 132 117 106 Table 5.4. Effect of service life and discount rate on life-cycle costs—installation costs.

Analyses and Results 95 The data in Table 5.4 is shown graphically in Figure 5.51. This figure shows that for higher discount rates the average annual cost is insensitive to changes in service life beyond 20 years. For discount rates in the “normal” range of 6 to 10 percent, insensitivity to service life occurs around 25 to 30 years. This analysis suggests that using a service life of 25 years for cable barrier projects is appropriate unless it is known that the service life of the system will be shorter because of planned or expected roadway modifications. Effects of Anchor Spacing on Installation Costs The spacing of anchors for high-tension cable systems varies widely from project to project. The impact of anchor spacing on barrier performance is discussed elsewhere in the report. Wider spacing of anchors reduces the cost of the system since fewer anchors are needed. If an anchor is destroyed in an impact, the longer the anchor spacing, the greater the length of barrier that is out of service until the anchor is replaced and the system is retensioned. Anchor cost varies widely because of different designs used by the manufacturers and because of soil condition variations. Undersized anchors have been a significant problem, particularly for states that have not required soil-specific designs. Several anchors have pulled out of the ground during periods of cold weather. Recent average TL3 anchor costs in Texas were $1,512 from December 2009 to February 2010 and $2,300 from October 2009 to December 2009. In 2009, TL3 anchors in South Dakota cost an average of $3,881. Part of the difference in costs between Texas and South Dakota can be attributed to the larger anchors required in northern climates. Stricter specifications in South Dakota may also be responsible for part of the higher costs. For the previously mentioned 11 recent projects in Colorado, the average and median win- ning bid for the 127 anchors in these projects was $2,718 and $2,500, respectively. In only 4 of the 11 projects was the unit price of the low bid (contract winner) the lowest price for the anchors. The overall average for the 57 anchor bids was $2,388 and individual bids ranged from $1,500 to $6,000. For the Colorado projects, the average spacing between anchors ranged from 74 m (241 ft) to 507 m (1,663 ft) with a median of 160 m (525 ft). One long 2010 project in Kentucky totaling 26.6 km (16.5 mi) of high-tension cable barrier used only 24 anchors, giving an average anchor spacing of 2,213 m (7,260 ft). The bid prices (9 contractors) averaged $2,621 per anchor and Figure 5.51. Effect of service life and discount rate on life-cycle costs—installation costs.

96 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems ranged from $2,000 (winning bidder) to $3,200. Two recent projects in Oklahoma had average anchor spacings of 2,500 m (8,200 ft) and 2,256 m (7,400 ft). Anchor costs were $3,900 for the first project and $2,000 for the other. To allow for the wide variation in anchor costs, a range of costs from $1,000 to $4,000 has been used in the anchor analysis. Using these anchor costs, the extra cost that the anchors add per linear foot of barrier is shown in Table 5.5 for anchor spacings of 61 to 6,100 m (200 to 20,000 ft). End-anchors can add a lot to the cost of a cable barrier system; but without adequate anchors, high-tension cable barrier systems are ineffective. The dynamic deflection from an impact increases with increased anchor spacing, and the increase in deflection varies for different cable barrier systems. Also, anchors that are undersized cost less, but can lead to failures of the system and costly repairs. Effective end anchors must be designed for site-specific soil and climate conditions. The values highlighted in the table represent typical values for many states and systems: anchor spacings from 300 to 1,500 m (1,000 to 5,000 ft) with anchor costs of $1,500 to $3,000. The impact on barrier installation cost ranges from minor for low-cost anchors separated by 1,500 m (5,000 ft) to major for higher-cost anchors separated by only 300 m (1,000 ft). There is a 10-fold difference in unit cost between these two conditions. Effects of Post Spacing on Installation Costs For most installations of high-tension cable barriers, post spacings vary between 10 and 20 ft. Post spacing narrower than 10 ft is used where deflections need to be low to avoid obstructions close behind the barrier. Post spacings greater than 20 ft have been used, but concerns about the increased risk of penetrations at wide post spacings have discouraged states from using very wide post spacings. Post costs depend on the cost of steel, the amount of steel in the post, the accessories (lock plates, hair pins, spacers, stiffeners, caps, etc.) required by the particular design, and the installation method (concrete foundation, steel sleeves, or driven). Most often these costs are included in the unit bid price for the barrier so that actual post costs