Criteria for Restoration of Longitudinal Barriers, Phase II (2021)

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.

28 CHAPTER 3 – LITERATURE REVIEW The literature search conducted in Report 656 focused primarily on published guidelines for assessment of barrier damage and repair. That study effort was relatively comprehensive and thus is not repeated here. The literature review herein was focused primarily on the many studies related to analysis and design/redesign of guardrail systems that have been conducted in the past two decades. From this general area of research, the literature to be searched, reviewed, and summarized is extensive; however, a great deal of the analysis work from these studies remains unpublished - that is, the final reports and published papers generally only include performance data related to final designs and not the analysis and testing details that lead up to the final design. In an attempt to gather some of the unpublished data, the roadside safety research community was asked to provide information or references involving parametric studies of guardrail system component design and analysis. Unfortunately, these data are often incomplete, documented in private notebooks, or deemed “fully privileged” (e.g., proprietary products, NCHRP quarterly reports, etc.); and researchers are generally not privileged to, or willing to, share the information. However, even without the complete detailed analysis data, the published reports provide a great deal of information about how the various elements of a barrier system function as well as their respective role in overall system performance, and were thus useful in identifying:  The guardrail types to be considered in the study,  The critical damage modes associated with each guardrail type,  The combination(s) of damage modes that are considered most likely to be characteristic of damaged systems, and  Which of those combinations are considered most critical to system performance. General Crash Phenomena that Lead to Poor Performance There are, in general, four crash phenomena that lead to poor performance of strong-post guardrail: (1) pocketing, (2) wheel snag, (3) barrier override/underride, and (4) rail rupture. These phenomena are described below with a brief explanation of their causes and effects. Pocketing Pocketing is the crash event in which the impacting vehicle causes a large deflection in a relatively low-stiffness section of a guardrail upstream and adjacent to a relatively high-stiffness section of the guardrail, which forms a “pocket” shape in the rail. For strong-post w-beam guardrails, this phenomenon occurs when the w-beam rail experiences relatively high lateral deflections in the span between guardrail posts, as shown in Figure 30. As the vehicle approaches the downstream end of the pocket, the rail element tends to bend around the post. This generally causes high stress concentrations in the rail element at the post location, particularly when a splice connection exists at that post, and often results in rupture of the rail element. Pocketing can be alleviated by making the stiffness along the guardrail more uniform. For example, by reducing post stiffness (e.g., weak-post guardrail systems) the downstream posts

29 will more readily deflect as the vehicle approaches. In this case the post stiffness is reduced to better match that of the unsupported section of rail. However, reducing post stiffness will generally result in greater system deflections, which may not be appropriate in many situations. Figure 30. Example of pocketing during a simulated impact with a w-beam median barrier. [Fang10] Another means of reducing the effects of pocketing is to decrease the spacing between posts. As the posts are placed closer together, the unsupported length of the w-beam becomes shorter and, consequently, more stiff. In this case the stiffness of the unsupported section of rail is increased to better match that of the stiff guardrail posts. However, the added increase in system cost (i.e., additional materials and construction costs), is generally not feasible for most guardrail applications. Wheel Snag Wheel snagging is a crash event in which the tire or rim of the vehicle directly impacts against a relatively stiff component of the system, usually a guardrail post, causing high vehicle decelerations and extensive damage to the wheel assembly. Increased vehicle decelerations are directly associated with increased occupant impact forces and decelerations; while a damaged wheel assembly is often times attributed to causing vehicle instability and rollover during redirection. Wheel snag may be caused by several factors, such as pocketing, underride of the rail element, shallow blockout depth or collapsed blockout. Wheel snag is seldom preventable in high speed, high angle impacts with standard strong-post guardrail; however, its effects are often reduced when the contact area between the vehicle and post is minimized and/or the capacity of the post (e.g., failure load and energy absorption) is reduced with respect to the longitudinal direction of the guardrail. Barrier Override Barrier override, as the name implies, is a crash event in which the impacting vehicle vaults over the barrier, exposing the vehicle to the hazards behind the barrier. Barrier override may be caused by many factors, such as low mounting height of the rail element, improper release of rail-to-post connection, pocketing, and ramping of a vehicle tire on a guardrail post, to name a few.

30 The top of the rail element should be high enough to prevent the front bumper of an impacting vehicle from going over the top of the rail; this is particularly true for the case of pickup trucks and SUV’s where, once the bumper goes over the rail, the rail will be in direct contact with the vehicle tire. The spinning tire and the friction between the tire and the rail are often sufficient to push the rail down, allowing the vehicle to override the system. The rail-to-post connection is also a critical aspect of a guardrail system. The post-rail connection must fail consistently and at the appropriate time to prevent the rail from being pulled down with the post during impact.[Engstrand00; Ray01a] As discussed in the preceding paragraph, if the rail drops enough to allow the front bumper of the vehicle to override, then there is an increased chance for the vehicle to override the barrier.[Bligh97a; Engstrand00; Ray01a; Gabler10] As will be discussed later, increased blockout depth can help alleviate the sensitivity of the rail-to-post release on system performance. Rail Rupture Rail rupture refers to the event of complete rupture or tearing of the metal guardrail component which allows the vehicle to penetrate the system, thereby exposing the vehicle to hazards behind the barrier. An additional hazard associated with rail rupture is subsequent vehicle rollover resulting from the “loose end” of the rail wrapping around a tire and tripping the vehicle or penetrating the occupant compartment of the vehicle.[Buth99a; Buth00a; Bullard10] One of the leading causes of rail rupture for most standard strong-post guardrail is pocketing and subsequent failure of the splice connection.[Bullard10; Buth99a; Buth06; Mak99b; Polivka99; Ross99]. It has been determined, based in part on the results of tensile tests on specimens taken from the failed w-beam, that guardrail rupture is not usually caused by defective or substandard material; but rather due to stress concentrations in the splice-bolt holes as the rail splice is bent around guardrail posts during impact.[Engstrand00; Polivka00a; Ray01a] Another cause of rail rupture is small tears in the rail that occur when the rail is directly exposed to other guardrail components with sharp edges, such as steel wide-flange posts when no blockouts are used (e.g., weak post guardrails) or when steel blockouts are used. [Engstrand00; Ray01a] Regardless of how the tear initiates, the tension in the rail element may cause the tear to propagate and lead to rail rupture. The various phenomena leading to rail rupture are discussed in more detail later in this review. General Functions and Influences of Guardrail Components The general function and influences of the various guardrail components are presented and discussed in this section. There are several studies that provide information directly related to the sensitivity of various components, while others provide supporting information from full- scale testing, component testing or numerical analyses. The discussion presented herein references specific studies which directly or indirectly support the conclusions with only brief summaries of the studies. Posts and Soil Strong-Axis Loading Sometimes, essentially identical full-scale tests can result in dramatically different outcomes.[Strybos97] One of the causes of these variations could be the result of varying post

31 strengths and soil conditions.[Patzner97; Patzner99] The response of the post has a significant effect on the energy absorption of the post-soil system in guardrail impacts. For example, typical deformation modes of a post-soil system include post rotation in the soil, fracture of the post, post bending, post twisting or a combination of these modes. The maximum energy absorption of the post-soil system occurs when all deformation occurs in the soil (i.e., post rotates through the soil with no post deformation or failure); therefore, the stiffness of the post should be greater than the stiffness of the soil for optimal performance.[Reid97] When evaluating guardrail strengthening techniques, Rossen et al. determined via computer simulation that, although dynamic deflection was consistently higher for guardrail systems using the steel W6x8.5 posts than for those with the 6x8-inch wood posts, the difference was relatively small compared to the differences in deflection arising from variation in soil type and soil density.[Rossen96] Patzner, Plaxico and Ray while at the University of Iowa conducted a study for the Federal Highway Administration to investigate the effects of post and soil strengths on the performance of a common generic guardrail terminal called the Modified Eccentric Loader Breakaway Cable Terminal (MELT), shown in Figure 31.[Patzner97; Patzner99] The MELT was tested at the Southwest Research Institute in San Antonio, Texas in 1997 and it was found that the system would meet performance criteria for NCHRP Report 350 Test 3-35 only after optimizing the post and soil conditions (i.e., “Grade 1 structural grade lumber for the posts and dry compacted strong soil”).[Mayer99] Although the current study is mainly concerned with strong-post guardrail systems, guardrail terminals such as the MELT are designed to function like a guardrail when struck along the side downstream of the end. For example, NCHRP Test 3-35 involves the 4,409-lb pickup impacting on the side of the guardrail terminal at 62.1 mph and 20 degrees. The posts used in the MELT are designed to break with relatively low forces when impacted in the weak direction (i.e., end-on impacts where it is desired to have the posts shear at the groundline as the vehicle passes over them) but are designed to function as guardrail line-posts when loaded in the strong direction (e.g., vehicle impact on the face of the w-beam rail). Patzner, et al. used finite element analysis to conduct a parametric study to examine the effects of post and soil strength on the overall performance of the MELT under impact conditions corresponding to Report 350 Test 3-35. The study included a matrix of twelve cases varying only post strength and soil conditions. Three post strengths and four soil strengths were investigated. The three post strengths corresponded to a Grade 1 Dense, Grade 2, and a Grade 2 with 30% strength reduction. The four soil strengths were modeled using properties of a granular soil (e.g., similar to the soil type used in the full-scale crash tests) with four different moisture conditions of 50, 75, 87.5 and 100 percent saturation.

