Crash Testing with a Massive Moving Barrier as an Accident Reconstruction Tool

Transcription

1 SAE TECHNICAL PAPER SERIES Crash Testing with a assive oving Barrier as an Aident Reonstrution Tool Ronald L. Woolley, Alan F. Asay and Dagmar Buzeman Jewkes Woolley Engineering Researh Corp. Chuk onson GH Engineering Reprinted From: Safety Test ethodologies (SP 1516) SAE 000 World Congress Detroit, ihigan arh 6 9, Commonwealth Drive, Warrendale, PA U.S.A. Tel: (74) Fax: (74)

2 The appearane of this ISSN ode at the bottom of this page indiates SAE s onsent that opies of the paper may be made for personal or internal use of speifi lients. This onsent is given on the ondition, however, that the opier pay a $7.00 per artile opy fee through the Copyright Clearane Center, In. Operations Center, Rosewood Drive, Danvers, A 0193 for opying beyond that permitted by Setions 107 or 108 of the U.S. Copyright Law. This onsent does not extend to other kinds of opying suh as opying for general distribution, for advertising or promotional purposes, for reating new olletive works, or for resale. SAE routinely stoks printed papers for a period of three years following date of publiation. Diret your orders to SAE Customer Sales and Satisfation Department. Quantity reprint rates an be obtained from the Customer Sales and Satisfation Department. To request permission to reprint a tehnial paper or permission to use opyrighted SAE publiations in other works, ontat the SAE Publiations Group. All SAE papers, standards, and seleted books are abstrated and indexed in the Global obility Database No part of this publiation may be reprodued in any form, in an eletroni retrieval system or otherwise, without the prior written permission of the publisher. ISSN Copyright 000 Soiety of Automotive Engineers, In. Positions and opinions advaned in this paper are those of the author(s) and not neessarily those of SAE. The author is solely responsible for the ontent of the paper. A proess is available by whih disussions will be printed with the paper if it is published in SAE Transations. For permission to publish this paper in full or in part, ontat the SAE Publiations Group. Persons wishing to submit papers to be onsidered for presentation or publiation through SAE should send the manusript or a 300 word abstrat of a proposed manusript to: Seretary, Engineering eetings Board, SAE. Printed in USA

3 Crash Testing with a assive oving Barrier as an Aident Reonstrution Tool Ronald L. Woolley, Alan F. Asay and Dagmar Buzeman Jewkes Woolley Engineering Researh Corp. Chuk onson GH Engineering Copyright 000 Soiety of Automotive Engineers, In. ABSTRACT Damage analysis methods in aident reonstrution use an estimate of vehile stiffness together with measured rush to alulate rush energy, losing speed, and vehile delta-v. Stiffness is generally derived from barrier rash test data. The aident being reonstruted often involves one or more onditions for whih vehile stiffness is not well defined by existing rash tests. assive moving barrier () testing is introdued as a tool to obtain additional and aident speifi stiffness oeffiients appliable for reonstrution. The impats a stationary vehile of similar struture as the aident vehile under aident-speifi onditions like impat loation, angle, over-ride / under-ride, offset and damage energy. A rigid or deformable struture is mounted to the front of the, representative of the impating struture in the aident. Four illustrative tests are presented. A 1984 Honda Civi frontal impat (x), a 1988 Dodge Caravan rear impat and a 199 Isuzu Rodeo frontal offset / over-ride impat were onduted using the. The tests demonstrated that the testing method is an effiient means to attain stiffness, rush energy and aeleration data for speifi aident onditions. INTRODUCTION assive oving Barrier () tests are intended to omplement existing data bases. Sine the late 1970 s, tests have been onduted in aordane with Federal otor Vehile Safety Standards (FVSS) 08, 14, and 301 for front, side and rear rash performane, respetively. Additionally, New Car Assessment Program (NCAP) tests have been performed, of whih some are onduted against a load ell barrier onsisting of 36 ells [NHTSA, 1999]. The rash database provides 30 and 35 mph frontal barrier data, side impats at 33.5 mph by a 3000 pound moving deformable barrier at a 7 degree rab angle, and rear impats at 30 and 35 mph by a 4000 pound rigid moving barrier. (3000 lb = N, 4000 lb = N, 30 mph = 48.3 kph, 33.5 mph = 53.9 kph, 35 mph = 56.3 kph) In aident reonstrution, the vehile damage may have ourred under one or more onditions for whih the vehile stiffness is not well defined by the existing rash test data. Suh situations our when rush is a result of over- or under-ride, when the diretion of fore differs markedly from above-mentioned test-onditions (oblique impat), when the involved struture is not represented in the database (side impat into axles) or in offset and narrow impats (poles). A versatile test method is needed, whih allows variation of impat onditions like amount of overlap, impat orientation and loation, as well as impat speed. OBJECTIVE The objetive of this paper was to introdue a new and versatile test method with a assive oving Barrier, as an effetive tool for aident reonstrution purposes. Four tests are presented to illustrate the test method and show that it provides a valid tool for estimation of vehile stiffness data, nonlinear foredefletion harateristis, aeleration pulses, vehile BEV s, and vehile rush energy. ETHOD GENERAL assive oving Barrier tests involve driving a large, speially reinfored vehile into a stationary target test vehile. The strutures of both vehiles experiene the same fore at eah instant in aordane with Newton s 3 rd Law. Typially the is about 10 times heavier than the test ar. Sine rash delta-v is proportional to the losing speed and the ratio of opposite vehile mass to total mass, a delta-v on the test ar of 9/10 th of the impat speed is ahieved while the experienes a speed hange of only about 1/10 th the impat speed. The low delta-v allows the to be 1

