Click for next page ( 34


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 33
33 A pilot study was conducted to demonstrate the feasibility ever, one of the limitations associated with using current finite of using finite element simulation for this research. Simulation element analysis codes to model concrete barriers relates to the was used as a tool to develop a set of preliminary guidelines material behavior. At the time of undertaking this research, that defined relationships between different design parameters there were no available robust concrete material models that for aesthetic surface treatments on safety shape barriers. Once could accurately and efficiently capture the failure/fracture of these preliminary guidelines were established, a full-scale concrete. Even though the FHWA was at that time sponsoring crash-testing effort was conducted. The initial crash-tested the development of such a material model, it was not available configurations were picked based on simulation results. The in time for use in this project. However, since the concrete bar- results from these crash tests were analyzed in conjunction rier profiles of interest in this project had been successfully with the preliminary guidelines to determine the asperity crash tested and their structural adequacy was not at issue, this geometries to be evaluated in subsequent crash tests. This pro- was not considered as a significant limitation. For this reason, cedure maximized the information available for adjusting the modeling the concrete barriers as a rigid material without fail- preliminary guidelines to yield the final design guidelines. ure was considered a reasonable and practical assumption. Simulations were performed using LS-DYNA. LS-DYNA Most of the effort devoted to the validation effort therefore is a general-purpose, explicit-implicit, nonlinear finite element focused on the vehicle models. program capable of simulating complex nonlinear dynamic The validity of the improved 820-kg passenger car and impact problems. LS-DYNA has been used extensively in 2,000-kg pickup truck vehicle models was established by simulations involving vehicular impacts with roadside safety comparing the results of simulations with the results of full- appurtenances, including safety shape barriers. The decision to scale crash tests. It should be noted that an accurate compari- choose this explicit finite element code for this research was son of a simulation with a successful test does not necessarily based on several reasons, including: constitute validation. It is important that some of the tests The availability of vehicle models that correspond to selected for use in the validation study include relevant failure NCHRP Report 350 design test vehicles (mainly the 820C modes. The two most critical failure modes associated with and 2000P vehicles). These vehicle models have been the performance evaluation of longitudinal barriers are vehic- used for roadside safety applications for the last 6 or more ular instability (i.e., rollover) and OCD. While the validation years, and their fidelity and limitations are reasonably study focused on these two evaluation criteria, other vehicular understood. acceleration-based criteria were also analyzed and compared. The ability to model the geometry of the safety shape barriers and the details of the aesthetic surface treatment 820C Vehicle Model (which affects the mechanics of the vehicle-barrier inter- action) with a high degree of fidelity. The researchers identified historical crash tests that could The availability of a large contact algorithm library. These be used to assist with vehicle model validation in the pilot contact algorithms provide means to model vehicular col- study. The number of crash tests useful for this purpose was lisions with roadside objects. very limited. One of the first simulations that the researchers performed was that of Caltrans Test No. 582.(19) In this test, PILOT STUDY AND FINITE ELEMENT a 1990 Geo Metro impacted a single-slope concrete barrier MODEL VALIDATION with an inclined fluted surface at a speed of 100 km/h at an angle of 20 degrees. These testing conditions conform to In all types of modeling, approximations must be made the impact conditions for Test 3-10 in NCHRP Report 350 when trying to represent reality. If finite element analysis is (see Figure 33). The slope of the single-slope barrier was to be used to assess the effects of surface treatments on con- 9.1 degrees from vertical, and the overall height of the barrier crete median barriers, the vehicle and barrier models must be capable of capturing the dynamic response associated with a concrete barrier impact. This capability was investigated by performing a pilot validation study. The pilot study had the following three objectives: Perform finite element simulation of previously available crash tests so as to identify potential modeling problems. Make necessary changes to improve the performance of the vehicle models and validate them for use in evaluat- ing the performance of barriers with surface asperities. Identify surrogate measures for assessing OCD. Accurately representing the geometry of a concrete barrier Figure 33. Caltrans single-slope barrier with fluted and added surface asperities is fairly straightforward. How- surface texture.