are difficult to identify. Table 5.6 shows the impact of post spacing and post cost on cable barrier installation costs. The table covers a wide range of post costs from $20 to $90. These costs include the material costs for the post and foundation as well as the installation costs. Driven steel posts are the least expensive, and steel posts in concrete foundations designed for poor soil conditions in cold climates are the most expensive. Material costs for posts also vary among the manufacturers due to post size and design. The post spacings in Table 5.6 range from 1.2 m (4 ft), which is used only in special, low- deflection cases, to 6.4 m (20 ft) which is the upper limit of post spacings for most new installations. Common post spacings are 3.0 m (10 ft) and 4.9 m (16 ft). Extra Cost ($) per linear foot due to end anchors Anchor Spacing (ft) Cost per end anchor ($) 1,000 1,500 2,000 2,500 3,000 3,500 4,000 200 10.00 15.00 20.00 25.00 30.00 35.00 40.00 500 4.00 6.00 8.00 10.00 12.00 14.00 16.00 1,000 2.00 3.00 4.00 5.00 6.00 7.00 8.00 2,500 0.80 1.20 1.60 2.00 2.40 2.80 3.20 5,000 0.40 0.60 0.80 1.00 1.20 1.40 1.60 7,500 0.27 0.40 0.53 0.67 0.80 0.93 1.07 10,000 0.20 0.30 0.40 0.50 0.60 0.70 0.80 15,000 0.13 0.20 0.27 0.33 0.40 0.47 0.53 20,000 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Table 5.5. Effect of anchor spacing on installation cost.

analyses and results 97 Posts are a major component of overall barrier installation costs, but they are less variable than anchor costs. For the highlighted values in Table 5.6, which represent typical values, the highest value is 2.4 times the lowest value. However, in the case of typical anchor costs as shown in Table 5.5, the highest value is 10 times the lowest value. Coincidentally, the highest typical unit value for post costs is exactly the same as the highest typical cost for anchor costs indicating an equal contribution of post and anchor costs to installation cost. However, the lowest typical post cost is more than 4 times as high as the lowest typical anchor cost. Effects of Cable Costs on Installation Costs Cable cost is the most predictable of all of the cable barrier components. Galvanized wire rope used for the cables is a standard commodity; the price varies mostly with the cost of steel. Now that so much prestretched wire rope is being used for highway barriers, there is no significant difference in cost between prestretched and regular wire rope. Installed cost for cable is approximately $1.00 per linear foot per cable, plus or minus 20 percent. All high-tension cable barriers have either three cables or four cables. The 4-cable systems provide coverage for a wider range of vehicles for a small incremental increase in cost over 3-cable systems. For barriers with 4.9 m (16 ft) post spacing, the cable costs and post costs are approximately the same. For barriers with closer post spacings, the cable costs are less than the post costs. End anchor costs are usually less than the cable costs except when anchor spacing is less than 610 m (2,000 ft). Increasing anchor spacing is a way to reduce cable barrier installation cost, but it does have a significant effect on barrier performance. Cable barriers have typically been crash-tested at anchor spacings of approximately 100 m (328 ft) under NCHRP Report 350 guidelines and at 183 m (600 ft) under MASH guidelines. Research discussed in Section 5.2 indicates that an anchor spacing of 305 m (1,000 ft) will result in increases in deflection of approximately 25 percent from crash-test-reported values. Going to an anchor spacing of 1,524 m (5,000 ft) will result in increases in deflection of up to 50 percent from crash-test-reported values, depending on the cable barrier system. Increasing post spacing is another way to reduce installation costs, but deflection is increased, which affects barrier performance. Increases in deflection resulting from increases in post spacing appear to be independent of anchor spacing. Increasing the post spacing from 3.2 m (10 ft) to 4.8 m (15 ft) can lead to an increase in deflection of approximately 20 percent. An increase of post spacing from 3.2 m (10 ft) to 6.4 m (20 ft) can result in an increase in deflection of approximately 30 percent. Many states have opted for 3-cable systems instead of 4-cable systems for cost reasons. Also, it may not be clear to these states that 4-cable systems provide greater safety than 3-cable systems Cost ($) per linear foot due to posts and foundations Post Spacing (ft) Cost ($) per installed post and foundation 20 30 40 50 60 70 80 90 4 5.00 7.50 10.00 12.50 15.00 17.50 20.00 22.50 6 3.33 5.00 6.67 8.33 10.00 11.67 13.33 15.00 8 2.50 3.75 5.00 6.25 7.50 8.75 10.00 11.25 10 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 12 1.67 2.50 3.33 4.17 5.00 5.83 6.67 7.50 14 1.43 2.14 2.86 3.57 4.29 5.00 5.71 6.43 16 1.25 1.88 2.50 3.13 3.75 4.38 5.00 5.63 18 1.11 1.67 2.22 2.78 3.33 3.89 4.44 5.00 20 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 Table 5.6. Impact of post spacing and post cost on cable barrier installation costs.

