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.

271 CHAPTER 11 – RE-ASSESS CAUSE OF FAILURE IN TEST C08C3-027 FOR THE MODIFIED G4(1S) WITH PRE-CRASH- INDUCED DEFLECTION. The undetached post-rail connection in Test C08C3-027.2 that led to the vehicle overriding the barrier was suspected by the authors to be a symptom of low rail tension resulting from the excessive movement of the anchor during the test. The focus of this chapter is to reevaluate the G4(1S) installation of Test C08C3-027.2 and investigate the cause for the improper rail detachment. The anchor system used in Test C08C3-027, shown in Figure 216(c) included only a single foundation tube, and was therefore not as stiff nor as strong as the standard anchor system used in the earlier tests of the G4(1S) (i.e., Figure 216 (a) and Figure 216 (b)), which included two foundation tubes connected via a groundline strut. Figure 216. Anchor used in evaluation of the modified G4(1S) with wood blockouts in tests (a) 405421-1 [Bullard96], (b) 2214-WB2 [Polivka06b] and (c) C08C3-027 [Fleck08a;08b]. A model of the G4(1S) guardrail was developed and finite element analysis was used to simulate full-scale crash test C08C3-027-1 to create low-level crash-induced damage of the system. The results from this low-speed impact case, including guardrail component deformations and residual stresses, were then used as initial conditions in a secondary high-speed impact simulation of full-scale crash test C08C3-027-2. The upstream and downstream end- terminals for the system were modeled using non-linear springs attached to the ends of the rail. To represent the response of the single-foundation tube anchor system used in the full-scale tests, the force-deflection properties of the end-terminal springs were scaled to 47 percent of the baseline anchor stiffness defined in Chapter 9. The results of the analysis were compared to those of the full-scale crash test to determine if the model reasonably reproduced the results of the full-scale test. If so, then it may be reasoned that the most likely cause for the override was the collective effects of the low- stiffness anchor system combined with the initial crash-induced rail deflections – rather than the rail deflections alone. (a) (b) (c)

272 Simulation of Test C08C3-027-1 Test C08C3-027-1 Summary Test C08C3-027-1 was conducted by MGA Research Corporation on August 6, 2008. The test article consisted of 12.5 feet long 12-gauge w-beam rail; the rail was supported with W6x9 structural steel posts that were 72 inches long, embedded 44 inches in the soil and spaced at 75 inches on center; the rail was blocked out from the post using 6x8x12 inch routed wood blockouts; the blockout and rail were attached to the post using 5/8-inch diameter carriage bolts. The top height of the guardrail was 27.8 inches. The overall length of the test installation for Test C08C3-027-1 was 162.4 feet including two single-foundation tube anchors – one at each end of the system. Full-scale test C08C3-027-1 involved a 1997 Chevrolet 2500 pickup impacting the G4(1S) guardrail at 30.0 mph and at an impact angle of 26 degrees; the impact point was at 45.8 inches upstream of the splice connection at Post 11. The gross static mass of the test vehicle was 4,632 lb. The damage to the test installation is shown in Figures 217 and 218. The maximum deflection of the rail was 14.5 inches and occurred at the midspan between Post 11 and Post 12. Figure 217. Damage to guardrail in low-speed impact test C08C3-027-1.[Fleck08a]

273 Figure 218. Damage to guardrail in low-speed test C08C3-027-1 (overhead view).[Fleck08a] FEA Model Development The finite element model of the G4(1S) guardrail is shown in Figure 219. The guardrail model consisted of eleven 13.5 feet lengths of 12-gauge w-beam rail, twenty-two W6x9 structural steel posts, and twenty-two 6x8x13 inch routed wood blockouts. The center-to-center distance between rail splices was 12.5 feet. The posts were spaced at 75 inches on center and the w-beam rail was positioned such that the top of rail was 27-5/8 inches above ground. The posts were embedded 44 inches in the ground. The model included 138.6 feet of the guardrail including Post 3 through Post 24. The boundary conditions at the ends of the rail were modeled using nonlinear springs with properties corresponding to 47 percent of the baseline anchor stiffness; this was based on the assumption that the single-foundation tube anchor would have approximately half the stiffness of the baseline two-foundation tube anchor system. The overall effective length of the model (including the simulated end-terminals) was 162.4 feet. The finite element model for the rail element, the blockouts and the connection hardware were adopted from the G4(2W) guardrail model developed in Chapter 7. The blockout was modified by including the “routed” section that fits over the flange of the W6x9 steel posts. The finite element model of the W6x9 posts were modeled with the fully-integrated Type 16 shell elements in LS-DYNA with warping stiffness. The nominal element size was 0.4 x 0.5 inches for the post flange and 0.6 x 0.5 inches for the post web. The material properties for the post were characterized based on the properties determined in an earlier study by Wright and Ray.[Wright96] This material model has been used in numerous analyses and has been validated in simulations of full-scale crash tests.[Plaxico02] For consistency with the G4(2W) model, the soil response was modeled using the non-linear spring concept with the springs attached directly to the post. The properties of the spring elements were defined according to [Plaxico98] using a soil density of 134 pcf and no moisture content (refer to Chapter 7 for more details). 10 11 12 13