32 Figure 31. Modified Eccentric Loader Breakaway Cable Terminal. [Patzner99] The results of the analyses showed that certain combinations of soil and post strengths increase the hazardous possibilities of wheel snagging, pocketing, or rail penetration and they identified conditions that would maximize the safety and reliability of the guardrail terminal system. In particular, the analyses indicated that the potential for pocketing tended to increase as soil stiffness increased and as post strength decreased. [Patzner99] In such cases, the premature failure of the posts resulted in significant reduction in stiffness of the unsupported section of rail leading up to the next downstream post, thereby increasing the potential for pocketing. The results from the study were also evidenced in full-scale tests. “In Test MLT-3 Grade No. 2 posts were installed in a very stiff soil. The posts in this test had been used in an earlier full-scale test and it was believed that some of the posts may have been damaged. This test resulted in several posts breaking and guardrail rupture. Test MLT-4, in which Grade No. 1 Dense posts were mounted in very wet soil conditions, resulted in wheel snag which caused the vehicle to roll over. In Test MLT-5 the Grade No.1 Dense posts were mounted in very stiff soil at near optimum moisture conditions and the test was successful.”[Mayer99] Hascall et al. performed a study for the U.S. Department of Agriculture to design and test a wood-post guardrail system for Report 350 TL-3 application.[Hascall07] The primary objective of the project was to determine the acceptable diameter and grading for three species of wood posts for use in the MGS guardrail system. A series of static and dynamic bogie impact testing was carried out on 8-, 10- and 12-inch diameter wood posts made from Douglas Fir, Ponderosa Pine, and Southern Yellow Pine to determine appropriate post diameter and length. Two full-scale crash tests were performed to evaluate system performance according to the testing and evaluation requirements of NCHRP Report 350 Test 3-11. Test MGSDF-1 used guardrail posts made from 7.25-inch diameter Douglas Fir, and Test MGSDF-2 used guardrail posts made from 8-inch diameter Ponderosa Pine. The test results showed that the MGS functioned adequately with posts made from either species. Although a full-scale test was not performed for the guardrail system with posts made from Southern Yellow Pine, the 7.5-inch diameter Southern Yellow Pine posts were also determined to be an acceptable for use in the

33 MGS system, based on comparable strength of the Southern Yellow Pine relative to the other two species. Another very useful result of the Hascall et al. study was the extensive literature review on wood-post testing and the comparison of the performance between wood and steel posts. The reader is referred to [Hascall07] for the complete review. A few of the relevant studies from that review are summarized here:  Jeyapalan et al. concluded, based on static testing of posts with embedment depths of 38 inches, that the 7-inch diameter Southern Pine posts had similar peak force and energy dissipation to that of the W6x8.5 steel posts regardless of soil type. In the dynamic tests, the peak force and energy dissipation for the 7-inch diameter Southern Yellow Pine post were lower, but within 10% of the values for the W6x8.5 steel posts installed in cohesive soil; dynamic response comparisons in noncohesive soils, however, could not be made since the wooden post fractured almost immediately upon impact. [Jeyapalan84]  Bronstad et al. conducted a study in 1988 involving pendulum tests on both wood and steel posts. The tests involved various sizes of wood posts with embedment depth of 36 inches and two types of steel posts (i.e., W6x8.5 and W6x15.5) with embedment depth of 44 inches. The results of those tests, as shown in Table 5, showed that the W6x8.5 steel posts at 44-inch embedment provided 1.6 times more stiffness in the strong axis than the 6x8-inch wood posts at 36-inch embedment; whereas the wood post provided 1.7 times more stiffness than the steel post in the weak direction.[Bronstad88]  Rohde and Reid conducted a study regarding grading specifications and requirements for wood posts used in w-beam guardrail systems, and concluded that there was no significant benefit to using Grade 1 over Grade 2 posts, and that “use of the Grade 2 posts would lower cost of guardrail installations without adversely impacting its safety performance”. [Rhode95]  Coon et al. showed that when wooden posts fractured the energy absorption was significantly lower than when the posts rotated in the soil.[Coon99] Hascall et al. concluded from their review of pendulum impact tests on guardrail posts embedded in soil that the response of the 6x8-inch wood post performed similarly to the W6x8.5 steel post when loaded in the strong direction.[Hascall07] Full-scale crash tests seem to indicate, however, that although performance is similar between these two systems, all other system components being the same, the steel-post systems provide lower impact forces in full-scale tests. For example, as shown in Table 5, tests on the modified G4(1S) with wood blockouts resulted in longitudinal Occupant Ridedown Accelerations (ORA) values of 7.9 G and 7.6 G [Bullard96; Bligh97]; while tests on the G4(2W) resulted in longitudinal ORA values of 10.2 G, 10.9 G and 11.6 G.[Bullard09; Bligh95; Mak99a]

34 Table 5. Pendulum tests on wood and steel posts. [Bronstad88; Hascall07] Weak-Axis Loading Posts with high bending stiffness about the weak-axis (i.e., loading of the posts in the longitudinal direction with respect to the guardrail system) can also result in severe wheel snags when contact between the tire and posts occurs. The steel W6x9 posts used in the modified G4(1S) and MGS guardrail systems, as well as the 6x8-inch wood posts used in the G4(2W) guardrail system, tend to have appropriate strength in the weak-direction to alleviate severe wheel snags during impacts with pickups (i.e., in Test 3-11). The steel posts tend to bend over when struck in the weak direction at high speeds by heavier passenger vehicles, while the wood posts tend to rupture in such cases. [Bullard96; Mak99a] On the other hand, when these posts are directly impacted by small and midsize passenger cars (e.g., Report 350 and Mash 3-10 test vehicles), the effects of the wheel snag are more pronounced, and in some cases result in relatively high vehicle decelerations and occupant compartment intrusion.[Ray04] This effect is magnified as the overlap between the posts and vehicle wheel increases, which will be discussed in more detail in the next section on blockouts. Karlsson investigated the use of failure fuses on timber guardrail posts to reduce the impact forces when vehicles directly impact against the side of the posts during collisions with strong-post guardrail systems.[Karlsson00] Karlsson’s study included evaluation of a 6x8-inch wood post that was modified by drilling a 3.5-inch diameter hole through the side of the post at approximately two-inches above the groundline, as illustrated in Figure 32. The study was carried out using finite element analysis, where it was shown that the post provided

35 approximately the same stiffness and load carrying ability as the standard 6x8-inch wood post when loaded in the strong direction (i.e., impacts on front of post), but would fail at 60% of the impulse required to fail the standard post in the weak direction (i.e., impacts on the side of the posts). The same basic technique has been widely used in the design of the breakaway mechanism in guardrail terminals for decades. Figure 32. Illustration of wood post with failure fuse. An existing post type, the CRT post, was also investigated in Karlsson’s study. The CRT post includes two 3.5-inch diameter holes, one at the ground line and another at 15.75 inches below ground. This post is used in most guardrail terminals (e.g., MELT, FLEAT, SKT, etc.) as well as in the MGS long-span guardrail. It was determined that the CRT post did not possess adequate strength in the strong direction for application in a length-of-need section of guardrail with standard post spacing of 75 inches. The weakened post could lead to excessive lateral deflections. The modified post concept was never evaluated in full-scale crash tests for a standard length-of-need section of guardrail. Embedment Depth The embedment depth of guardrail posts must be optimized to achieve the desired resistance and energy absorption from the post-soil system. The stiffness of the soil increases as embedment depth increases. Recall that pocketing can be alleviated by ensuring that the stiffness along the guardrail is more or less uniform. Thus, if the stiffness of the post-soil system greatly exceeds that of the unsupported section of rail, pocketing may result. When the soil stiffness exceeds that of the post, the deformation of the post-soil system is primarily due to bending of the post (e.g., steel posts) or rupture of the post (e.g., wood posts) with minimal soil displacement. This tends to (1) cause posts in the immediate impact area to yield or fail prematurely, (2) decrease deflection of the posts located immediately down-stream of the impact, due to the increased rigidity of the post-soil system, (3) reduce the amount of 2 inches Drilled hole