98 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems for very little extra cost. Penetrations of barriers are caused when vehicles go under, through, or over cable barriers. By adding a fourth cable, a wider range of heights can be achieved, as well as a closer vertical spacing of cables to reduce the likelihood of penetrations. The inflation-adjusted 2010 societal cost of a highway fatality is estimated to be approximately $3.575 million by FHWA. The average annual cost of adding a fourth cable to a 3-cable barrier is $413 per (1.6 km) mile for a barrier with a 25-year service life and a 6 percent discount rate (see Table 5.4). Adding the fourth cable to 1,600 km (1,000 miles) of barrier has an average annual cost of $413,000. Thus, if the added cable would reduce the number of fatalities by 1 over this 1,600 km (1,000 mi) of highway over an 8-year period, it would offset the cost of the extra cable. The small incremental cost of adding a cable opens the possibility of designing barri- ers with more than four cables particularly when used on steep slopes within medians. As mentioned earlier in the report, the wide range of trajectories of vehicles on steep median slopes makes it difficult for a conventional 3-cable or 4-cable barrier to capture all vehicles. Adding one or two cables increases the range over which the barrier can capture vehicles, allowing the barrier to be located in more areas of the median and reducing the probability of penetrations. 5.7 Cable Barrier Maintenance: Tolerances, Repairs, and Systemwide Maintenance Tolerances: Construction and Maintenance Variations in cable barrier dimensions, especially cable heights, could affect the ability of the barrier to perform adequately. Additionally, variations in the terrain shape and slope could also affect barrier performance. The vehicle dynamics and finite element simulations performed under this study, as well as findings from the literature, were used to develop suggested tolerances for cable barrier installations. Table 5.7 lists these tolerances. The tolerances were established such that the variations within the suggested limits should not have significant effects on cable barrier performance. Based on vehicle dynamics analyses, increasing the height of the lowest cable or decreasing the height of the highest cable could adversely affect the barrier’s ability to engage a vehicle and lead to underrides or overrides. Thus, a zero tolerance is suggested for the lowest and highest cables for the critical side (upper for lowest cable and lower for highest cable). Cable heights affect a barrier’s ability to engage and capture an impacting vehicle so tolerances must be tight to insure that the barrier will perform as designed. For the middle cables and non-critical side of the top and bottom cables, a tolerance of 25 mm (1 in.) is suggested. This tolerance may be difficult to achieve on vertical curves without having post connections that resist vertical movements of the cables. In particular, extra care is needed around drainage inlets to insure that the cables have the correct heights to reduce the probability of underrides. Cable Barrier Parameter Tolerance Top Cable Height -0,+25 mm (-0, +1 in) Bottom Cable Height -25,+0 mm (-1, +0 in) Middle Cable(s) ±25 mm (±1 in) Barrier Lateral Position ±150 mm (±6 in) Average Post Spacing ±150 mm (±6 in) Consecutive Post Spacing ±600 mm (±2 ft) Cable Tension ±2 kN (±0.45 kips) Table 5.7. Suggested cable barrier tolerances.