OCR for page 33
34 was 1.42 m. The surface of the barrier was modified to incor- ment impact simulation of the barrier using the modified Geo porate inclined flutes or ribs. The flutes were oriented at a 45- Metro model. A comparison of roll, pitch, and yaw angles degree angle from the ground, rising in the direction of vehi- versus time is shown in Figures 36 through 38, respectively. cle travel. The cross section of each flute was 19 mm high A significant improvement in correlation of the roll angle and 19 mm wide. The flutes were spaced 50.8 mm on the cen- was achieved, but with minor divergence in the pitch angle ter along the length of the barrier. The vehicle was redirected correlation. but rolled over as it exited the barrier. The vehicle model used in the initial simulation of this impact was the reduced Geo Metro model that was developed Single-Slope Barrier by the National Crash Analysis Center (NCAC) under FHWA sponsorship. This model contains approximately 16,100 ele- After the modified Geo Metro model demonstrated the ments. Initial simulation results did not show a good correla- ability to capture interaction with the inclined asperities on tion with the test results. Several changes were made to the the fluted single-slope barrier, a baseline simulation using a original model to improve its performance in interacting with smooth-faced single-slope barrier was performed. The pur- the surface asperities. Changes focused primarily on the vehi- pose was to verify that the changes made to the Geo Metro cle's front suspension and the tires. The suspension was mod- model did not adversely affect other areas associated with ified to include deformable control arms and some of the other concrete barrier impacts. Figures 39 through 41 compare the linkages for the suspension mechanism (see Figure 34). Sim- angular displacements of the vehicle obtained from the crash ulation results with the modified 820C vehicle model showed test(20) and simulation of the single-slope barrier. Improved better correlation with the Caltrans fluted-surface, single- correlation was observed with the modified model for both the slope concrete barrier test results. roll and pitch behavior. Both models showed good correlation In addition to the Caltrans fluted barrier, a smooth single- with the test data for the yaw angle. slope barrier was used as a baseline system to validate the modified Geo Metro model. A comparison of vehicle dynam- ics between crash tests and finite element simulations of the Summary of the 820C Vehicle Model Validation Caltrans fluted single-slope barrier and those of standard single-slope barrier follows. At the start of the simulation study, a significant effort was put into the improvement and validation of the 820-kg, small- car model for impacts into single-slope barriers with and with- Caltrans Single-Slope Barrier with Angled Flutes out surface asperities. As described above, results from full- scale crash tests performed by Caltrans were used to help Figure 35 shows sequential images comparing the actual assess validity of the model for this purpose. The suspension crash test of the fluted single-slope barrier with the finite ele- on the original reduced Geo Metro model was extensively (a) (b) Figure 34. 820C front suspension: (a) modified model; (b) actual vehicle.

OCR for page 33
35 Figure 35. Sequential comparison of test and simulation of angle fluted barrier. Figure 36. Comparison of roll angles from crash data with vehicle simulations for the angle fluted barrier.

OCR for page 33
36 Figure 37. Comparison of pitch angles from crash data with vehicle simulations for the angle fluted barrier. Figure 38. Comparison of yaw angles from crash data with vehicle simulations for the angle fluted barrier.

OCR for page 33
37 Figure 39. Comparison of roll angles from crash data with vehicle simulations for the single-slope barrier. Figure 40. Comparison of pitch angles from crash data with vehicle simulations for the single-slope barrier.