36 energy dissipation absorbed by the post-soil system, and (4) cause the posts to rotate about a point nearer to ground level. Issues (1) through (3) may result in pocketing due to an increase in relative deflection between the unsupported section of rail and the downstream post. Issue (4) results in an increased potential for the rail to be pulled down with the posts as the posts rotate about a shorter radius. In this case, the release of the rail-to-post connection becomes more critical. Kuipers and Reid conducted a series of bogie impact tests on W6x16 steel guardrail posts with embedment depths ranging from 34 to 43 inches. [Kuipers03] The purpose of the tests was to determine the appropriate embedment depth for the posts in the MGS guardrail system. They determined that embedment depths less than 40 inches resulted in the posts being pulled out of the ground during impact with little or no post deformation, while embedment depths equal to or greater than 40 inches resulted in higher levels of soil failure with slight deformations of the posts. In the development of the MGS long-span guardrail system, Bielenberg et al. used a reduced embedment depth for the posts at the up- and down-stream ends of the long-span section of the guardrail system.[Bielenberg07] The long-span system was developed to shield culvert headwalls and was designed to accommodate up to 25 feet of unsupported w-beam rail. The unsupported length is composed of a single layer of w-beam rail, which may experience relatively large deflections during collisions. The reduced embedment depth of the posts adjacent to this unsupported section prevents pocketing by acting as a transition from the low-stiffness of the long-span section to the high-stiffness of the standard MGS guardrail section. This transition allows deflections to be spread over longer distances of the guardrail, creating a more effective load distribution along the rail, which reduces the rail force and rail strain. Increased Post Spacing Mak and other researchers at TTI performed a study for the Washington State Department of Transportation to crash test and evaluate a strong-post guardrail system with 12.5- ft post spacing under Report 350 TL-2 impact conditions. At the time of the study, Washington state was using TL-3 guardrail systems for both low-speed and high-speed facilities. If successful, the modified strong-post guardrail with reduced post-spacing would provide a more cost-effective alternative for installations on low-volume, low-speed roadways. The test installation consisted of a 100-foot long section of guardrail and two 25-foot long end-terminals. The guardrail was a standard G4(2W) with modified post spacing of 12.5 feet (double post spacing). The wooden posts were 6 inch x 8 inch in cross-section and 6 feet long. The wooden blockouts were also 6 inch x 8 inch cross-section and were 14 inches long. The w-beam rail was attached to the posts with standard 20-inch long, 5/8-inch diameter bolts. The mounting height of the guardrail was 27-inches. The test resulted in the vehicle vaulting over the barrier, as shown in Figure 33 (taken from the test report).[Mak93] The vaulting was attributed to “lack of torsional rigidity of the w- beam rail element and increased post spacing.” As the rail deflected the w-beam twisted allowing the front bumper to slide over the top of the rail. This exposed the front tire to direct contact with the w-beam rail, which resulted in the tire immediately beginning to climb the rail. The damage to the guardrail was minor and was confined to a 12.5-ft section of the rail. The maximum dynamic deflection was 20.4 inches.

37 Figure 33. Summary of TTI Test 0482-1 on the G4(2W) with 12.5-ft post spacing. [Mak93] Gabler, Gabauer and Hampton evaluated the effects of missing guardrail posts in a modified G4(1S) guardrail with routed wood blockouts under NCHRP Report 350 Test 3-11 conditions using finite element simulations.[Gabler10] Two impact points were used in the analyses: (1) at a post located at the upstream end of the unsupported section and (2) at the midspan of the unsupported section. The analyses indicated that, for the case of a single post missing, impacts at the mid-span of the unsupported section resulted in 25 percent higher rail deflections compared to the results of the undamaged system; however, the potential for system override was not evident in the simulations. As discussed earlier, a major concern of pocketing is rail-splice rupture. Gabler et al. evaluated the potential for splice rupture by recording and assessing increased magnitude in rail tension through the unsupported section of rail. The results indicate that rail tension increases as the damaged section included more and more missing posts, as shown in Figure 34. It was concluded by Gabler et al. that “the likelihood of the rails rupturing during impact increased as more posts were removed.”

38 Figure 34. Maximum rail tension as a function of missing posts.[Gabler10] Marzougui, Mahadevaiah, and Opiela used finite element analysis to evaluate various design modifications to the MGS guardrail for TL-2 applications.[Marzougui10] Five design alternatives were explored which involved doubling post spacing and reducing blockout depth, as listed below: 1. Doubling the post spacing from 6.25 ft. to 12.5 ft., 2. Reducing the blockout depth by half; from 11.811 in. to 5.906 in., 3. A combination of doubling the post spacing and reducing the blockout depth by half, 4. A combination of doubling the post spacing and having no blockouts, and 5. A combination of doubling the post spacing, having no blockouts, and lowering the height of the rail from 31 in. top of rail height to 27.75 in. Simulations of TL-2 impact tests were made for each design option, under both NCHRP Report 350 and MASH test requirements. Modification #1 – Doubled Post Spacing The simulations for both the small car and the 2000P pickup truck showed a smooth redirection and limited rail deflection. Major bending of only one post was observed with the pickup truck. The simulation of the heavier pickup with the higher center of gravity at a sharper angle resulted in greater system deflection, pull-out of two posts from the soil, and significant bending at one post in the MASH test. The vehicle’s front wheel was partially severed in the impact. The analyses of the system indicated that the system would pass all required safety performance criteria of Report 350 and MASH. The Report 350 tests showed a potential for

39 intrusion into adjacent traffic lanes for both the passenger car and the pickup, with both exit angles at 14 degrees.2 This criterion is preferable but not required. Modification #2 – Reduced Blockout Depth The small car simulation showed smooth redirection, but increased deflection and minor wheel snagging on the posts. For the Report 350 2-11 test there was a greater amount of deflection and more post bending than there was for the increased post spacing modification, but both were well within limits and both tests passed Report 350 criteria. The simulation results for the MASH 2-11 test were very similar to the increased post spacing modification, including the indication of the vehicle’s front tire being partially severed during impact. The critical values of occupant ridedown velocity and acceleration were higher than with the increased post spacing modification, but still within desired levels. The analyses of the simulations indicated that the system would pass all required safety performance criteria of Report 350 and MASH, and as such the researchers deemed reducing the blockout depth to be a viable design option. Modification #3 – Doubled Post Spacing and Reduced Blockout Depth As would be expected, there was increased deflection and minor wheel snagging on the posts for the small car when compared to either of the individual previous modifications. The results were still well within the limits of NCHRP Report 350, and were considered as passing. The Report 350 Test 2-11 simulation was not performed on this design modification, but the more stringent MASH test with the larger test vehicle was evaluated with computer simulation. The vehicle in this scenario behaved similarly to its response in each of the two previous individual design alternatives and met all MASH TL-2 criteria. Modification #4 – Doubled Post Spacing and No Blockouts For this design alternative, the researchers conducted both 2-10 and 2-11 simulations to MASH standards. The small car test met all of the crashworthiness requirements but exhibited significant wheel snagging. The pickup truck test also passed all the crashworthiness requirements, but displayed greater instability and greater lift as it was redirected. As such, the researchers stated that although the design met all safety performance requirements, it was considered less desirable than the previous design alternatives. Modification #5 – Doubled Post Spacing, No Blockouts, and Lowered Rail Height MASH Tests 2-10 and 2-11 simulations were performed for the system with double post spacing, no blockouts, and a 27.75-inch rail height. Although the lower rail height reduces the amount of underride for the small vehicle in the Test 2-10 simulation, it was noted that the vehicle experienced greater vertical trajectory and greater pitch. The Test 2-11 simulation showed similar deflection behavior as the same system with the higher rail height, but also showed more lift during redirection. The degree of upward pitch was greater for this modification than in any of the other tests. As with Modification #4, the high degree of vehicle instability renders this design alternative less desirable than the previous design options. 2 Note that this test is similar to the test conducted on the Washington State guardrail, but the guardrail in this case was the steel post MGS with double post spacing, whereas the Washington State guardrail was the G4(2W) with double post spacing.

40 Table 6 shows the results for the different simulation cases. Although all of the design alternatives met safety performance requirements, the authors stated that “…other considerations may render some of these alternatives undesirable (i.e., potentials for snagging, costs, sensitivity to factors).” Table 6. TL-2 computer simulation results. [Marzougui10] Design Option Report 350 Criteria MASH Criteria Test 2-10 Test 2-11 Test 2-10 Test 2-11 820C 43.5 mph (70 km/h) 20 degrees 2000P 43.5 mph (70 km/hr) 20 degrees 1100C 43.5 mph (70 km/h) 25 degrees 2270P 43.5 mph (70 km/hr) 25 degrees Modification #1 - Increased Post Spacing Pass Pass Pass Modification #2 - Reduced Blockout Width Pass Pass Pass Modification #3 - Increased Post Spacing & Reduced Blockout Width Pass Pass Modification #4 - Increased Post Spacing & No Blockout Pass Pass Modification #5 - Increased post Spacing, No Blockout, & Lower Rail Height Pass Pass Reduced Post Spacing Rosson, Bierman and Rhode conducted a study for the Midwest States Regional Pooled Fund Research Program to assess the effectiveness of three common strengthening techniques for strong-post w-beam guardrail: (1) nesting w-beam rail, (2) reduce post spacing by half (i.e., 3.1-ft post spacing), and (3) combination of nested w-beam rails and half post-spacing. The evaluations were carried out using computer simulation and full-scale crash testing.[Rosson96] Four full-scale tests were conducted and the results are shown in Table 7. The baseline test included the standard G4(1S) with W6x8.5 steel posts and blockouts with a single layer of 12-gauge w-beam rail and post spacing of 6ft-3in. All tests were conducted under the crash testing guidelines in NCHRP Report 230 Service Level 2, which involved a 4400-lb large car sedan impacting at a nominal speed and angle of 60 mph and 25 degrees, respectively. [Michie81]. Test KSWB-3 with the half-post spacing resulted in a 16 percent reduction in dynamic rail deflection, compared to the baseline test KSWB-1. Test KSWB-4 with the combination of nested rail and half-post spacing resulted in a 37 percent reduction in dynamic deflection, compared to the baseline test. Test KSWB-2 with the nested rails, on the other hand, resulted in an increase in lateral deflection, compared to the baseline test. It was stated by the researchers that they believed that the moisture content of the soil on the day that KSBW-2 was tested was