analyses and results 99 The lateral position of the barrier relative to the roadway or median centerline is important for several reasons. These reasons were explained in detail in Section 5.1. Small deviations from the design alignment may be needed to accommodate issues in the field such as imbedded rocks, drainage facilities, and other construction constraints. The suggested tolerance for lateral position is 150 mm (6 in.), which should never be exceeded if the barrier is being placed near the centerline of the median because of the sensitivity of barrier performance to location as discussed in Section 5.1. Post spacing is less critical since it primarily affects the magnitude of deflection and not the ability to engage a vehicle unless the post spacing is very wide (wider than the maximum spacing used for the crash tests). Most cable barriers have been crash-tested and accepted for more than one post spacing, confirming that barriers can perform adequately at varied post spacings. The suggested tolerance for consecutive post spacings is 600 mm (2 ft). This tolerance should provide enough flexibility to account for field obstacles that prevent a post from being installed in the exact location specified on the plans. The suggested tolerance on average post spacing, in the region free of obstacles, is 150 mm (6 in.). Cable tension makes cable barriers work, particularly for high-tension systems. However, cable tension changes with temperature, increasing in cold weather and decreasing in warm weather due to thermal contraction/expansion. Simulations of cable barriers at different tensions were discussed in Section 5.2. Minor changes in tension do not have significant effects on deflection. Lower tensions allow some increase in deflection, but higher tensions exert higher static loads on anchors, particularly during cold weather. Because of the day-to-day variations in tension due to climatic changes, specifying a tight tolerance on tension would increase the need for barrier maintenance without a commensurate increase in safety. Therefore, the suggested tolerance for cable barrier tension (as measured against the design tension for the given cable temperature) is 2 kN (450 lb). Cable Barrier Repairs Most maintenance costs associated with cable barriers result from crash damage. Since cable barriers are often placed where they can be hit frequently, these costs can be significant. In police-reported crashes, it is usually possible to get the offending driver’s insurance company to reimburse the highway agency for these costs. But, because cable barriers are so “forgiving,” drivers often are able to drive away after a crash, which makes it difficult to collect from an insurance company. Data from a few states indicates that police reports are usually available for slightly more than half of the crashes. However, non-reported crashes are less severe, which means their repair costs will be lower. Kentucky found that non-reported crashes had approximately half as many damaged posts as the more severe crashes where a police report was filed. Frequent inspections are needed to identify crash damage, but these inspections can be accomplished by highway agency and police personnel reporting damage they observe as they drive by the barriers during their normal work activities. In addition, periodic inspections should be conducted to identify unreported damage. In the state survey conducted for this study, the frequency of inspections ranged from daily to annually. Previous crash history can be used to determine how frequently inspections should be conducted for each highway. Crash frequency depends on the average daily traffic (ADT), speed limit, location of the barrier relative to the edge of pavement, and weather conditions. In most crashes, only the posts are damaged. It is rare for the cable to be damaged unless a vehicle gets tangled in the cables or the cables are cut by emergency response personnel. In the majority of crashes, the cables can be reinstalled on the replacement posts without the need to retension. If enough posts have been destroyed to cause the cables to be on the ground, then the

100 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems tension in the repaired system should be checked. If any of the individual cable wires has been broken or if significant kinks in the cable are visible, the cable should be evaluated to determine if it needs to be replaced. In the rare case of an extremely large deflection caused by a heavy vehicle impact, the cable should be examined to make sure that plastic deformation of the cable has not occurred. The number of posts damaged in a crash depends on a number of factors including the velocity and mass of the impacting vehicle, impact angle, post spacing, cable tension, barrier type, and vehicle-barrier interactions. Most states have found that the average number of posts damaged in a crash is between four and eight. A few crashes with tractor trailers have resulted in more than 50 posts being damaged. The cost to repair posts depends on a number of factors. Driven posts are more expensive to replace than posts in socketed concrete foundations. Post repairs in