OCR for page 33
38 Figure 41. Comparison of yaw angles from crash data with vehicle simulations for the single-slope barrier. modified, and this modified version of the Geo Metro was con- TTI librarian and the FHWA, the report and film for this test sidered to be adequately validated against single-slope barrier could not be located. tests with and without surface asperities (i.e., angled flutes) to TTI researchers ultimately simulated impacts of the Geo proceed with its use in this project. Metro into an F-shape barrier and the New Jersey safety shape At the interim panel meeting, the focus of the research barrier with the 75-mm reveal/lip covered by a pavement changed from single-slope barrier to New Jersey safety shape overlay. It was discovered that for these safety shape barriers, barriers. Consequently, the validation of the small-car finite neither the original NCAC model nor the TTI-modified model element model had to be revisited. The number of crash tests exhibited adequate correlation. Although correlation was into rigid concrete safety shape barriers with 820-kg cars was achieved with the single-slope barrier, the safety shape barri- found to be very limited. Some New Jersey safety shape bar- ers interact differently with the vehicle's tires, wheels, sus- rier tests that were identified were conducted on a modified- pension, and so forth, and a model validated for one barrier barrier profile in the early 1980s under NCHRP Report 230. shape will not necessarily work for another barrier shape. The barrier modification consisted of a 75-mm pavement Further, the tests that were simulated were conducted with a overlay in front of the barrier that covered the 75-mm Honda Civic rather than a Geo Metro. Therefore, it is not reveal/lip at the bottom edge of the barrier. Further, the tests known how much of the observed differences between the were conducted with a different vehicle (i.e., Honda Civic) tests and simulations were attributed to differences in vehicle- at a 15-degree angle rather than the 20-degree angle currently barrier interaction versus differences in vehicle type. specified in NCHRP Report 350. These factors limited the While the validation effort for the 820C vehicle was going usefulness of these tests for validating the Geo Metro model on, TTI researchers were working in parallel on the 2000P for NCHRP Report 350 impacts into a New Jersey safety vehicle model validation, details of which follow in subse- shape barrier. quent sections in this chapter. As mentioned previously, the A reference to a 1981 test of an unmodified New Jersey 2000P pickup truck design vehicle is believed to be more crit- safety shape barrier at a 20-degree impact angle was identi- ical than the 820C in regard to evaluation of OCD and stabil- fied. The test was conducted by Dynamic Science, Inc., under ity in impacts with concrete barriers. As an example, con- FHWA contract DOT-FH-11-9115. Despite considerable sider the Texas T411 aesthetic bridge rail shown in Figure 42. efforts by the research team, including consultation with the An impact into this barrier with an 820-kg passenger car at

OCR for page 33
39 ing the panel's desire to revisit the validation of the small-car model for impacts with safety shape barriers. The research team believed that with additional time and resources, improved correlation of the Geo Metro for impacts into safety shape barriers could be achieved through further modification to the model. However, the benefits derived from such an effort needed to be weighed against the cost of the effort and the delay it would have imposed on the full-scale crash-testing program. Use of the small-car model was limited to providing a check of the preliminary guidelines established by the pickup truck. This objective could also be accomplished with a full- scale crash test. This approach was approved by the project panel. Conse- quently, further validation of the small car for the New Jersey Figure 42. Texas T411 aesthetic bridge rail. shape barrier was discontinued, and a crash test was performed to check the validity of the guidelines for the small car. Details of the crash-testing phase are presented in Chapter 6. 97 km/h and 21.2 degrees was successful, while an impact into this barrier with a 2000-kg pickup truck at 101 km/h and 2000P Vehicle Model 24.9 degrees failed due to excessive OCD (see Figure 43). Since the small car was not considered to be the critical Initial validation efforts for the 2000P vehicle were carried design vehicle from the standpoint of evaluating OCD or sta- out with the reduced element pickup truck model that was bility, the researchers planned to use the pickup truck as the developed by the NCAC. Simulations with the vehicle impact- primary vehicle for developing the preliminary guidelines. ing a smooth-surface, single-slope barrier and a New Jersey The role of the small-car model was to be limited to check- shape barrier were performed and compared with available ing the preliminary guidelines established by the pickup crash test data.(21,22) The correlation between test and simula- truck. Therefore, rather than undertaking another extensive tion was not considered acceptable. Certain vehicle suspension effort to improve the validity of the Geo Metro model for parts in the reduced vehicle model (e.g., control arms) are impacts into the New Jersey safety shape barrier while retain- modeled as rigid materials. The lack of deformability in ing sufficient fidelity to detect surface asperities, the research the front suspension was believed to be the primary cause team shifted its focus to the development of preliminary guide- of the observed discrepancies between test and simulation. lines based on parametric simulations with the pickup truck TTI researchers then simulated these crash tests using model. the NCAC detailed pickup truck model. This model, which Once the preliminary guidelines were established based on contains approximately 54,800 elements, incorporates a the parametric simulations conducted with the pickup truck, deformable front suspension. A comparison of the vehicle the research team sought input from the project panel regard- dynamics resulting from the simulation and crash test showed reasonable correlation for both barriers. Figures 44 through 46 compare the roll, pitch, and yaw displacements, respectively, for the single-slope barrier. Figures 47 through 49 provide a similar comparison of angular displacements for the New Jersey safety shape barrier. While reasonable correlation was obtained for both barriers, it can be seen from these figures that the single-slope barrier showed better correlation. This is likely due to the more prominent role of the vehicle suspension in impacts with the New Jersey shape barrier and limitations in the suspension model of the finite element pickup truck. Having demonstrated reasonable correlation, the detailed NCAC pickup truck model was selected for use in the devel- opment of the preliminary design guidelines. Surrogate Measure of OCD Figure 43. Pickup truck after impact with Texas T411 As mentioned previously, a common cause of barrier failure bridge rail. in a crash test is excessive OCD. As an example, OCD failure