41 higher than it was when KSWB-1 was tested, “despite attempts to ensure the same density of soil for both tests.” Table 7. Full-Scale crash test impact conditions and results. [Rosson96] KSWB-1 (single w-beam with standard post spacing) KSWB-2 (nested w-beam with standard post spacing) KSWB-3 (single w-beam with 1/2 post spacing) KSWB-4 (nested w-beam with 1/2 post spacing) Test Vehicle Weight - lbs. (kg) 4398.2 (1995) 4486.4 (2035) 4486.4 (2035) 4501.8 (2042) Impact Speed - mi/hr (km/hr) 61.8 (99.6) 60.5 (97.4) 59.7 (96.1) 60.3 (97.2) Impact Angle - deg. 25.1 25.4 24.8 28.4 Impact Severity - ft lbf (kJ) 101783.5 (138) 101046 (137) 93670.3 (127) 123910.4 (168) Exit Speed - mi/hr (km/hr) 40 (64.5) 52.6 (84.7) 46.1 (74.2) 46.7 (75.3) Exit Angle - deg. 14.5 12.9 9.6 14.1 Maximum Dynamic Deflection - in. (cm) 27.9 (71.1) 32.4 (82.5) 23.5 (59.7) 17.6 (44.8) Maximum Permanent Set Deflection - in. (cm) 22.9 (58.4) 24.8 (63) 20.3 (51.6) 15.9 (40.4) In order to better isolate the variables in the system to only those variables being studied, computer simulation was undertaken to evaluate the three strengthening techniques. The computational analyses were carried out using the BARRIER VII program. In addition to the three strengthening alternatives, the parametric study also included three different soil conditions: high density clay, (2) low density clay and (3) granular soil. Table 8. Test matrix for post-in-soil bogie tests. [Rosson96] Class No. Post Type Soil Type Moisture Content Embedment Depth No. of Tests Conducted 1 Timber Clay Low Standard 2 2 Timber Clay Low Extended 2 3 Timber Clay Optimum Standard 2 4 Timber Clay High Standard 2 5 Timber Sand Unsaturated Standard 2 6 Steel Clay Low Standard 2 7 Steel Clay Low Extended 1 8 Steel Clay Optimum Standard 2 9 Steel Clay High Standard 2 10 Steel Sand Unsaturated Standard 2

42 Table 9. BARRIER VII post input variables. [Rosson96] Class Post Material Soil Material Embedment Moisture Content (%) KB (kip-ft) MA (kip-ft) DB (in) FB (kip) 1 Timber Clay Standard Low 443 19.8 20 14.1 2 Timber Clay Extended Low 324 21.2 15 15.2 3 Timber Clay Standard Optimum 359 23.8 20 17.0 4 Timber Clay Standard High 190 8.4 17.5 6.0 5 Timber Sand Standard Unsaturated 347 11.3 20 8.0 6 Steel Clay Standard Low 349 16.1 20 11.5 7 Steel Clay Extended Low 339 18.8 20 13.4 8 Steel Clay Standard Optimum 587 17.2 20 12.3 9 Steel Clay Standard High 139 8.2 20 5.8 10 Steel Sand Standard Unsaturated 291 9.1 20 6.5 The input parameters for the BARRIER VII program for the material and mechanical properties of the post-soil model were determined through physical testing of guardrail posts embedded in different soil types and at different embedment depths. A total of 21 dynamic bogie-impact tests were conducted in the test series, as shown in Table 8. Five different soil types were used in the tests including a well-graded medium grain unsaturated sand, silty clay with low moisture content, silty clay with high moisture content, and silty clay with optimum moisture content. Two post types were used: a W6 X 8.5 steel post and a 6x8-inch timber post. Two embedment depths were used: 44 inches and 50 inches. The BARRIER VII properties which were derived from the tests are presented in Table 9. The BARRIER VII model for each of the four guardrail cases were then performed and compared to the results of the four full-scale tests KSWB-1, KSWB-2, KSWB-3, and KSWB-4 as shown in Figure 35. The BARRIER VII model was then used to conduct a parametric investigation to evaluate the effectiveness of three different strengthening techniques and the effects of soil properties on the dynamic deflection of the systems. The results from the BARRIER VII simulations, shown in Table 10, indicated that soil density has a much greater influence on dynamic deflection than any of the other variables, while nesting of the w-beam rails was shown to have the least effect on deflection. As can be seen, the posts with the highest dynamic deflection are those that were embedded in either low density clay or sand. Although the dynamic deflection is consistently higher for steel posts than for wood posts, the difference is much smaller in all cases than it is for the soil type/density variation. Low Soil Confinement/Support Polivka et al. performed a study for the Midwest States Regional Pooled Fund Research Program to design and test a w-beam guardrail for installation at the break point of 2:1 slopes.[Polivka00b] The guardrail was to be evaluated according to NCHRP Report 350 Test 3-11 requirements. The study involved (1) conducting dynamic bogie-impact tests on guardrail posts installed in soil to determine post-soil response, (2) develop and calibrate a post-soil for use in the BARRIER VII program based on the results from the dynamic bogie-impact tests, (3) use

43 BARRIER VII to determine optimum barrier design, and (4) verify new barrier design via full- scale testing. Figure 35. Maximum dynamic deflection comparison plots: (a) KSWB-1, (b) KSWB-2, (c) KSWB-3, (d) KSWB-4. [Rosson96] Six bogie tests on steel posts installed at the break point of a 2:1 foreslope were conducted. The soil conformed to AASHTO M147-65 gradation “B” specifications (NCHRP Report 350 Strong Soil). The posts were impacted with a 2,143 lb bogie vehicle at the target speeds of 15 mph for the first five tests and 20 mph for the last test. The impact point in all tests was at 21.65 in. above the ground line and perpendicular to the front face of the posts. The test matrix for the six bogie tests is provided in Table 11 and a typical test-installation is shown in Figure 36. The bogie test results are shown in Table 12. The primary mode of failure occurred in the soil rather than post yielding in all six tests.

44 Table 10. BARRIER VII maximum dynamic deflection results. [Rosson96] Maximum Dynamic Deflection (cm) (% Decrease of Max. Defl. Compared with Std. Installation) Standard Installation W-Beam Nesting Only Half-Post Spacing Only Both Nesting & Half-Post Spacing Clay High Density Steel 68.1 62.7 (8%) 54.4 (20%) 46.2 (32%) Timber 62.7 59.9 (4%) 48.0 (23%) 41.4 (34%) Low Density Steel 102.4 95.8 (6%) 72.9 (29%) 69.6 (32%) Timber 98.8 91.9 (7%) 73.4 (26%) 69.1 (30%) Sand Steel 94.7 88.6 (6%) 69.3 (27%) 65.0 (31%) Timber 88.9 80.8 (9%) 65.0 (27%) 59.9 (33%) Table 11. Steel post bogie impact test matrix. [Polivka00b] Figure 36. Typical steel post test installation for bogie impact testing.[Polivka00b] Post Type Embedment Target Test No. ASTM Designation Depth Speed (in) (mph) MSB-1 W6x12 by 8 ft long 67 15 MSB-2 W6x12 by 8 ft long 55 15 MSB-3 W6x9 by 7 ft long 55.25 15 MSB-4 W6x9 by 7 ft long 55.25 15 MSB-5 W6x9 by 7 ft long 55.25 15 MSB-6 W6x12 by 9.68 ft long 87.25 20

45 The researchers used the results from the bogie tests to develop and calibrate a post-soil model for use in the BARRIER VII program. BARRIER VII was then used to evaluate four W- beam guardrail design alternatives on both level and sloped terrain conditions: 1. A system with 7-ft posts spaced 6.25 feet on center 2. A system with 8-ft posts spaced 6.25 feet on center 3. A system with 7-ft posts spaced 3.125 feet on center 4. A system with 8-ft posts spaced 3.125 feet on center The BARRIER VII results are shown in Table 13. The results of the analyses indicated that excessive guardrail deflections would occur with posts installed on the break-point of slopes for the standard post spacing of 6.25 feet for both embedment depths. It was theorized that the excessive deflection would greatly increase the possibility of severe wheel snag on posts. The analyses also indicated that for both embedment depths the guardrail with post spacing at 3.125 feet (i.e., half post-spacing) would result in less dynamic deflection than standard strong-post w- beam on level terrain conditions. The design selected for full-scale testing included 7-ft posts with 3.125-ft post spacing. Table 12. Steel post bogie test results. [Polivka00b] Table 13. BARRIER VII computer simulation results. [Polivka00b] Impact Peak Deflection at Test No. Speed Load Peak Load Results (mph) (kip) (in) MSB-1 14.5 26.5 2.67 Small soil failure at front of post, post rotation MSB-2 15.7 17.8 2.39 Large soil failure at front of post MSB-3 14.4 14.5 2.5 18-in radius soil failure at front of post MSB-4 14.4 17.4 2.11 15-in radius soil failure at front of post MSB-5 13.5 9.3 2.87 12-in radius soil failure at front of post MSB-6 21.2 30.6 3.79 No soil failure at the front of the post, post slightly twisted during rotation, post bent 55 in from the top with maximum deflection of 4.75 in.