274 Figure 219. Finite element model of G4(1S) for simulation of Test C08C3-027-1. Post 11 Post 3Post 24 138.6 ft 44 in 28 in Soil Spring Model

275 FEA Simulation Finite element analysis was then used to simulate the impact conditions of Test C08C3- 027-1. At the beginning of the analysis the post-bolts that fasten the w-beam and blockouts to the post were tightened to approximately 2,000 lb axial force by imposing an initial strain-time history to the bolt elements via the LOAD_THERMAL card in LS-DYNA. The vehicle model used in the analysis was the NCAC C2500D version 5B with modifications described in Chapter 7. The total mass of the vehicle model was 4,568 lb. The vehicle model struck the guardrail at 45.8 inches upstream of the splice connection at Post 11 at an impact speed and angle of 30.0 mph and 26 degrees, respectively. The analysis was conducted with a time-step of 1.26 microseconds for a time period of 0.6 seconds. Figure 220 shows sequential snapshots of the impact event from an overhead viewpoint comparing the results from the FE analysis with the full-scale test. The yaw angle of the model was greater than that of the test, due primarily to the difference in the response of the wheel assemblies during impact, as illustrated in Figure 221. In the full-scale test the wheel snagged on Post 11 causing the front wheels to turn “full-steer” toward the barrier; whereas in the simulation the wheel’s contact with the post was not sufficient to significantly alter the steer angle. In the full-scale test, the steer angle of the wheels resulted in higher deceleration of the vehicle and also altered the trajectory during the impact event, compared to the results of the FE analysis. The W-Beam rail element was deformed from Posts 9 through 14 as shown Figure 222. The maximum permanent deflection of the rail in the FE analysis was 15.5 inches and occurred at 47 inches upstream of Post 12 (compared to 14.5 inches deflection at 37.5 inches upstream of Post 12 in the full-scale test). All the post-bolt connections remained attached throughout the impact for both the full-scale test and the FE analysis. The maximum permanent groundline deflections of Posts 10 through 13 in the FE analysis were 0.7 inches, 7 inches, 4.2 inches and 0.24 inches, respectively. Although it cannot be confirmed, it appears from visual inspection of the damaged test article that the deflections of the posts are of similar magnitude. The maximum deflection of the rail at the upstream boundary was 1.2 inches. Figure 223 shows the post-test photograph of the upstream anchor after Test C08C3-027-1. The separation between the back of the post and the soil in the photo is evidence of upstream anchor movement, but the actual amount of displacement was not included in the test report. Overall, the barrier damages resulting from the simulated impact were reasonably representative of that observed in the full- scale test.

276 Figure 220. Sequential views of FEA results compared with Test C08C3-027-1. 0.112 s

277 Figure 221. Impact response at 0.18 seconds for (a) full-scale test and (b) FEA illustrating wheel orientation. (b) (a)

278 Figure 222. Comparison of guardrail damage for (a) Test C08C3-027-1 and (b) FEA model. Figure 223. Photo of upstream anchor after Test C08C3-027-1.[Fleck08a] (b) (a)

279 Simulation of Test C08C3-027-2 Test C08C3-027-2 Summary The damaged guardrail from Test C08C3-027-1 was then subjected to a second impact at high-speed to evaluate the performance of the crash-damaged system. Test C08C3-027-2 was performed by MGA Research Corporation on August 7, 2008 under Report 350 Test 3-11 conditions. No repairs of the pre-damaged guardrail were made prior to this second test. In Test C08C3-027-2, a 1997 Chevrolet 2500 pickup impacted the damaged rail at 38.5 inches upstream of Post 11 at an impact speed and angle of 62.1 mph and 26.4 degrees, respectively. At 16 milliseconds after impact Post 11 started to deflect back, and at 44 milliseconds Post 12 began to deflect. At 56 milliseconds the right-front tire impacted against Post 11, pushing the post over. As the tire interacted with the post, the tie-rod released and the tire steered 90 degrees toward the barrier. The left-front tire remained straight. At 60 milliseconds Post 10, which was upstream of the truck, began to twist such that the blockout was rotating in the downstream direction. At 98 milliseconds the front bumper was at Post 12. At 108 milliseconds the height of the w-beam at Post 12 began to reduce as the post and blockout began to rotate. At 112 milliseconds Post 12 was deflected significantly, the blockout on the post was rotated almost 90 degrees, and the post-bolt at Post 12 released. The rail began to drop and at 120 milliseconds the right-front corner of the truck bumper was visible above the rail. At 146 milliseconds Post 12 was pushed to the ground as the truck overrode the post and the front bumper of the truck at this time was completely over the top of the rail. At 174 milliseconds, the front of the truck was at Post 13; all the upstream posts had rotated essentially 90 degrees about their vertical axis at this time, which inferred substantial movement of the upstream anchor system. As the test continued, the front bumper continued to pass over the top of the rail. At 270 milliseconds Post 13 was pushed to the ground without releasing the post-bolt connection; the w- beam rail was pulled down with the post and the vehicle overrode the barrier. Model Setup – Including Pre-Damage The results from the low-speed impact case were used as initial conditions in a secondary high-speed impact simulation of full-scale crash test C08C3-027-2. The initial conditions for the pre-damaged guardrail included component deformations and residual stresses. These values were recorded for all shell, solid and beam elements in the model via the SPRINGBACK option in LS-DYNA. The residual forces for discrete elements (e.g., non-linear spring elements for the soil and rail end-boundaries), however, could not be recorded and saved with this option. For initialization of all the spring elements in the model, LS-PrePost was used to determine the final displacements of the spring nodes; these displacement values were then used to define the initial offset for each spring element in the FE model. This methodology directly imposed the force- deflection response of the spring elements for the soil and the end-anchor from the low-speed analysis onto the model for the high-speed impact case. Figure 224 shows a contour plot of effective plastic strain for the initial state of the guardrail model used in the simulation of Test C08C3-027-2.