OCR for page 33
40 Figure 44. Comparison of roll angles of crash data with detailed pickup truck vehicle simulation on the single-slope barrier. Figure 45. Comparison of pitch angles of crash data with detailed pickup truck vehicle simulation on the single-slope barrier.

OCR for page 33
41 Figure 46. Comparison of yaw angles of crash data with detailed pickup truck vehicle simulation on the single-slope barrier. Figure 47. Comparison of roll angles of crash data with detailed pickup truck vehicle simulation on the New Jersey safety shape barrier.

OCR for page 33
42 Figure 48. Comparison of yaw angles of crash data with detailed pickup truck vehicle simulation on the New Jersey safety shape barrier. Figure 49. Comparison of pitch angles of crash data with detailed pickup truck vehicle simulation on the New Jersey safety shape barrier.

OCR for page 33
43 was the most predominant type of failure in the Caltrans study, structed for the crash test and the associated LS-DYNA model "Crash Testing of Various Textured Barriers."(19) There has used in the simulation of the system. been little research performed to assess or improve the ability of vehicle models to accurately capture and predict OCD Deep Cobblestone Barrier. The deep cobblestone barrier resulting from a barrier impact. Part of the pilot study con- (shown in Figure 51) is a single-slope barrier with a random ducted under this project was devoted to assessing the ability cobblestone surface treatment. This barrier was tested by of existing vehicle models to predict OCD, either through Caltrans as part of its project to develop guidelines for aes- direct measurement of the maximum deformation inside the thetic surface treatments for single-slope barriers. The bar- passenger compartment (similar to the procedure used in a rier failed the test due to excessive OCD of the pickup truck crash test), or by means of a surrogate measure correlated caused by the interaction of the wheel with the cobblestones. against the OCD measurements obtained in full-scale crash The maximum amount of relief on the cobblestone surface tests. was 64 mm. Several crash tests of concrete barriers with the 820C pas- For the simulation, the cobblestone surface was modeled senger car and the 2000P pickup truck were identified. How- using hemispherical and ellipsoidal shapes with the same ever, the number of useful crash tests was limited, especially depth as the actual surface treatment. Because this was one for the small car. This was because OCD was not measured of the few pickup truck tests with a solid (i.e., without win- and reported prior to the publication and adoption of NCHRP dows) concrete barrier that failed due to excessive OCD, it Report 350 and many of the small-car compliance tests with provided a useful data point for correlation of the surrogate standard concrete median barrier shapes were conducted before OCD measures. NCHRP Report 350. All of the identified concrete barrier crash tests with measured OCD were modeled and simulated. Each simulation was set up to collect several potential measures of Shallow Cobblestone Barrier. After the failure of the OCD. The objective was to determine a measure that would deep cobblestone barrier, the depth of the cobblestone sur- demonstrate the best correlation with the maximum OCD face treatment was reduced to 19 mm and retested by Cal- reported in the crash tests. trans. Typical relief of the shallow cobblestone surface is shown in Figure 52. In the pickup truck crash test of this bar- Simulated Barrier Designs rier, the drive shaft became dislodged from the transmission. Oregon Bridge Railing. The Oregon bridge rail is a con- Although the vehicle remained upright during the test, this crete beam and post bridge rail similar to the Texas T411. type of damage was considered by Caltrans to represent a When the impact performance of this barrier was evaluated potential rollover risk. As a result, Caltrans decided that the with a pickup truck, the OCD was 475 mm, which signifi- barrier did not meet NCHRP Report 350 evaluation criteria. cantly exceeded the 150-mm limit imposed by the FHWA.(23) However, since the shallow cobblestone reduced the maxi- Therefore, this test served as one of the failure points in the mum OCD of the vehicle to within acceptable limits, this test OCD pilot study. Figure 50 shows an image of the rail con- illustrated the effect of surface asperity depth on vehicle (a) (b) Figure 50. Oregon bridge railing: (a) actual; (b) simulation model.