46 The test installation, shown in Figure 37, was constructed adjacent to a pit with 2:1 slope, 9 ft wide, 3 ft deep and 62.5 ft. long. The guardrail system was impacted with a 1994 GMC 2500 ¾-ton pickup weighing 4,462 lbs., in accordance with NCHRP Report 350 TL-3 testing standards. The guardrail safely contained and redirected the vehicle and successfully met all safety performance criteria specified in NCHRP Report 350. The researchers noted, however, that “…it may be possible to obtain acceptable safety performance from a guardrail design which incorporates longer posts, a wider post spacing, or combinations thereof.” Marzougui et al. recently performed a study for the Federal Highway Administration to investigate the effects of raising the rail height of the G4(2W) strong-post w-beam guardrail placed at the break-point of a 2:1 slope.[Marzougui09] The increase in rail height was achieved by raising the blockouts three inches which, in effect, achieves the increase in rail height without altering post height or embedment depth. Figure 37. W-beam guardrail adjacent to 2:1 foreslope for full-scale crash test.[Polivka00b] The study involved the use of finite element analysis to simulate Report 350 Test 3-11 on four different cases:  Case 1: Standard G4(2W) on level terrain (baseline case)  Case 2: Standard G4(2W) placed at break-point of 2:1 slope  Case 3: G4(2W) with blockouts raised 3 inches, placed at break-point of 2:1 slope  Case 4: Same as Case 3, but with two bolts used to fasten the blockout to the post The results of the analyses indicated that the standard G4(2W) placed at the break-point of a 2:1 slope would result in an increase in rail deflection of approximately 2-inches, relative to the baseline rail deflection of 22.4 inches. The performance of the modified system with raised

47 blockouts was essentially the same as the standard system regarding barrier response, vehicle response and occupant risk measures. They concluded that there is no evidence from this analysis that the raised blockout design would increase the likelihood of barrier failure or have adverse effects on impacting vehicles. It was also stated that the proposed method of adjusting guardrail height by raising the rails and blockouts may be a low cost option for retrofitting existing field installations until system upgrades can be made. It was not stated in the report, however, what the embedment depths were. Blockouts The blockouts used in most strong-post guardrail systems serve a dual function: (1) to reduce the possibility of wheel-snag on the guardrail posts by creating separation distance between post and rail and (2) to ensure that the rail element maintains a critical height as the system deflects to allow enough time for the rail-to-post connection to release properly. The connection of the rail to blockout (or post if no blockout is used) is designed to release under relatively low loads. The failure mechanism for standard strong-post guardrails involves the bolt- head pulling through the slot in the w-beam rail. Although this release occurs rather quickly, it is not instantaneous so it is important that the rail remain above the critical height until it is released. Effects of Blockout Depth Figure 38 shows an illustration of how the blockouts serve to reduce the probability of wheel snag. In Figure 38(a) the eight-inch blockout provides adequate separation of the rail from the post to prevent wheel contact with the post; whereas in Figure 38(b) the wheel impacts against the post even at very low rail deflection. Full-scale tests on w-beam guardrail systems have shown that the front wheel on the impact side of the vehicle often snags on the guardrail posts as the top of the posts are pushed back by the rail element, exposing the lower section of the post to impact by the wheel, as shown in the test photo in Figure 39.[Mak94] Figure 38. Illustration of how blockouts help reduce possibility of wheel-snag.[Ray04]

48 Figure 39. Wheel snag against guardrail post in TTI Test 471470-26.[Mak94] For small, light-weight vehicles, a direct impact with a guardrail post can lead to high decelerations and occupant compartment intrusion, and/or vehicle instability.[Ray04] In such cases, the release of the rail from the post may be critical in reducing the effective resistance of the post. That is, if the rail does not release properly from the post, then the post is effectively a “fixed-pinned” beam restrained at the top by rail-to-post connection and at the ground line by the soil, which will not easily yield, particularly under the impact forces from a small car. Figure 40 shows an illustration of how the blockouts function to keep the rail above the critical height as the posts deflect. During impact, the rail element imposes a lateral load on the posts and as the posts are pushed back they rotate at a point below grade. For a 6-ft long strong- post with a rail height of 27 inches, the rotation point is typically between 25 and 30 inches below grade.[Rohde96] At post rotations less than approximately 30 degrees, the post with the blockout (Figure 40(a)) maintains the critical rail height, while the post without the blockout (Figure 40(b)) results in the rail being pulled down significantly by the rotating post. The release of the rail from the post is therefore much more critical when a blockout is not present. Karlsson used finite element analysis to evaluate the interaction between the vehicle wheel and guardrail posts in simulations of Report 350 TL-3 impact conditions.[Karlsson00] Several design alternatives for alleviating the effects of wheel snagging were investigated, including the use of longitudinal failure fuses on the posts as discussed earlier. Another alternative was the use of deeper blockouts to increase the separation distance between the rail and posts. The results from Karlsson’s study are shown in Figure 41 where the peak force between tire and post is plotted as a function of blockout depth. These results correspond to strong-post guardrails with a 27-inch rail-mounting height. The analyses indicated that (1) the impact forces decrease with increasing blockout depth, (2) the impact forces are approximately constant for blockout depths between 4 to 8 inches, and (3) a minimum blockout depth of 10 inches would be required to avoid wheel snag on strong-post w-beam guardrails (Note: the standard blockout depth on most strong-post guardrail systems is 8 inches). Figure 41 also suggests there are diminishing returns for using block-outs much greater than 10 inches (254 mm).

49 Figure 40. Illustration of how blockouts help to maintain critical rail height during post deflection. Figure 41. Peak tire-post impact force as a function of blockout depth for Report 350 TL-3 impacts.[Karlsson00] Karlsson’s conclusions were similar to those made by Sicking and Ross in their 1987 study on structural optimization of strong-post w-beam guardrail.[Sicking87] The results of the Sicking and Ross study indicated that a blockout depth of 10 inches for strong-post w-beam guardrails was sufficient to prevent wheel snag for both the small car and full-size sedan. (mm)

50 It is worth noting, that the MGS guardrail uses a 12-inch blockout but still experiences slight to moderate wheel snag on guardrail posts in Report 350 TL-3 tests.[Sicking02] Recall that the MGS has a standard rail height of 31 inches; whereas the earlier studies by [Sicking87] and [Karlsson00] pertained to strong-post guardrail with a 27-inch height so there is clearly a geometric relationship between the rail height and the optimal blockout depth. In a recent study at MwRSF, full-scale crash tests were conducted on the MGS guardrail with 8-inch blockouts and no blockouts.[Rosenbaugh12] The tests were performed according to MASH TL-3 criteria. This work is not yet published, however, a brief summary of the test results regarding occupant risk measures were presented at the Transportation Research Board (TRB) Roadside Safety Design Committee (AFB20) mid-year meeting in Irvine, California; those results are shown in Table 14. The results showed that occupant risk measures generally increase as blockout depth decreases. The pickup test for the non-blocked MGS resulted in increases of 12 percent and 20 percent for longitudinal and lateral occupant impact velocities, respectively, and increases of 40 percent and 87 percent for longitudinal and lateral occupant ridedown accelerations, respectively, compared to the MGS with 12-inch blocks. For the small car test, reducing blockout depth primarily results in increased occupant impact velocity (e.g., higher accelerations at the beginning of the impact event), where the MGS with 8-inch block results in a 42 percent increase in longitudinal Occupant Impact Velocity (OIV) compared to the 12-inch block; and the MGS without blocks results in a 111 percent increase in longitudinal OIV compared to the 12-inch block. Table 14. Comparison of full-scale test results for MGS with 12-inch blockouts, 8-inch blockouts and no blockouts.[Rosenbaugh12] Ray, Plaxico and Oldani evaluated the performance of the Alberta weak-post w-beam guardrail under Report 350 TL-3 impact conditions using finite element analysis.[Ray04] The Alberta weak-post system is composed of 6x8 inch wooden posts supporting a w-beam rail. The posts are spaced at 12.5 ft and the rail is fastened to the posts using 5/8-inch diameter bolts. This system is basically the G4(2W) strong-post system with increased post spacing and no blockouts. The results of Ray’s study indicated that vehicle decelerations caused by wheel-snag during the simulation of Report 350 Test 3-10 (i.e., small car test) would result in Occupant Ridedown Accelerations (ORA’s) that would only marginally pass the Report 350 safety criteria and it was suggested that small changes in event timing could lead to unacceptable results. The forces induced by the wheel-snag also resulted in excessive deformation inside the occupant compartment, which was deemed unacceptable based on Report 350 criteria. The results from

51 the simulation of Report 350 Test 3-11 (i.e., pickup truck test) indicated marginal performance of the guardrail system. The analysis results showed that the vehicle was very unstable when exiting the system, as the vehicle was airborne with a maximum roll angle of more than 45 degrees. Several design alternatives were investigated involving a smaller size wood post (5-inch diameter), different rail-to-post connection strengths (5.6k, 6.7k and 9.0k), w-beam splice moved to midspan between posts, and rail height (27, 30 and 32 inches). The use of this smaller sized post was based on the results of an earlier study performed by Bronstad and Burkett for the State of Ohio who recommended the use of 5-inch posts with ¼-inch bolts inserted through a pipe sleeve in the post for weak-post w-beam guardrail systems.[Bronstad71] Ray et al. concluded that “good performance cannot be expected consistently unless the guardrail height is at least 32- inches above grade and the connection strength is below 6.7k. The recommend design included:  Standard 12-gauge w-beam rail  Splices moved to the mid-span  32-inch rail-mounting height (top of rail)  5-inch diameter wood posts  Rail connected to posts with standard 5/8-inch diameter button-head bolts It was also suggested that post types other than wood be considered, which may provide better performance. The new design was never put in place since a suitable anchor system was not available. Effects of Collapsed Blockouts Blockouts used for most strong-post guardrail systems are designed to be “non- collapsible.” Earlier versions of the G4(1S) guardrail (i.e., steel-post w-beam guardrail) used W6x9 steel blockouts which have a relatively low torsional rigidity. The poor performance of that system was a direct result of the blockout collapsing in a lateral torsional buckling mode during impact, leading to wheel snag on the posts.[Mak96a; Mak99a] The wheel-snag was exacerbated by the twisting deformation mode of the blockout which caused the posts to fail prematurely in torsion resulting in low deflection of the posts at the groundline. The solution was to replace the steel blockout with the 6x8 inch wooden blockout that performed successfully in the G4(2W) guardrail.[Bullard96] Other versions of the modified G4(1S) incorporate “non- collapsible” blockout designs made from other materials (e.g., recycled plastic).[Bligh97b] This behavior is common in all strong-post guardrail systems that use structural-steel posts with low torsional rigidity (e.g., wide-flange posts), including the modified thrie-beam guardrail system shown in Figure 42. The modified thrie-beam guardrail (SGR09b) uses the same W6x9 steel posts as the G4(1S) system with a W14x22 blockout. This system was successfully tested for NCHRP Report 350 TL-3 at the Texas Transportation Institute. [Mak99a] It was concluded from the results of the test (i.e., Test 471470-30) that: “The relatively large deflection sustained by the guardrail system and snagging of the left wheel assembly with post 17 were somewhat unexpected given the stiffness of the thrie-beam element and the (14-inch) deep blockout. The soil condition was checked and found to be a little damp, but not to the extent that it would adversely affect the bearing