280 Figure 224. Contours of effective plastic strain for the initial state of the guardrail model for simulation of Test C08C3-027-2. FEA Simulation Finite element analysis was then used to simulate the impact conditions of Test C08C3- 027-2. Two analysis cases were conducted:  Case 1: The crash-induced damage of the guardrail (including displacements and residual stresses) from the analysis of C08C3-027-1 was imposed directly onto the model as initial conditions. The post-bolt position at Post 12 was then re- positioned to the corner of slotted hole in the w-beam (i.e., to be consistent with the test article) and re-tightened. – Analysis Model: FEA_C08C3-027_HS5 – LS-DYNA version: smp s R6.1.2, revision 85139.  Case 2: The same initial conditions as Case 1. The model used in Case 1 was modified to include a finite rigidwall which effectively simulated the boundary at the backside of the soil pit at the test site. – Analysis Model: FEA_C08C3-027_HS7a – LS-DYNA version: smp s R6.1.2, revision 85139. Analysis Case 1 The vehicle model used in the analysis was the NCAC C2500D version 5B with modifications described in Task Report 4A-1. The vehicle model impacted the damaged rail at approximately 38.5 inches upstream of Post 11 at an impact speed and angle of 62.1 mph and

281 26.4 degrees, respectively (i.e., the same impact conditions as the full-scale test). At 25 milliseconds after impact, Post 11 started to deflect back, and at 55 milliseconds Post 12 began to deflect. At 60 milliseconds the right-front tire impacted against Post 11; however, there was no tire snag on the post and the tie-rod did not fail. At 80 milliseconds Post 11 reached its maximum lateral groundline deflection of 14 inches. At 50 milliseconds Post 10, which was upstream of the truck, began to twist such that the blockout was rotating in the downstream direction. At 100 milliseconds the front bumper was at Post 12, and Post 13 started to deflect. At 110 milliseconds the head of the post-bolt at Post 12 pulled through and released the rail; the rail did not begin to drop prior to release. At 120 milliseconds Post 14 began to deflect. At 130 milliseconds Post 12 reached its maximum groundline deflection of 12.3 inches; the post continued deflecting about the groundline until 135 milliseconds when the top of the post was pressed to the ground. The front tire of the vehicle contacted Post 12 at 140 milliseconds and proceeded to roll over the post and blockout. At 185 milliseconds, the front of the truck was at Post 13. At this time all the upstream posts had rotated significantly and the post-bolt connection had released at all upstream non- splice locations. These factors inferred substantial movement of the upstream anchor system. At 200 milliseconds the post-bolt at Post 13 pulled through the double-slot of the splice and released the rail; at this time the rail was still at its original height. Post 15 also began to deflect slightly at this time. At 210 milliseconds Post 13 reached its maximum groundline deflection of 15 inches, and at 215 milliseconds the top of the post contacted the ground. At 225 milliseconds the front tire on the vehicle contacted Post 13 and its blockout; the tire proceeded to ride over the post and passed behind the blockout. At 280 milliseconds the front of the vehicle was at Post 14. At 295 milliseconds the post- bolt at Post 14 released from the rail; at this time the rail was still at its original height. Also at this time deflection of Post 15 began to increase. The vehicle was parallel to the guardrail system at 315 milliseconds. Also at this time Post 14 reached its maximum groundline deflection of 10.5 inches. At 325 milliseconds the front tire on the vehicle contacted the post just as the top of the post contacted the ground; also, the deflection of Post 15 began to increase at this time. At 395 milliseconds the front of the vehicle was at Post 15. At 405 milliseconds the front fender of the vehicle impacted against the blockout at Post 15, and at 435 milliseconds the front tire of the vehicle contact the post. The tire steered toward rail as the vehicle passed Post 15. At 455 milliseconds Post 16 began to deflect. At 490 milliseconds Post 15 reached its maximum groundline deflection of 8.3 inches. At 500 milliseconds the vehicle lost contact with the rail as it was redirected from the system. At 552 milliseconds the vehicle reached its highest roll angle of 18.5 degrees. The analysis was terminated at 0.6 seconds, at which time:  The roll angle of the vehicle was 17.3 degrees and decreasing,  The pitch angle was -18.4 degrees and stable,  The yaw angle was 38.8 degrees relative to the barrier and increasing slightly, and  The forward velocity of the vehicle was 25 mph. Figure 238 and Figure 239 show sequential snapshots of the impact event from a downstream-oblique view point and an upstream viewpoint, respectively. The maximum plastic strain in the rail was 0.91 and occurred in the lower, down-stream splice-bolt hole at Post 13.