4 driven into the pre-positioned test vehile at various impat angles and/or loations. Repeated hits into the same struture an be performed easily. testing simplifies utilization of previously wreked ars for test vehiles sine they remain stationary until impat. testing is dynamially the opposite of fixed barrier testing in that a stationary test vehile is aelerated through a hange in veloity and then slides to rest over some distane whereas in fixed barrier testing the ar is deelerated to a stop from an initial impat speed. TEST PROCEDURE One of the more obvious differenes between testing and the more onventional test trak method is the absene of a test trak. Instead, a paved road is required of suffiient length and width to aelerate the up to impat speed and to aommodate post-impat run-out of both the test vehile and. Sine both vehiles are moving post-impat at a speed of roughly 9/10 th of the impat speed, this beomes an essential safety onsideration. The four tests presented in this paper were run on low traffi roads, losed briefly during testing. INSTRUENTATION Instrumentation onsisted of two tri-axial piezo-resistive aelerometers (range = 00 g s, auray = %) mounted to the body struture at the base of the B pillars on the test vehiles. Two apaitive type aelerometers (range = 10 g s, % auray) were mounted on the frames behind the ab for the Honda and Caravan tests. All aelerometers were onneted to Data Brik aquisition systems provided by GH Engineering. The data was aquired in aordane with the SAE J11/1 AR95 Reommended Pratie. All hannels were sampled at 1.8 khz with anti-aliasing filters of the hannel frequeny lass (CFC) Right and left aeleration pulses were averaged and filtered with a SAE CFC 60 filter when used for omparisons of aeleration or fore data, and with a SAE CFC 180 filter before integration to veloity and displaement time-histories. The speed of the just prior to impat was measured with two laser speed traps (auray of 0.5%) provided by GH. DOCUENTATION Honda and Caravan tests were video taped from various angles. All tests were photographed with 35 mm print film to doument damage and test onditions. CRUSH PROFILES Residual or post-impat rush profiles were obtained by measuring a set of pre-defined points on the vehile both before and after testing, using a total-station surveying instrument. The two 3- dimensional data sets were then aligned by mathing three widely spaed points on the least damaged part of the tested ar with orresponding points on the undamaged ar. Displaement of orresponding points in the damaged zone provides both D and 3D residual rush profiles and displaement vetors. ANALYSIS NOTATION In the equations whih follow, subsripts a and b represent aident vehiles, subsript represents the test ar whih orresponds to vehile a in the aident and subsript represents the massive moving barrier. V is the average speed of the as measured by the speed trap. TEST DELTA-V Test vehile delta-v is given by the equation of onservation of momentum with zero restitution. Delta V = + V VEHICLE DYNAICS The impat fore time-history on the test vehile was alulated as the average longitudinal aeleration pulse, multiplied by the mass of the. It is possible to use the aeleration of the for this alulation, sine the onservation of its mass is guaranteed by its rigid struture. The aeleration of the test ar annot be used for this purpose, sine the vehile mass to be deelerated dereases with ar deformation. An overestimate of the barrier-fore would be attained from initial-mass times test-ar aeleration [Fossat,1994]. Collision or fore was adjusted by a small offset to attain a zero fore at time zero. F ( t) = a( t) () Veloity time graphs were obtained by integration of the average vehile aeleration trae. The result was mathed with the pre-impat speed as measured by the speed trap. V ( t) = V (0) a( t) dt V ( t) = a( t) dt () 4 Longitudinal displaements in these tests were approximated by double integration of the average longitudinal aeleration pulse of both vehiles. The differene between the and test-ar displaements was then taken as an approximation of dynami rush, X(t) d. The result s offset was set to zero. Aelerometer alibration errors and vibration errors were then minimized by saling the entire dynami rush trae to math the alulated residual rush, X,r, to the measured rush, X m,r. (1) () 3