52 capacity of the soil. Review of the high-speed film showed that posts 16-18 were severely twisted from the vehicle impact as the thrie-beam element deflected. The W6x9 steel posts are relatively weak in torsion to begin with. The added moment arm due to the deep blockout aggravated the torsional moment acting on the posts. As the posts twisted, the blockouts essentially collapsed. This in effect increased the dynamic deflection of the guardrail by 18 inches … Also, the collapse of the blockout allowed the left front wheel assembly of the vehicle to come into direct contact with post 17 …”[Mak99a] Sheikh and other researchers at the Texas Transportation Institute recently used finite element analysis to evaluate the performance of the SGR09b for high-speed impact conditions (i.e., 4,577-lb pickup, 85 mph, 25 degrees) for the Texas Department of Transportation (TxDOT).[Sheikh09] TxDOT embarked on an effort to expand the state’s transportation system and expressed interest in the development of road systems with higher design speeds. Figure 42. Modified thrie-beam guardrail (SGR09b).[Sheikh09] The results of their analyses indicated that the performance of the SGR09b guardrail would not likely be acceptable in high-speed (e.g., 85 mph) impact events. They also concluded that the overall deformation mode of the blockout-and-post components, shown in Figure 43, leads to the poor performance of the system and suggested that these components be replaced by wood posts and blockouts. Although this system has been shown to perform well in Report 350 TL-4 conditions, it was concluded that performance would be further improved if the post and blockout were designed such that they had sufficient lateral stiffness to allow the post to rotate properly in the soil as the system deflected to reduce the potential for wheel-snag on the posts. Recall that the thrie-beam guardrail with 6x8-inch routed wood blockouts met the crash test requirements of MASH, but the vehicle appeared to be very unstable during redirection.[Bullard10] It is not known, however, how a solid block (e.g., wood, plastic, etc.) would affect the performance of the modified thrie-beam guardrail. Recall that the modified thrie-beam rail is 34 inches high (1.5 inches higher than the standard thrie-beam system) and has a 14-inch deep blockout with a unique cutout at the bottom of the blockout to help to keep the face of the rail vertical as the posts rotate back during collisions.

53 Figure 43. Deformation of thrie-beam guardrail during impact with 4,577-lb vehicle at 85 mph and 25 degree impact.[Sheikh09] Effects of Missing Blockouts Hampton and Gabler conducted pendulum testing on missing blockouts in Report 656.[Hampton10] The problem with pendulum testing alone is that it cannot predict the possibility of wheel-snagging due to the missing blockout. To account for this, the researchers continued on their previous work by conducting finite element modeling of the pendulum testing using LS-DYNA software. The model was created by modifying an existing model of the modified G4(1S) guardrail with routed wood blockouts obtained from the National Crash Analysis Center (NCAC). Parts of the model were deleted until only a small section that matched the original setup remained. They also increased the strength of the connections between the splice bolts and nuts, increased the mesh density of the rail, and lowered the strength of the soil in which the posts were embedded to better replicate the results of the actual pendulum tests. As a result, the pendulum simulations were able to predict the same test outcome as was observed in the real tests, as can be seen in Table 15. The researchers then combined the modified guardrail model with a model of a Chevrolet 2500 pickup truck obtained from the same NCAC library. They conducted a series of eight simulations (four impact points for each missing blockout location), keeping the initial vehicle speed and angle constant for all simulations (62.1 mph and at 25 degrees). “In all of the missing blockout simulations, the vehicle was observed to show more roll and pitch than was seen in a simulation of an impact into a guardrail not missing any blockouts… [however] …the roll and pitch were not high enough to conclude that the vehicle was unstable.” Most notably, however, is that no evidence was found in any of the simulations of major snagging of the vehicle tires.[Gabler10]

54 Table 15. Summary of pendulum test and model results. [Hampton10] Pendulum Test 1 Pendulum Test 2 Pendulum Test 3 Real Test Model Real Test Model Real Test Model Test Outcome Contained Contained Splice Fail Splice Fail Contained Contained Maximum Rail Deflection (inch) 25.9 25.5 27.2 27.2 24.9 25.2 Splice Post Bolt Intact Intact Broken Pulled out Intact Intact Non-splice Post Bolt Pulled out Intact Pulled out Pulled out Pulled out Intact Rail-to-Post Connection As mentioned earlier, the release of the rail-to-post connection is critical for proper performance of the guardrail. If the release forces are too low, the rail will release too soon allowing the rail to drop ahead of the vehicle, increasing the probability of the vehicle overriding the rail and penetrating behind the system. If the release forces are too high, the rail may release too late, or not at all, resulting in the rail being pulled down by the posts during impact.[Bligh97] Thus, the post-rail connection must fail consistently and at the appropriate time to ensure that the rail maintains the proper height during impact.[Engstrand00; Ray01a; Sicking02] It was also demonstrated earlier that increased blockout depth can help alleviate the sensitivity of the rail-to- post release on system performance (refer to the section on blockouts for more information). Ray, Engstrand and Plaxico at Worcester Polytechnic Institute and McGinnis at Bucknell University conducted a study for the Pennsylvania Department of Transportation to improve the crash performance of the weak post w-beam guardrail system (G2);[Engstrand00; Ray01a] This system is very popular in Connecticut, New York, Pennsylvania and, to a lesser extent, in Virginia and North Carolina. The original G2 design successfully met performance criteria of Report 230 but failed to meet that of Report 350. The weak-post W-beam guardrail is composed of W-beam rails supported on weak S3x5.7 steel posts with rectangular soil plates. The system performs much like the cable guardrail in that the posts hold up the rail at the proper height until the guardrail is struck by an errant vehicle. The posts are spaced at 12.5 feet and the rail is connected to the posts using 5/16- inch diameter bolts with 1.75 inch square washers under the head. The bolts are designed to fail in an impact allowing the rail to separate easily from the post. The rail separation from the post is an important feature of the design since this action allows the rail to remain in contact with the vehicle instead of being pulled to the ground by the post. Once the rail is separated from the post, the posts are bent back as the w-beam deflects and slips over the top of the posts. The W-beam rail then redirects the vehicle, acting like a cable that is anchored at the ends. Although the basic functionality of weak post systems is quite different from its strong-post “cousin,” the weak post w-beam guardrail shares many of the same components which perform the same basic functions at the local component level. Ray et al. demonstrated that relatively small changes in several important design details can significantly affect performance of the weak-post w-beam guardrail.[Ray01a] They found that: (1) the post-rail connection must fail consistently and at the appropriate time to prevent the rail from being pulled down with the post during impact; (2) rupture of guardrail splices were largely a result of high stress concentrations around the splice-bolt holes as the splice bends

55 about the post and blockout, which could easily be avoided by simply moving the splices to the mid-span (i.e., non-post locations); (3) backup plates are needed when the rail is directly exposed to other components with sharp edges, such as steel I-beam posts when blockouts are not used, to prevent small tears in the rail; and (4) that guardrail height must be sufficient to prevent override of vehicles with high centers of gravity (e.g., pickup trucks) while still preventing under-ride of smaller vehicles (e.g., small cars). These modifications were initially investigated through FEA analysis and later through a series of four full-scale crash tests conducted at TTI from November 1999 through June 2000. [Buth99a-c; Buth00e] The final modified system consists of the following components:  12 gauge w-beam guardrail mounted 32.25 inches (820 mm) above the ground with splices at mid-span,  S3x5.7 (S75x8.5) weak steel posts with soil plates spaced at 12.5 ft (3,810 mm) and attached to the rail at non-splice locations  W-beam backup places at each post, and  A post-rail connection consisting of one 5/16-in diameter ASTM F568 Class 4.6 bolt with two 1.6-in square washers and two nuts. The resulting modified design was successfully crash tested to both Report 350 and MASH performance criteria.[Buth00c; Buth00d; Bullard10] The importance of this study is that many of the design modifications that were made to the G2 system were later implemented into the strong-post system (i.e., the MGS guardrail), which demonstrated marked improvement over the standard G4(1S) and G4(2W) systems.[Sicking02] As with the design of the weak post w-beam guardrail [Engstrand00; Ray01a], the rail- to-post connection is also an important design consideration for strong post systems. In a study by Plaxico et al., the force and energy required to fail the post-bolt connection (i.e., bolt head pulling through the w-beam slot) in strong post guardrail systems was determined through quasi- static laboratory testing.[Plaxico03] Four scenarios were investigated which are illustrated in Figure 44:  Case 1. One layer of w-beam with the bolt positioned in the center of the slot in the w- beam,  Case 2. One layer of w-beam with the bolt positioned at the edge of the slot in the w- beam,  Case 3. Two layers of w-beams (e.g., a splice connection at a post) with the bolt positioned at the center of the slots in both sections, and  Case 4. Two layers of w-beams with the bolt positioned at the edge of the slots in both sections. The results of the tests are shown in Figure 45. The minimum pull-through force required to fail the connection occurred in case 1 (i.e., one layer of w-beam with the bolt in the center of the slot) with an average load of 4,046 lb (18 kN). The maximum pull-through force occurred in case 4 (i.e., two layers of w-beams with the bolt positioned at the corner of the slot) with an average load of 14,545 lb (64.7 kN); which was a difference of 3.6 times compared to the minimum failure force. The results shown here were measured from quasi-static tests, thus the