282 Figure 225. Sequential Views of Test C08C3-027-2 and FE analysis Case 1 from downstream-backside view perspective. 0.05 seconds 0.10 seconds 0.15 seconds 0.00 seconds 0.20 seconds FEA Case 1Test C08C3-027-2 Post 11 12 13 14

283 Figure 220. [CONTINUED] Sequential Views of Test C08C3-027-2 and FE analysis Case 1 from downstream-backside view perspective. 0.30 seconds 0.35 seconds 0.40 seconds 0.45 seconds 0.25 seconds FEA Case 1Test C08C3-027-2 Post 11 12 13 14

284 Figure 226. Sequential Views of Test C08C3-027-2 and FE analysis Case 1 from an upstream view perspective. 0.05 seconds 0.10 seconds 0.15 seconds 0.00 seconds 0.20 seconds Test C08C3-027-2 FEA Case 1

285 Figure 226. [CONTINUED] Sequential Views of Test C08C3-027-2 and FE analysis Case 1 from an upstream view perspective. 0.30 seconds 0.35 seconds 0.40 seconds 0.25 seconds 0.45 seconds FEA Case 1Test C08C3-027-2

286 The acceleration time-histories of the vehicle during the event are shown in Figure 228 through Figure 230 and the angular displacement-time histories are shown in Figure 231. Data from the accelerometer located at the center of gravity of the vehicle were collected and input into the Test Risk Assessment Program (TRAP) Version 2.3.2 to determine occupant risk factors. The data as provided directly from TRAP is presented in Figure 232. In the longitudinal direction, the occupant impact velocity was 16.4 ft/s (5.0 m/s) at 0.1584 seconds, the highest 0.010-second occupant ridedown acceleration was -8.5 g from 0.4146 and 0.4246 seconds, and the maximum 0.050-second average acceleration was -5.6 g between 0.3404 and 0.3904 seconds. The sudden drop in acceleration at times 0.13 seconds, 0.21 seconds, 0.27 seconds and 0.32 seconds corresponded to the release of post-bolts and buckling of posts. The maximum 0.010-second acceleration occurred when the vehicle impacted against Post 15, as shown in Figure 227. Figure 227. Vehicle impacts against Post 15 in analysis Case 1. In the lateral direction, the occupant impact velocity was 15.4 ft/s (4.7 m/s) at 0.1584 seconds, the highest 0.010-second occupant ridedown acceleration was -8.9 g from 0.2575 and 0.2675 seconds, and the maximum 0.050-second average acceleration was -5.2 g between 0.2203 - 0.2703 seconds. Figure 228. Longitudinal acceleration-time history at C.G. of pickup truck model in local coordinates for analysis Case 1. Post 15 X Acceleration at CG 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -15 -10 -5 0 5 Time (sec) L o n g it u d in a l A c c e le ra ti o n ( G ) Time of OIV (0.158367 sec) SAE Class 60 Filter 10-msec average Post-Bolt 12 released Post-Bolt 13 released Post 14 buckled Post-Bolt 14 released

287 Figure 229. Lateral acceleration-time history at C.G. of pickup truck model in local coordinates for analysis Case 1. Figure 230. Vertical acceleration-time history at C.G. of pickup truck model in local coordinates for analysis Case 1. Figure 231. Acceleration-time history at C.G. of pickup truck model in local coordinates for analysis Case 1. Y Acceleration at CG 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -10 -8 -6 -4 -2 0 2 4 Time (sec) L a te ra l A c c e le ra ti o n ( G ) Time of OIV (0.158367 sec) SAE Class 60 Filter 10-msec average Z Acceleration at CG 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -15 -10 -5 0 5 10 Time (sec) V e rt ic a l A c c e le ra ti o n ( G ) SAE Class 60 Filter 10-msec average Roll, Pitch and Yaw Angles 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -50 -40 -30 -20 -10 0 10 20 30 Time (sec) A n g le s ( d e g re e s ) Roll Pitch Yaw