5 X ( t) X ( t) X ( t) d = = ( a( t) dt + 0 ) dt () 5 ( a( t) dt + V ) dt X () 6 = o o [ X ( t) X ( t) X m, r ] () 7 DAAGE ANALYSIS assive oving Barrier testing is onduted to determine missing or inadequately defined rash stiffness data. Test onditions are seleted with the objetive of approximating either damage energy or vehile delta-v by using rough estimates of stiffness oeffiients in a damage analysis model. Or, given an estimate of the ratio of stiffness of the ollision partners, an estimate of the damage energy for one ar may be obtained through Newton s 3 rd law, a rush model, and an estimate of the damage energy for the other ar. The vehile stiffness and subsequent test onditions (if needed) are then refined by the new test data. Damage analysis, as used in aident reonstrution, is generally based upon a fore defletion model suh as the linear spring model of CRASH3 [Campbell, 1974; NHTSA, 1981]. Extensions to the basi model have been made for non-linear effets suh as fore saturation [Strother, 1986; Strother, 1990; Fonda, 1990; Woolley, 1991; Varat, 1994; Wood, 1997]. The basi linear spring model is written as an integral over the rush profile with stiffness oeffiients: A, B, G [Campbell, 1974]. However, this is a quadrati, two parameter model with G=A /B. X, r saturation extension in whih f s represents the fore saturation level and x s is the orresponding rush value at saturation [Woolley, 1991]. E w = k( x + xo ) dw n k( x xs ) Where k = B n = 0 n = 1 x o = for x <= x for A B w x > x G = ( x + x ) = A B When the rush profile is essentially uniform, as in barrier rush, then the integral may be represented by an average rush integrated over a harateristi width, w. For the onstant stiffness model (only the first term of Equation 8), the square root of the equation provides a linear result, with slope, k, and interept value, x o [Woolley, 1983]. TEST / ACCIDENT ENERGY AND DELTA-V In the aident reonstrution, the rush energy for both ollision partners must be added to obtain the total rush energy in the aident (E a +E b ). This total damage energy equals the differene in the kineti energy before and after the impat, negleting tire fores. By ombining the priniples of onservation of energy and onservation of momentum the damage energy an be related to the losing veloity (Equation 9a). The vetor differene of the delta-v s of the two ollision partners is equal to the vetor differene between losing and separation speed. s s 1 1 f s kx = k o dw s o (8) ( E a + E ) = b 1 a b + a b ( V l V sep ) (9a) The same method is used to alulate the impat veloity neessary to math the damage of the test ar,, and aident ar, a (Equation 9b): E a A E Test = 1 V + (9b) Figure 0. FFore Saturation odel Notation. To emphasize the linear spring basis of the model and for analytial onveniene, this integral may be written in fatored form using linear spring notation where k is the spring stiffness per unit width, w, and x o is the strutural reovery distane between dynami and residual rush. The first term of Equation (8) represents the basi Campbell model and the seond term provides the fore Usually, the mass of the test ar,, is seleted to losely math the mass of the aident ar, a. However, it should be noted that the delta-v experiened by the test ar generally will not math that of the aident ar when mathing the damage energy. The delta-v math is related to the total rush energy and mass ratio of the two aident vehiles (ar a and ar b), as well as the mass of the. The damage energy in the aident and test an be rewritten in terms of the delta-v: 3