56 effects of strain-rate were not invoked in the tests but the relative difference in magnitude between the four connection cases should be similar in impact events. Figure 44. Position of bolt in slotted hole in w-beam for each load case.[Plaxico03] Figure 45. Results from uniaxial tests on rail-to-post connections.[Plaxico03] Moving the rail splice to the mid-span between guardrail posts improves crash performance for strong post guardrail systems by serving two very important functions: (1) it Load case 1 Load case 2 Load case 3 Load case 4 (Single Layer of W-Beam)(Single Layer of W-Beam) (Two Layers of W-Beam) (Two Layers of W-Beam) Load case 1 Load case 2 Load case 3 Load case 4

57 alleviates high-stresses in the splice connection by moving the splice away from the post (see earlier discussion) and (2) results in a more consistent failure load for the rail-to-post connection by limiting the connection to a single layer of w-beam, as illustrated in load cases 1 and 2 in Figure 45.[Engstrand00; Polivka00a; Plaxico03] Recall that in the development of the MGS guardrail, the rail splice was also moved to the midspan for these reasons.[Sicking02] The design of the rail-to-post connection was further improved for the MGS by lengthening the post- bolt slot in the w-beam rail, which further reduces the connection strength and improves consistency of the failure load by making it less likely for a post-bolt to be positioned near the edge of the slot during installation.[Sicking02] Rail Element Splice Failures Splice failures have been observed in a wide variety of w-beam guardrail types, including weak-post w-beam guardrail [Kilareski99], strong-post w-beam guardrail [Buth99a; Ross99; Mak99b; Polivka00a], as well as w-beam end-terminals and transitions [Mak96b; Mak96c]. Essentially all w-beam barriers use the same 8-bolt splice connection for connecting the ends of w-beam rails to each other and all such systems occasionally experience splice failures. Based on a comprehensive review of these tests it was found that:  Guardrail rupture is not usually caused by defective or substandard material, based on the results of tensile tests on specimens taken from the failed w-beam,  In every case where it could be determined, the rupture occurred downstream of the vehicle,  In the only test that was instrumented with a strain gauge on the rail, rail tension was no more than 29.2 kip (i.e., approximately 32% of the tensile capacity of the splice connection),  Dynamic deflections when noted were usually modest,  Failures were observed with both large and small vehicles  Tears always pass through at least one splice hole and the bottom-downstream hole is usually located on the tear line – the only exception being Test RF476460- 1-5 on the G4(2W). These results suggest that splice failures cannot be adequately explained by material deficiencies or axial rail capacity. The splice connections in most guardrail and guardrail terminal systems are located at the guardrail posts so the loading experienced by the splice is a combination of the axial guardrail tension, torsion in the guardrail section about its longitudinal axis as well as lateral bending due to displacements of the posts. Ray explored the mechanics of w-beam splice connections using finite element analysis, laboratory tests and full-scale tests. While rail ruptures occur on all types of w-beam guardrail systems, Ray’s study focused on the weak-post w-beam guardrail. The results of his study demonstrated that rail ruptures are commonly caused by (1) high stress concentrations in the splice connection (e.g., as the splice bends around the post), as illustrated in Figure 46 and Figure 47, and (2) small tears/cuts in the rail caused by the rail sliding over relatively sharp edges (e.g., top of posts) while the rail is in tension.

58 Figure 46. Effective plastic strains in the back layer of w-beam in a guardrail splice showing the formation of a plastic hinge. [Ray01b] Figure 47. Von Mises stress contour plot showing relatively low stresses on the top layer of rail in the rail splice.[Ray01b] Ray et al. concluded that W-beam guardrail splice failures are usually caused by the complex multi-axial state of strain experienced by the splice when it is located near a guardrail post. When subjected to these multidirectional loads, stress concentrations develop around the bolt holes in the back layer of w-beam in the splice connection and this often results in a small fracture or tear in those locations. Once a tear is initiated, the tension in the rail may cause the tear to propagate through the whole w-beam section causing the guardrail to rupture completely.[Engstrand00; Ray01a] A similar conclusion was also made by Polivka et al. in a study for the Midwest States Regional Pooled Fund Research Program to investigate the effects of curbs placed in combination with strong-post guardrail.[Polivka00a] In their study a four-inch tall curb was

59 placed under a G4(1S) guardrail with the edge of the curb flush with the w-beam rail. The test (i.e., NEC-1) resulted in rupture at a w-beam splice. It was stated that the “reduction in (w-beam) cross-section (at splice connections) tends to localize strain in the splice region and leads to rail rupture near the point that the full cross-section begins to yield … Subsequently, a tear in the w- beam rail was observed at the bottom downstream bolt location of the rail splice which later propagated upward through the reduced-area cross-section.”[Polivka00a] Gabler, Gabauer and Hampton evaluated the effects of splice damage using pendulum tests in Report 656.[Gabler10] The test article was an 18-ft section mockup of the modified G4(1S) with routed wood blockouts, as illustrated in Figure 48. Each end of the w-beam rail was anchored to rigid posts with two ¾-inch diameter swaged cables (i.e., AASHTO component FCA01]. Figure 48. Pendulum test setup for the Gabler et al. study. [Gabler10] In their tests the damage mode was fabricated by cutting out the material around one of the bottom splice bolts, as shown in Figure 49. The damage mode was essentially representative of a missing splice bolt at that location. The test article was impacted perpendicular to the face of the rail with a 4,545-lb pendulum at a nominal speed of 20 mph. The test article contained the pendulum and “no serious splice separation was observed.” The results of the Gabler et al. study verifies the conclusions from Ray et al. and Polivka et al. that splice failure is not likely the result of reduced rail capacity. A simple yet very effective means of minimizing the chance of a guardrail rupture is to relocate the splice to the mid-span of the guardrail. As discussed earlier, this design alternative was implemented in the modified weak-post w-beam guardrail system [Engstrand00; Ray01a] and the MGS strong post guardrail system.[Sicking02] The splice connection performed well in all cases and both systems passed all safety and structural adequacy requirements of Report 350 and MASH.[Buth00c; Buth00d; Bullard10; Sicking02; Polivka06c-e]

60 Figure 49. Fabricated splice damage mode evaluated by Gabler et al. [Gabler10] Rail Tear Another critical failure mode related to rail rupture is small tears in the rail that propagate when the rail is tensioned. This type of damage is typically caused during an impact event when the rail slides over, or bends around, sharp edges such as the edge of a steel guardrail post. In most strong-post guardrails the rail element is not directly exposed to such dangers since the rail is generally separated from the posts by wooden or plastic blockouts. In the case of weak-post guardrail systems, however, the rail is fastened directly to the post. The posts used in the weak- post w-beam system are structural steel, small-flange posts (i.e., I-shape with small flanges) which twist as they deflect due to their low torsional stiffness. During impact, as the rail detaches and slides up and over the top of post, the rail element is exposed directly to the edge of the twisted post. This failure mode was demonstrated in a full-scale test on the weak-post w-beam system conducted by the Texas Transportation Institute under Report 350 Test 3-11 impact conditions.[Buth00a] A small “nick” formed at the bottom edge of the w-beam rail as the rail was pulled over the top of the second post downstream of the impact point, as shown in the top photo of Figure 50. During impact, the friction between the vehicle and the w-beam rail causes the section of rail upstream of the vehicle (i.e., behind the vehicle) to be under much higher tension than the down-stream sections of rail. As soon as the vehicle moved past the tear, the tear was suddenly on the tension side of the rail, resulting in crack propagation and, finally, complete rupture of the rail, as shown in Figure 50. The solution to the problem was to use a w-beam back-up plate as a sacrificial element to shield the rail from the sharp edges of the post. Ray et al. determined, through the use of finite element analysis, that the backup plate would reduce the effective plastic strain in the rail by 38 percent.[Ray01a] Subsequent tests on the weak-post w-beam system with this modification successfully met performance requirements for Report 350 and MASH for TL-3.[Buth00c; Buth00d; Bullard10]