288 Figure 232. Summary report of occupant risk measures for the analysis Case 1. General Information Test Agency: RoadSafe, LLC Test Number: FEA_C08C3-027-2_5 Test Date: Test Article: Damaged G4(1S) with 15.5 inches initial deflection Test Vehicle Description: C2500D-v5b-R131202.k Test Inertial Mass: 4472 lbm Gross Static Mass: 4472 lbm Impact Conditions Speed: 100.0 km/h Angle: 26.4 degrees Occupant Risk Factors Impact Velocity (m/s) at 0.1584 seconds on right side of interior x-direction 5.0 y-direction 4.7 THIV (km/hr): 23.4 at 0.1525 seconds on right side of interior THIV (m/s): 6.5 Ridedown Accelerations (g's) x-direction -8.5 (0.4146 - 0.4246 seconds) y-direction -8.9 (0.2575 - 0.2675 seconds) PHD (g's): 10.7 (0.2231 - 0.2331 seconds) ASI: 0.72 (0.3411 - 0.3911 seconds) Max. 50msec Moving Avg. Accelerations (g's) x-direction -5.6 (0.3404 - 0.3904 seconds) y-direction -5.2 (0.2203 - 0.2703 seconds) z-direction 2.0 (0.4172 - 0.4672 seconds) Max Roll, Pitch, and Yaw Angles (degrees) Roll 21.7 (0.5603 seconds) Pitch 7.9 (0.4453 seconds) Yaw -42.3 (0.5999 seconds)

289 Comparing the results of Analysis Case 1 to the full-scale test (C08C3-027-1) revealed that the groundline deflections of Posts 11 and 12 were notably higher in the analysis. This phenomenon resulted in the posts in the analysis rotating about a lower point below grade, which affected the tire’s interaction with Post 11 and the height of the w-beam as the posts deflected and rotated back. After further review of the full-scale test it was determined that the limited width of the soil pit at the test site may have restricted the posts’ deflection through the soil. Figure 233 shows the test setup for the low-speed Ttest C08C3-027-1 illustrating the limited distance between the back of the posts and the back edge of the soil pit. The ground surface outside the soil pit area was composed of asphalt material. From a review of the high-speed video of the test, the lateral deflection of Post 11 was stopped abruptly at around 160 milliseconds of the impact event. Figure 234 shows snapshots from the test video for low-speed test C08C3-027-1 illustrating the position of the guardrail posts relative to the backside of the soil pit at the beginning of the test and at 0.16 seconds. Post 11 showed no signs of further deformation (e.g., torsional deflection, collapsing of the post cross- section, or further bending at the groundline). It was therefore assumed that the soil pit area was sufficient for the low-speed test. Figure 233. Test setup for low-speed test C08C3-027-1 illustrating the limited distance to the back-edge of the soil-pit for the test article. Edge of soil pit

290 Figure 234. Snapshots from low-speed test C08C3-027-1 illustrating the position of the guardrail posts relative to the backside of the soil pit at the beginning of the test and at 0.16 seconds. Figure 235 shows the test setup for high-speed test C08C3027-2. The initial deflections for Posts 11 and 12 were such that the back of the posts were already in contact (or nearly in contact) with the back edge of the soil pit prior to the start of the test. The reduced distance from the back of Posts 11 and 12 to the back-edge of the soil pit restricted the movement of the posts at the groundline. Figure 235. Test setup for high-speed test C08C3-027-2 illustrating the reduced distance from the back of Posts 11 and 12 to the back-edge of the soil-pit. Analysis Case 2 As mentioned previously, analysis Case 2 included the same model used in analysis Case 1, except that the backside of the soil pit in the test article installation was included in the model. The soil-pit boundary was modeled using the “finite rigidwall” option in LS-DYNA. The lateral offset of the rigid wall relative to the guardrail was approximated based visual inspection of an Time = 0.16 secondsTime = 0.00 seconds Apparent Edge of Soil PitPost 11 Post 12

291 overhead view of the test article just before impact, as shown in Figure 235. The resulting model is shown in Figures 236 and 237 from an overhead and a downstream viewpoint, respectively. Figure 236. FE model for Case 2 with vertical “rigidwall” located just below grade to simulate back-edge of soil pit (overhead viewpoint). Figure 237. FE model for Case 2 with vertical “rigidwall” located just below grade to simulate back-edge of soil pit (downstream viewpoint). At 30 milliseconds after impact Post 11 started to deflect back, and at 55 milliseconds Post 12 began to deflect. At 60 milliseconds the right-front tire impacted against Post 11 and began to push the post back as it rode over the post; the tie-rod did not fail as the wheel steered slightly toward the guardrail during interaction with the post. The post-bolt connection at Post 11 (location of a w-beam splice connection) released at 65 milliseconds. Post 11 was pushed completely to the ground at 95 milliseconds. At 50 milliseconds Post 10, which was upstream of the truck, began to twist such that the blockout was rotating in the downstream direction. At 85 milliseconds the posts upstream (e.g., Post 13, Post 14, etc.) began to twist, such that the blockouts where rotating in the upstream direction. At 100 milliseconds the front bumper was at Post 12. The w-beam at Post 12 began to drop at 110 milliseconds. At 115 milliseconds the upstream anchor reached its maximum deflection of 5.5 inches, and the post-bolt at Post 12 Vertical “rigidwall” located just below grade Vertical “rigidwall” simulating the back-edge of soil pit