OCR for page 33
44 (a) (b) Figure 51. Cobblestone barrier: (a) actual; (b) simulation model. response and represented another useful data point for devel- Single-Slope Barrier, New Jersey Safety Shape oping a surrogate measure for OCD. Barrier, and Modified Texas T203 Bridge Rail Cobblestone Reveal Barrier. An alternative treatment The standard single-slope barrier, New Jersey safety shape, developed by Caltrans to address the OCD problems asso- and Texas T203 bridge rail were also modeled and evaluated. ciated with the deep cobblestone barrier was to provide a Each of these tests had acceptable OCD (i.e., < 150 mm) and smooth reveal at the bottom of the barrier. The 610-mm-tall met NCHRP Report 350 guidelines. These "passing" crash reveal, which had a smooth, sandblasted finish (see Figure 53), tests provide confidence in establishing a "passing" threshold was intended to reduce the snagging contact between the for the selected surrogate OCD criterion. barrier and wheel assembly and, thereby, reduce the result- ing OCD. This test successfully passed NCHRP Report 350 Results criteria and provided another point for use in establishing The first and most obvious measure of OCD was to take a the thresholds for a surrogate OCD measure. This barrier direct measurement of the maximum deformation to the also possessed some similarity to the safety shape barriers floorboard and toe pan of the vehicle in a manner similar to that were to be addressed in this study, since the surface as- that used in crash test evaluation. An example of the OCD perities were to be applied to the upper-wall portion of the generated in the pickup truck model is shown in Figure 54. safety shape barrier, while the toe of the barrier was to be left The deformation shown in Figure 54(b) is caused by induced smooth. buckling resulting from compression of the floorboard. (a) (b) Figure 52. Shallow cobblestone barrier: (a) actual; (b) simulation model.

OCR for page 33
45 (a) (b) Figure 53. Cobblestone reveal barrier: (a) actual; (b) simulation model. Direct measurements of OCD were obtained from the simu- time of initial impact, and total impulse were all computed and lations and compared with measured full-scale crash test OCD analyzed to investigate their correlation to OCD measure- values. As shown in Table 1, there is some correlation observed ments. These measures of contact force between the wheel between the simulation and the test data. However, the relia- and the barrier have been tabulated in Table 2. The correlation bility of predicting the outcome of a test, based on a single num- between these measures and actual OCD measurements was ber from simulation, was not considered very high. In an actual found to be poor. The amount of variation that exists in these crash test, the wheel, wheel well, fender, and other parts may data between acceptable and failed crash tests was not ade- contact the floorboard and cause additional OCD. The accurate quate to confidently use these forces as a surrogate measure representation of this mode of deformation requires failure in for OCD. This is possibly due to the unreliable values of one or more components of the suspension that are not repre- force between parts undergoing such severe deformation. sented in current vehicle models. For this reason, direct mea- Finally, the internal energies of the vehicle parts in the surement was not used as a measure for predicting OCD. crushed region of the vehicle were obtained and checked for As mentioned above, much of the OCD in a barrier crash correlation to OCD measurement. The internal energy in a test results from the wheel and suspension assembly being part is related to the overall deformation experienced by the deformed and shoved back into the toe pan area. An option part. Internal energies obtained from the floorboard and wheel was set into the model to collect the direct impact forces well showed the best correlation to the actual crash test re- between the wheel and barrier. These forces were evaluated sults among the measures evaluated (see Table 3). Between using several criteria. The XY and XYZ resultants of the peak these, the truck floorboard was selected as the surrogate force, peak 10-ms moving average force, impulse over the measure of OCD because it had slightly better correlation, (a) (b) Figure 54. Buckling floorboard: (a) undeformed; (b) deformed.