winter can be more expensive if the posts are frozen in the sockets. Little data is available on the difference in cost of using driven posts rather than socketed posts. Many states now use socketed posts routinely because of the ease of post replacement after crashes. Ohio, one of the first states to install high-tension cable barriers, purposely installed some of each kind of post to be able to observe the differences. Shortly into the program they began to replace all damaged driven posts with socketed posts because of the difficulty associated with repairing driven posts. Data from Oklahoma shows an installation cost savings of approximately $3 per LF for driven posts. For a 25-year service life and a 6 percent discount factor this savings is about $1,240 per 1.6 km (1.0 mi) per year. Texas data shows an average of about 6 crashes per 1.6 km (1 mile) per year on their highways with cable barriers while 2007 data from Oklahoma shows an average of about 14 crashes per 1.6 km (1.0 mi). Using the more conservative Texas crash rate, the savings in cost for driven posts is approximately $200 per crash. In Oklahoma, the savings is only $90 per crash. Repairs for driven posts require post-driving equipment, which often requires a lane closure. The extra cost for driven-post repairs will easily exceed the $200 saving per crash. Thus, the added cost of including posts with socketed concrete foundations may be justified except possibly for very lightly traveled roads where the expected crash rates are low. Repair costs, on a per-post basis, are useful for forecasting future repair costs. Some states have collected good information on repair costs for cable barriers at least for the first couple of years of service. In the state survey, responders were asked to give the average number of posts damaged per impact and the average cost per impact. The 18 responses for cost per impact fell into two groups of nine, the first being rather precise numbers ranging from $240 to $870 and the other having what appeared to be estimates with all numbers rounded to the closest $100 and most rounded to the closest $1,000. These values ranged from $1,000 to $10,000. It appears that the numbers in the first group, which averaged $520 per crash, were probably based on studies while the numbers in the second group, which averaged $2,320 per crash, were rough estimates. Almost all the state in-service performance studies, primarily on the Brifen system, provided results similar to the first group. For example, Oklahoma DOT kept track of the cost of each of the crashes on the Lake Hefner Parkway during its first year the cable barriers were in service, the first high-tension cable barrier installation in the United States. This 4-cable system experienced 126 crashes which, on average, cost $284 (2002 dollars) to repair an average of 4.7 posts per impact. For this system, the average repair cost per post, including traffic control and all other repair costs, was $60. Repair costs depend partly on how long it takes a repair vehicle to get to the scene. The Lake Hefner Parkway represents a “best case” situation since it is only 11 km (7 mi) long and is located in Oklahoma City close to response units. In rural areas, crash locations may be more than an hour’s driving distance from the nearest response unit, which will make repairs more expensive.

analyses and results 101 The lower repair costs appear to be mostly associated with repairs done by highway agency crews rather than by private contractors. It is possible that some of the states reporting costs included only materials costs and not labor costs. Minnesota reported a total repair cost of $58,500 for 45 crashes and gave a breakdown of the cost for labor, material, and equipment. The average $1,300 cost per crash included $631 for materials, $478 for labor, and $191 for equipment. A major factor in cable repair costs is the price the highway agency pays for replacement posts. Since the high-tension cable barrier systems are proprietary, once the system is installed the manufacturers have control over the price of replacement posts unless the highway agency has included replacement posts in the initial (or follow-up) contract. Since it is known that a significant number of posts will be damaged in impacts, agencies should consider including a sufficient number of replacement posts in the original construction contract. Typically six posts are damaged in an average impact into a high-tension cable barrier. If posts are socketed in concrete foundations, their repair is usually quick and relatively inexpensive. Often, the repairs can be done in less than an hour without traffic control. Repair costs depend on the extent of the damage, the remoteness of the crash location, and whether highway agency crews or a contractor does the repairs. Systemwide Maintenance On-going, maintenance issues for cable barriers are discussed separately from crash-related repairs since they are systemwide issues rather than site-specific crash-related needs. On-going issues include inspecting and maintaining specified cable heights and correct cable tension, as well as ensuring proper functioning of cable connectors