292 failed releasing both the rail and the blockout. Also at this time Post 12 was twisted almost 90 degrees. Between 115 and 120 milliseconds the right-front corner of the truck bumper was visible above the rail and Post 14 began to deflect laterally. The tire of the vehicle impacted Post 12 at 145 milliseconds and continued to push the post to the ground as it rode over the post. At 185 milliseconds, the front of the truck was at Post 13. At this time all the upstream posts had rotated significantly; however, none of the upstream post-bolt connections had released. Post 13 continued to deflect and at 220 milliseconds the top of the post contacted the ground with the w- beam still attached. At 225 milliseconds the right front tire of the vehicle impacted against the blockout of Post 13 and began to climb the rail (the top of the rail was 17 inches above ground at this time). At 275 milliseconds the front of the vehicle was at Post 14. At 320 milliseconds the right front tire overrode the rail. At 430 milliseconds the rear tire of the vehicle overrode the rail at Post 13 and the post-bolt at Post 13 released from the w-beam. Figure 238 and Figure 239 show sequential snapshots of the impact event from a downstream-oblique view point and an upstream viewpoint, respectively. Table 2 provides a list of phenomenological events and their time of occurrence for both the full-scale test and the FE analysis. The model appears to reasonably simulate the basic kinematic behavior of the pickup until approximately 0.2 seconds of the impact event. After this time there was noticeable difference in the attitude of the vehicle, particularly in the yaw and pitch angles and the vertical trajectory. The front bumper and the front tire of the vehicle overrode the rail 0.05 seconds later in the analysis than it did in the full-scale test. As a result, the front of the vehicle remained in contact with the rail for a longer period of time in the analysis, resulting in greater yaw and pitch angles. This phenomenon is clearly evident in the sequential views beginning at 0.25 seconds. Of primary concern, however, was that the post-rail connection at Post 13 did not release properly as the system deflected, which allowed the vehicle to override the guardrail in both the test and the analysis. Note that the only difference in this case, compared to Case 1, was the inclusion of the rigid boundary representing the soil pit wall, which limited the groundline deflection of the posts and resulted in the posts rotating about the top edge of the soil pit. Based on the results of the analysis, it is the authors’ opinions that the override was likely the result of a combination of two events: 1) as the posts deflected laterally, the shorter radius of rotation of the posts resulted in the rail-height dropping more quickly, and 2) the low stiffness of the end- terminal resulted in lower tension in the rail, which consequently reduced the forces acting between the post-bolt and rail. That is, without sufficient tension in the rail there is not enough force to pull the post-bolt head through the w-beam slot and release the rail from the post, so the rail simply follows the deflection of the post to the ground – similar to a slack cable.

293 Figure 238. Sequential Views of Test C08C3-027-2 and FE analysis from downstream- backside view perspective. 0.05 seconds 0.10 seconds 0.15 seconds 0.00 seconds 0.20 seconds FEA Case 2Test C08C3-027-2 Post 11 12 13 14

294 Figure 238. [CONTINUED] Sequential Views of Test C08C3-027-2 and FE analysis from downstream-backside view perspective. 0.30 seconds 0.35 seconds 0.40 seconds 0.45 seconds 0.25 seconds FEA Case 2Test C08C3-027-2 Post 11 12 13 14

295 Figure 239. Sequential views of Test C08C3-027-2 and FE analysis from an upstream view perspective. 0.05 seconds 0.10 seconds 0.15 seconds 0.00 seconds 0.20 seconds FEA Case 2Test C08C3-027-2

296 Figure 239. [CONTINUED] Sequential views of Test C08C3-027-2 and FE analysis from an upstream view perspective. 0.30 seconds 0.35 seconds 0.40 seconds 0.25 seconds 0.45 seconds FEA Case 2Test C08C3-027-2