OCR for page 33
46 TABLE 1 Direct measurements for truck OCD study Crash Test Direct OCD Measurement Name Pass/Fail [mm] [mm] Oregon Fail 475 170 Cobblestone Fail 160 225 Cobblestone with Reveal Pass 98 50 Single Slope Pass 140 50 New Jersey Pass Not Reported 80 Modified T203 Pass 130 80 Shallow Cobblestone Pass 133 105 TABLE 2 Wheel to barrier contact forces and impulses for truck OCD study XYZ Resultant Crash Test Max 10 ms Total OCD Max Force Moving Avg. Impulse Impulse Name Pass/Fail [mm] [N] [N] [N-s] [N-s] Oregon Fail 475 1,290,000 459,000 28,800 29,900 Cobblestone Fail 160 1,340,000 450,000 34,500 40,100 Cobblestone with Reveal Pass 98 278,000 195,000 14,500 14,800 Single Slope Pass 140 510,000 164,000 10,700 15,200 New Jersey Pass Not Reported 229,000 197,000 12,100 12,800 Modified T203 Pass 130 290,000 231,000 12,800 21,100 Shallow Cobblestone Pass 133 910,000 459,000 18,700 18,700 XY Resultant Oregon Fail 475 1,290,000 455,000 27,700 28,300 Cobblestone Fail 160 1,176,000 438,000 31,900 35,900 Cobblestone with Reveal Pass 98 276,000 195,000 14,300 14,600 Single Slope Pass 140 498,000 164,000 10,600 15,100 New Jersey Pass Not Reported 228,000 196,000 12,000 12,700 Modified T203 Pass 130 263,000 123,000 11,900 19,800 Shallow Cobblestone Pass 133 901,000 449,000 18,000 18,000 TABLE 3 Internal energies for truck OCD study Crash Test Floorboard Wheel Well OCD Part Part Name Pass/Fail [mm] [N-mm] [N-mm] Oregon Fail 475 9,826,000 14,140,000 Cobblestone Fail 160 10,783,000 11,040,000 Cobblestone with Reveal Pass 98 782,400 3,980,000 Single Slope Pass 140 721,300 2,469,000 New Jersey Pass Not Reported 1,130,000 2,870,000 Modified T203 Pass 130 1,172,000 3,300,000 Shallow Cobblestone Pass 133 2,150,000 7,540,000 because its deformation is less influenced by contact with surrogate measure for evaluating OCD. Using the internal other parts of the vehicle, and because use of floorboard energy from the simulations and the reported OCD values deformation is more intuitively appealing given the nature of from the crash tests, thresholds for the surrogate measure OCD that occurs in a crash test. were established. As shown in Figure 55, the passing limit was selected as 2,200 N-m and the failure limit was tenta- Conclusions for the 2000P Study tively set at 10,700 N-m of internal energy in the floorboard of the pickup truck. As mentioned above, the internal energy of the floorboard The failure point that occurs at an internal energy of of the pickup truck was selected as the most appropriate 7,100 N-m is associated with the Oregon bridge rail. It is