and end anchors. Line posts and their foundations may occasionally need maintenance. Determining if maintenance is needed requires that the cable barriers be inspected. Modern high-tension cable barriers require little on-going maintenance if they have been designed and installed correctly. Problems can occur when end anchors and post foundations have not been designed for in situ soil and local climate conditions. Under-designed anchors and foundations often rotate out of the ground or move enough to cause reductions in cable tension. Inspections are needed to identify these problems as well as to locate damage to the barrier caused by non- reported crashes. Annual inspections to identify non-crash-related issues should be sufficient unless design flaws in anchors and post foundations are known to exist or if extreme weather events have occurred. Flooding can cause erosion, which will affect the heights of the cables, and very cold weather will exert heavy static loads on the anchors, which could cause anchor movement. If non-prestretched cables were used when the barrier was installed, cable tensions should be checked and adjusted every 6 months for the first several years. Once the wires in the cables have seated themselves, semi-annual retensioning should not be needed. Cable heights should be checked at the bottom and top of vertical curves to make sure the cables have not moved up or down from their specified heights. On sharp horizontal curves, cable heights can be affected by leaning line posts. Cable heights are also affected by discontinuities in the ground under the cables. Erosion, winter maintenance, crashes, and other activities around the cable barriers can cause ruts or mounds that will affect the effective height of the cables. These discontinuities should be removed if they cause the cable heights to be out of tolerance. Mow strips, as described below, can be effective in minimizing these problems. Maintaining adequate tension in cable barriers is important particularly for high-tension systems. Cable tension is affected by temperature changes. When the temperature drops, the cable contracts, which increases its tension. Increases in temperature cause the cable to expand,

102 Guidance for the Selection, Use, and Maintenance of Cable Barrier Systems which lowers its tension. Each cable barrier manufacturer has a temperature-tension table that gives the required tension as a function of cable (not air) temperature. Most manufacturers also sell tension meters to measure cable tension. Friction between the cables and posts, particularly for interwoven systems, can delay the migration of tension changes along the barrier. After retensioning, the cable tension should be checked at several different locations along the barrier to make sure the tension has equalized and that the system has the correct tension. As discussed above, a suggested tolerance for tension is ±2 kN (±0.45 kips) from the specified tension given in the manufacturer’s table. High tensions resulting from cold weather could cause improperly manufactured/installed cable connections to fail. Strength specifications and testing requirements for turn buckles, swaged-on fittings, and other cable connectors should be included in contract documents. For example, Kansas requires a minimum tensile strength of 164 kN (36.8 kips). The strength of the connection should be at least as strong as the breaking strength of the cable. Mow strips are used by some states to reduce the cost of mowing grass around the cable barriers and to strengthen the post and anchor foundations. Mow strips are expensive and can cost as much or more than the cable barrier itself. In the survey of states, 15 indicated that they use mow strips and 19 indicated they did not. Costs ranged from $6 to $25 per LF (0.3 m). Mow strips can also strengthen the embedment of line post foundations and end anchors, which lessens the chance for anchor/foundation failure. Mow strips also reduce the discontinuities in the natural ground, which should help to maintain proper cable heights. It is difficult to estimate the monetary value of the strengthening of the posts and foundations and reducing ground discontinuities, but they are significant. Table 5.8 shows how many hours of labor need to be saved each year for 1 mile of highway to justify the addition of a mow strip solely on grass cutting benefits. This table is based on a 25-year life for the mow strip and a discount rate of 6 percent. For example, if a mow strip costs $15 per LF and labor and equipment costs for grass mowing is $40 per hour, to justify a mow strip, a total of 155 hours of labor would have to be saved each year for each mile of highway where the mow strip is installed. These hours should be reduced by anticipated benefits from the strengthening of the posts and anchors and reducing ground discontinuities. 25-Year Service Life and 6 percent Discount Rate Labor & Equip Cost per Hour Cost of Mow Strip per Linear Foot ($) 5 10 15 20 25 $20 103 207 310 413 516 $30 69 138 207 275 344 $40 52 103 155 207 258 $50 41 83 124 165 207 $60 34 69 103 138 172 Table 5.8. Hours of labor savings/mile/year required to justify mow strip.