297 Table 66. Summary of phenomenological events of full-scale test C08C3-027-2 and FEA simulation. Event Test C08C3-027-2 FE Analysis Post 11 began to deflect 0.016 sec 0.025 – 0.030 sec Post 12 began to deflect 0.044 sec 0.050 – 0.055 sec Right-front tire impacted Post 11 0.056 sec 0.055 – 0.060 sec Post 10 began to twist clockwise 0.060 sec 0.050 – 0.055 sec Post-bolt at Post 11 released 0.060 – 0.070 sec 0.060 – 0.065 sec Post 11 pushed to the ground 0.088 sec 0.090 – 0.095 sec Post 13 began to twist counter-clockwise 0.088 sec 0.085 – 0.090 sec Front of vehicle was at Post 12 0.098 sec 0.095 – 0.100 sec W-beam at Post 12 began to drop 0.108 sec 0.105 – 0.110 sec Post-bolt at Post 12 released 0.112 sec 0.110 – 0.115 sec Maximum displacement of upstream anchor N.R. 5.5 in @ 0.115 sec Right-front corner of bumper was visible over top of rail 0.120 sec 0.115 – 0.120 sec Post 14 began to deflect laterally 0.128 – 0.130 sec 0.115 – 0.120 sec Post 12 was pushed to the ground 0.146 sec 0.145 – 0.150 sec Front bumper was completely overtop of rail 0.146 sec 0.225 – 0.230 sec Front of vehicle was at Post 13 0.174 sec 0.180 – 0.185 sec Post 13 was pushed to the ground without releasing rail 0.240 sec 0.220 – 0.225 sec Maximum displacement of downstream anchor N.R. 3.4 in @ 0.240 sec Front-right tire overrode the rail 0.270 sec 0.320 – 0.325 sec Rear-right tire overrode the rail 0.425 – 0.430 sec 0.425 – 0.430 sec Post-bolt at Post 13 released N.A. 0.425 – 0.430 sec N.R.: Not reported. Although the maximum deflection of the end-terminal was not reported in the full-scale tests, the magnitude could be approximated from the post-test photos of the deflected anchor post. Figure 240 shows the post-test photo of the upstream anchor in Test C08C3-027-2. The permanent deflection of the anchor post at the groundline was estimated from the photo to be approximately 8-9 inches, and the deflection of the post at the post-bolt location was approximately 13-14 inches. The maximum dynamic and permanent deflections of the rail boundary in the analysis, as shown in Figure 241, were 5.5 inches and 3.7 inches, respectively, which were considerably lower than those of the full-scale test. This difference in anchor deflection is also apparent from the maximum rotation of the upstream posts, as illustrated in Figure 242. Since the end-terminal was modeled with less than half the stiffness of a standard two-post anchor system, it is not apparent why the displacement was so much greater in the test. Regardless, the seemingly low magnitude of rail tension in the full-scale test was likely a major contributing factor in how quickly and easily the rail was pulled down allowing the vehicle to override the guardrail.

298 Figure 240. Permanent displacement of the upstream anchor in Test C08C3-027-2. Figure 241. Displacement of upstream rail boundary in FEA analysis Case 2 (modeled with 47% baseline anchor stiffness). 13-14 in 8-9 in 5.5 in 3.7 in(a) Max dynamic (b) Permanent

299 Figure 242. Maximum rotation displacement of Post 10 in (a) full-scale test and (b) FEA analysis Case 2. Summary and Discussion The purpose of this task was to resolve any discrepancies or to confirm the differences between the results in Chapter 10 for the G4(2W) guardrail and those in Report 656 which were based on damages to the G4(1S) guardrail. The primary discrepancy was related to the cause of failure. The results in Chapter 10 indicated that the most significant effect of pre-existing crash- induced rail deflection on the performance of the G4(2W) guardrail system was an increased potential for rail rupture. For example, Test 3-11 impact on the G4(2W) with an initial rail deflection of 14 inches resulted in maximum plastic strain values of 1.4 in the w-beam rail at the splice-bolt holes. The analyses for the G4(2W) also indicated that the potential for override was relatively low for all cases investigated. This was contradictory to the full-scale crash test results for the G4(1S) guardrail performed in Gabler’s study, where a post-bolt did not release properly resulting in the rail being pulled down with one of the guardrail posts, allowing the vehicle to override the barrier. The results in Chapter 10 lead to recommendations for future work to include analyses of both the G4(2W) and G4(1S) with pre-existing crash-induced rail deflections in combination with varying anchor strength. That activity was beyond the scope of this project; however, it was decided that the results of full-scale test C08C3-027-2 would be further investigated to determine if the cause of vehicle override was the combination of rail deflection and weak anchor rather than rail deflection and an “over-strong” post bolt connection. Due to the differences in torsional rigidity of the posts for these two systems it was expected that the G4(1S) would have a greater (a) Test C08C3-027-2 (b) FEA Case 2