6 1 a b A b a ( E + E ) = ( + ) V a (10) a b A E Test = 1 Test ( + ) V (11) The test delta-v thus relates to the aident delta-v by: V V a A Test = E E a A Test E 1 b + E a A m 1 + m m 1 + a m b (1) When the damage energy is mathed, the test delta-v approximates the aident vehile s when the rush energy ratio of the two aident vehiles is inversely proportional to their mass ratio, and the mass is many times larger than that of the aident ar. In the speial ase omparing an test with a rigid fixed barrier test, m b = and E b =0. Then the delta-v s would math if damage energy in the test were larger than in the fixed barrier by the mass ratio of test ar to ( E Test /E aa = 1+m /m ). Equations 9 1 apply to a ollinear ollision. For oblique ollision testing in whih rotational terms are signifiant, use of a D ollision model omputer program will provide more aurate delta-v and damage energy omputations. It is not essential to math either the damage energy or delta-v given the objetive of vehile stiffness determination via testing for an unusual rash onfiguration. The task is simpler and fewer test runs are needed if the goal is to exeed the aident damage in the test in order to provide interpolation rather than extrapolation stiffness data. Considering the potential delta-v mismath, the added ompliation of additional instrumentation to provide oupant injury data during testing may not be worthwhile. VALIDATION TEST 1984 HONDA CIVIC FRONTAL IPACT The assive oving Barrier was driven into the front of a stationary 1984 Honda Civi at a speed of 41.4 mph (66.6 kph) (Table 1). The front fae of the onsisted of a flat, rigid barrier similar to the fixed barrier fae in FVSS 08 and NCAP tests. The Honda Civi was positioned suh that the impat was head-on with full engagement of the front strutures (Figure 1). The impat speed was seleted suh that the delta-v in the test exeeded the delta-v of the NCAP test. Figure 1. Frontal Crash Test of the with a 1984 Honda Civi. 4

7 Table 1. Frontal Crash Test; 1984 Honda Civi with 1984-Honda Civi 4dr US units SI US Units SI ass 1960 lb 889 kg 6,760 lb 1138 kg Speed 0 mph 0 km/h 41.4 mph 66.6 km/h Delta-V 38.6 mph 6.1 km/h -.8 mph -4.5 km/h Ave. Crush 6.6 inh 67.6 m 0 inh 0 m ax. Crush 7. inh 69.1 m 0 inh 0 m PDOF 0 degree 0 degree Crush Energy 104,637 ft-lbf 141,869 Nm 0 ft-lbf 0 Nm E/w 01.3 lbf 44.6 N 0 lbf 0 N ax. Fore 170,744 lbf 759,640 N 170,744 lbf 759,640 N Peak Aeleration 64.8 g 6.7 g BEV 40.0 mph 64.4 km/h 0 mph 0 km/h Figure. Aeleration, Veloity, and Displaement Data for the Civi / Test. (3 data traes). Figure presents the longitudinal aeleration, veloity and displaement of the and the test-ar as measured and alulated in aordane with equations 3-6. NHTSA onduted a load ell barrier test on a 1984 Honda Civi as part of the NCAP test program (NCAP- 694). In this test the Honda was towed into a fixed barrier whih was fitted with 36 load ells (4 rows of 9 ells eah). The output from eah ell was added, and the total fore of all 36 ells was given for the test. The foredefletion urve of the to Civi test was validated against that measured in NCAP-694 test. The dynami deformation of the Civi was obtained using equation (7) and is ompared to that found through double integration of the aeleration in the NCAP test (Figure 3). The rush in the test inreased more quikly, as expeted from the higher losing speed (41.4 mph (66.6 km/h) vs 35 mph (56.0 km/h) in the NCAP). Figure 3. Dynami Crush of the Honda Civi. Figure 4 gives the fore - time history for the NCAP test 694 and for the test as alulated from the mass times aeleration of the. Both methods show similar fore peaks and rise times. 5