61 Gabler, Gabauer and Hampton evaluated the effects of vertical tears in the w-beam rail using pendulum tests in Report 656.[Gabler10] The test setup was shown earlier in Figure 48. Three tests were conducted: Test 01-3 and Test 03-5 included a 4-inch cut made on the top of the rail element, and Test 08-2 included a 0.5-inch cut on the top of the rail. In each test case, the test article was impacted perpendicular to the face of the rail with a 4,545-lb pendulum at a nominal speed of 20 mph. The “tear” propagated in all three tests and resulted in complete rupture in Tests 01-3 and 03-5. It was concluded that “no vertical tear is safe” and that “vertical rail tears of any length … should be repaired with high priority.” [Gabler10] Effects of Rail Height Another important design change made by Ray et al. to the weak-post w-beam guardrail was to raise the height of the rail from 27 inches to 32.25 inches.[Engstrand00; Ray01a] A full- scale test of the modified weak-post w-beam system (i.e., improved rail-to-post connection, splices moved to the mid-span and back-plates at the posts) with the rail height at 27 inches resulted in the 2000P test vehicle (i.e., 4409-lb pickup) overriding the rail in a full-scale test conducted by TTI (Test 473750-2) under Report 350 Test 3-11 conditions.[Buth00b] Finite element analysis was used to investigate the effects of rail height and it was determined that a rail height of 32.25 inches was sufficient to prevent override of the pickup truck while preventing underride of the 820C test vehicle (i.e., 820 kg small car). The modified weak-post w-beam guardrail system with the higher 32.25 mounting height was subjected to a full-scale crash test according to NCHRP Report 350 Test 3-10 (i.e., small car test) and Test 3-11 (e.g., pickup test), as shown in Figure 51 and Figure 52, respectively. The tests demonstrated that the modified design met all Report 350 performance requirements for TL-3 [Buth00c; Buth00e] The system was later successfully tested under MASH TL-3 conditions.[Bullard10] Marzougui et al. performed a study for the Federal Highway Administration to investigate the effects of rail height on the safety performance of the modified G4(1S) with routed wood blockouts using finite element analysis and full-scale testing.[Marzougui07b] Marzougui’s analyses involved simulations of Report 350 Test 3-11 on a model of the modified G4(1S) with the rail height of ±1.5 inches and ±3 inches, relative to the standard rail height of 27 inches (i.e., rail heights of 24, 25.5, 27, 28.5 and 30 inches). The results of Marzougui’s study indicated that a rail height of 27 inches for the G4(1S) guardrail would successfully redirect the pickup under Report 350 TL-3 conditions; however, “reducing rail height by as little as 1.5 inches could hinder the ability of the barrier to redirect pickup trucks and large SUVs.” Figure 53 shows the results of the analyses for rail heights of 27 inches (i.e., standard height), 25.5 inches (i.e., rail lowered 1.5 inches) and 24 inches (i.e., rail lowered 3 inches). Figure 54 shows the results of the full-scale tests for rail heights of 27 inches and 24.5 inches (i.e., rail lowered 2.5 inches).

62 Figure 50. Test 473750-1 on a weak-post w-beam guardrail resulting in rail rupture. [Buth00a]

63 Figure 51. Sequential view of Report 350 Test 3-11 on the modified G2 at 32.25 inch rail height.[Buth00c] Figure 52. Sequential view of Report 350 Test 3-10 on the modified G2 at 32.25 inch rail height.[Buth00e] Figure 53. FEA results for Report 350 Test 3-11 on the G4(1S) guardrail with rail heights of (a) 27 inches, (b) 25.5 inches and (c) 24 inches.[Marzougui07a]

64 Figure 54. Full-scale test results for Report 350 Test 3-11 on the G4(1S) guardrail with rail heights of (a) 27 inches and (b) 24.5 inches.[Marzougui07a] End Terminals The primary purpose of end-terminals is to anchor the ends of a guardrail, but end- terminals must also be crashworthy themselves. In fact, end-terminals are much more complex systems than guardrails. Guardrails are designed for one basic type of loading, (i.e., lateral impact on the traffic-facing side of the system). Accordingly, only two tests are required in MASH to assess the crashworthiness of guardrails: Test 3-10 (small car impact at 62 mph and 25 degrees) and Test 3-11 (pickup truck impact at 62 mph and 25 degrees).[AASHTO09] End-terminals, on the other hand, can be impacted on the side or on the end. The loading of an end-terminal’s components, and hence their intended function, will vary depending on several factors, including: (1) which end of the guardrail the end-terminal is located (e.g., down- stream or upstream), (2) where the end-terminal is struck (side hit or end-on), (3) the orientation of the vehicle (frontal impact or side impact), (4) impact direction relative to up-stream or down- stream placement (forward or reverse hit) and (5) the end-release function of the end-terminal (i.e., gating or non-gating). These additional functions of an end-terminal are reflected in the number of tests and impact conditions that are required in the crash testing guidelines. The buried-in-backslope end-terminal is the only non-proprietary end-terminal that meets FHWA eligibility for NCHRP Report 350 TL-3; this system, however, has limited application since it can only be installed in locations where there is a backslope. There are several older, obsolete end-terminal installations still in service, but it is expected that most of these will be replaced by proprietary systems in the near future. Unfortunately, proprietary systems all function somewhat differently (e.g., some are gating, some are non-gating, each has a unique energy absorbing mechanism, etc.). Because of the complexity and the proprietary nature of these systems, a comprehensive assessment would require the consent and participation from manufacturers. Although a great deal of attention was focused on the crashworthy design of guardrail terminals in the late 1990’s and early 2000’s, little change was made to the basic design of the guardrail anchor mechanism. For example, the anchor mechanism used on most FHWA eligible end terminals including the BEST, SKT, ET2000, ETplus, and FLEAT, were adopted from a non-proprietary design developed for the Modified Eccentric Loader breakaway cable Terminal

65 (MELT). The basic design of the anchor mechanism, illustrated in Figure 55, entails an anchor cable (3/4-inch diameter) with one end fastened to the rail element and the other fastened to the end-post near the groundline. A bearing plate is used to prevent the cable from pulling through the hole in the end-post and to distribute the load to the foundation tube (i.e., the bearing plate overlaps the top of the foundation tube). The end-post and the adjacent post are both installed in 5-foot long steel foundation tubes with a soil-bearing-plate (18 inches long by 24 inches wide) attached to provide additional resistance from the soil. A steel strut is then laid at the groundline which connects the two foundation tubes together making them work as a system. When the guardrail is struck, the anchor cable transfers the load to the foundation tubes to resist axial movement of the rail; the resulting tension in the rail helps to limit lateral deflections of the guardrail in the impact zone. Figure 56 shows a photo of a generic end-anchor for the modified G4(1S) with wood blockouts that was used in full-scale crash test 2214-WB1.[Polivka06a] Figure 55. Sketch of typical guardrail anchor system. Figure 56. Generic end-terminal used in full-scale Crash Test 2214-WB1 of a strong-post guardrail system.[Polivka06a] Foundation Tubes BCT PostBCT Post Anchor Cable Soil Bearing Plate Soil Bearing Plate Groundline Strut Cable Bearing Plate Cable Anchor to Rail

66 If the anchor fails or experiences excessive movement during loading, the “slack” in the rail generally results in pocketing. For example, Polivka et al. and Bullard et al. showed through full-scale testing that poor guardrail performance is often associated with large anchor movement.[Polivka00a; Bullard00] In a full-scale test (i.e., Test NEC-1) conducted at the Midwest Roadside Safety Facility at the University of Nebraska-Lincoln to evaluate the effects of a 4-inch curb placed flush behind the face of a modified G4(1S) guardrail system, the anchor posts split during the collision, as shown in Figure 57, causing a loss of tension in the w-beam which resulted in pocketing. The test was conducted under NCHRP Report 350 TL-3 conditions.[Polivka00a] The guardrail ruptured at a splice connection allowing the vehicle to penetrate behind the system. The splice failure was attributed to contact and snagging of the post blockout against the w-beam rail splice. The post twisted as it was pushed back in the soil, causing the bottom corner of the blockout to push up against the corner of the w-beam rail splice. This resulted in a tear in the w-beam at the lower downstream bolt location. Figure 57. Test results for Test NEC-1 on modified G4(1S) with wood blockouts and 4-inch curb. [Polivka00a] Researchers at the Texas Transportation Institute conducted a NCHRP Report 350 TL-3 test on a G4(2W) guardrail with a 4-inch curb set out one inch from the face of the w-beam rail (i.e., Test 404201-1).[Bullard00] During the test, there was significant movement of the anchor system as the foundation of the anchor posts moved in excess of 2.75 inches. The test was successful; however, there was considerable damage to the guardrail system due to the movement of the anchor, as shown in Figure 58, and it was stated in the test report that the guardrail was at its performance limit.

67 Figure 58. Guardrail damage in Test TTI 404201-1.[Bullard00] Report 656 provides some guidance on criteria for repair of generic end-terminals which was based on Ohio Department of Transportation’s check list for repairing energy absorbing end terminals.[Gabler10] The repair criteria shown in Table 16 “was based solely on engineering judgment; no finite element simulations of pendulum tests evaluating these end terminal damage modes were conducted.”[Gabler10] Table 16. Summary of generic end terminal repair guidance. [Gabler10]

68 Summary of Literature Review A review of the literature on guardrail design and evaluation studies was conducted to garner information regarding how the various elements of a guardrail system function as well as their respective role in overall system performance. The review considered four crash phenomena that lead to poor performance of strong-post guardrail: (1) pocketing, (2) wheel snag, (3) barrier override/underride, and (4) rail rupture. The results of several studies related to the sensitivity of various components were presented, including supporting information from full- scale testing, component testing and/or numerical analyses. These studies provided a great deal of information on the general function and influences of post strength, soil strength, post embedment depth, blockout material, blockout depth, blockout shape, rail-to-post connections, rail height, splice damage and rail tears. The information presented here was used for identifying:  The guardrail types to be considered in the study  The critical damage modes associated with each guardrail type,  The combination(s) of damage modes that are considered most likely to be characteristic of damaged systems, and  Which of those combinations are considered most critical to system performance.