300 sensitivity to anchor strength. This was further evidenced from the results of Test MGA C08C3- 027-2 on that system which resulted in significant anchor deflection and vehicle override. A model of the G4(1S) guardrail was developed and finite element analysis was used to simulate full-scale crash test C08C3-027-1 to create low-level crash-induced damage of the system. The results from this low-speed impact case, including guardrail component deformations and residual stresses, were then used as initial conditions in a secondary high-speed impact simulation of full-scale crash test C08C3-027-2. In these analyses, the upstream and downstream end-terminals for the system were modeled using non-linear springs attached to the ends of the rail. The single-foundation-tube anchor system in the full-scale tests was modeled via non-linear springs characterized by scaling the force-deflection properties of the baseline anchor defined in Chapter 9 by approximately 50 percent. From review of the full-scale test video it was determined that there may have been an additional factor influencing the results of the test; specifically, an insufficient width of the soil pit behind the posts. It was apparent from the video that two of the posts contacted the backside of the pit during the test which arrested the lateral deflection of the posts at the groundline. Thus, two analysis cases were conducted: Case 1 which did not include the rigid boundary of the soil pit, and Case 2 which did include this boundary condition via the “finite rigidwall” option in LS- DYNA. For analysis Case 1 with 15.5 inches of initial guardrail deflection and without the influence of the restricted soil pit size, the system successfully contained and redirected the 2000P pickup under NCHRP Report 350 impact conditions with no “apparent” likelihood for vehicle override. Case 2, on the other hand, which included the soil pit boundary, resulted in the vehicle overriding the barrier in a similar manner to that which occurred in the full-scale test. In both analysis cases, however, the deflection of the anchor was significantly less than the anchor deflection in the full-scale test. The force-deflection characterization of the anchor used in the FEA model is shown in Figure 243. Also shown in the figure is the static force- deflection responses for a two-foundation-tube anchor (i.e., Test 13001B) and a single- foundation-tube anchor (i.e., Test 14001D on a standard anchor with groundline strut removed) measured via quasi-static tests in Chapter 124. The plot shows that the anchor stiffness characterization of the FEA model was similar to that of Test 14001D up to approximately 5 inches deflection, at which point the resistance in the physical test suddenly dropped significantly. In the FE analyses, the deflection of the upstream anchor reached magnitudes of 11 inches and 5.5 inches for Case 1 and Case 2, respectively. It can therefore be assumed that the anchor system would likely have failed (i.e., experienced significant reduction in stiffness) in both of these analysis cases, which would have resulted in reduced tensile force in the w-beam and increased the potential for vehicle override. 4 The physical tests conducted in Task 4B were performed after this study was completed.

301 Figure 243. Anchor stiffness used in FEA model compared to anchor stiffness measured in physical tests. The maximum plastic strain in the w-beam was 0.91 and was located at the edge of a splice-bolt hole. Strains of this magnitude for steel are generally associated with a high potential for material failure. These strains, however, are restricted to a very localized area at the end of the splice-bolt holes and are compressive (i.e., caused from the bearing load between the splice- bolt and the edge of the w-beam hole), which results in a lower potential for tear initiation – compared to the same magnitude of strain under tension. Further, the maximum strains in these analyses were considerably less than those from similar analysis cases involving the G4(2W). For example, the maximum effective plastic strains in the splice for the G4(2W) with 14 inches initial deflection was 1.4 (see Chapter 10); and the maximum effective plastic strains in the splice for the G4(2W) with weak anchor was 1.2 (see Chapter 9). Conclusions The analyses showed that the simulation of the full-scale test, including the effects of the single-foundation-tube anchor and the limited size of the soil pit in the model, resulted in essentially the same response as the test, in which the rail was pulled to the ground with Post 13 without releasing the post-bolt connection allowing the vehicle to override the rail. Thus, based on the results of the analyses presented herein and the subsequent assessment of the characterization of the anchor stiffness used in the FE model, it was concluded that the vehicle override in full-scale test C08C3-027-2 on the G4(1S) was likely augmented by the limited width of the soil pit as well as the reduced strength of the anchor system. However, further analyses or full-scale testing would be required to determine conclusively if the system would have been successful without these influences. It was also ascertained that the influence of pre-existing crash-induced rail deflection was somewhat different for the G4(1S) compared to the G4(2W). That is, the G4(2W) has a higher probability for rail rupture for this damage mode; whereas, the G4(1S) is more likely to result in 0 5 10 15 20 25 30 0 2 4 6 8 10 12 Fo rc e (k ip s) Longitudinal Deflection at Rail Height (in) Baseline (Test 13001B) Missing Strut (Test 14001D) FEA_C08C3-027 - Rail Boundary

302 the vehicle overriding the rail – particularly for cases when the movement of the posts in the soil is restricted (e.g., frozen soil, posts driven through asphalt, etc.). Recommendations Additional analyses should be conducted, in which the end-anchors are modeled using the revised stiffness and strength properties measured in Chapter 13 for the single-foundation- tube anchor system, to confirm the conclusions made herein regarding analysis results. Additional analyses may also be warranted to evaluate the performance of the G4(1S) with the standard anchor strength with and without the influence of the restricted soil pit size. As discussed in Chapter 10, future work should also include analyses of both the G4(2W) and G4(1S) with pre-existing crash-induced rail deflections in combination with varying anchor strength. Due to the low torsional rigidity of the W6x9 steel posts of the G4(1S) guardrail it is expected that the G4(1S) may have a greater sensitivity to anchor strength. Also, the low-severity impact case evaluated herein involved low speed and a high impact angle which resulted in damage to a relatively localized section of the guardrail; i.e., the damaged area spanned only 3 to 4 posts. It is not known how the guardrail will respond to subsequent impacts when the rail deflections are spread over a longer length of the guardrail. It is recommended that future studies on the effects of pre-existing crash-induced rail deflections include higher impact speeds and smaller impact angles to create initial guardrail damage with similar magnitudes of rail deflection spread over a longer length of the guardrail.