8 Figure 4. Comparison of NCAP-694 Load Cell Barrier Fore time-history with Fore on the Honda Civi. Fore-defletion urves as measured by the NHTSA load ell barrier and agree well. Both show loal fore peaks at similar defletions of approximately 6 inhes (15 m), 14 inhes (36 m) and 3 inhes (58 m) (Figure 5). Differenes in the two urves reflet vibrations of the frame to whih the aelerometers were mounted, and differenes between otherwise idential ar-models. The maximum dynami rush and the rush energy were higher in the test, due to the greater delta-v in the test than in the NCAP. Figure 6. Crash Plot ( E/w vs Residual Crush) of Honda Civi Frontal Test Data. A straight line through the data is indiative of a onstant stiffness over the range of the data. A urve with dereasing slope with greater rush is indiative of dereasing stiffness oeffiient or fore saturation. The 1984 Civi data and the test data point reflet a fore saturation trend in Figure 6. The test data point at high residual rush is the result of a repeated rash by the into the Civi (see below). Damage to the Civi in the 1 st test was doumented as desribed previously, and the resulting damage profile is shown in Figure 7. Figure 7. Honda Civi Damage Profile after 1 st Impat TEST : SECOND REPETITION 1984 HONDA CIVIC FRONTAL IPACT Figure 5. Comparison of Fore-Defletion Curves for NCAP-694 Load Cell Barrier Test and Test on the Honda Civi. Several FVSS-08 and NCAP tests have been onduted on 1984 and 1983 Honda Civis. These are NHTSA tests 694 and 705 for the 1984 model Civi and NHTSA tests 189, 066, 000, 1801, and 175 for the 1983 model year Civi. The rush energy parameter, (E/w), and average residual rush were alulated for these tests and ompared to those of the test (Figure 6). The Honda Civi from test 1 was again positioned on the roadway with equal orientation as the first test, and the post-rash vehile attitude was not adjusted. A seond full frontal impat test was onduted at an speed of 41.4 mph (66.6 kph), Figure 8. Table gives the onditions for this repeated test. In this test, the high rash severity aused the ables to be ut on the Civi by deformation of the ompartment, and the aelerometers mounted on the to exeed the range. Hene, the fore-defletion plot for this rash was invalid. aelerometer extremes at these high fore levels appear to be a ombination of exessive vibration of the frame to whih the aelerometers were mounted and the relative motion of various masses suh as engine, bed, rear axle, et. The aeleration trae from the surviving instrument on the Civi is shown in Figure 9, with a peak aeleration of 88 g. 6

9 Table. Repeated Frontal Crash Test; 1984 Honda Civi with 1984 Honda Civi 4dr US Unit SI US Unit SI ass 1960 lb 889 kg 6,760 lb 1,138 kg Speed 0 mph 0 km/h 41.4 mph 66.6 km/h Delta-V 38.6 mph 6.1 km/h -.8 mph -4.5 km/h Equivalent Delta-V 54.6 mph 87.9 km/h Ave. Crush, 48.5 inh 13. m 0 inh 0 m ax. Crush 49.0 inh 14.5 m 0 inh 0 m PDOF 0 degree 0 degree Crush Energy 08,8 ft-lbf 8,319 Nm 0 ft-lbf 0 Nm E/w 83.9 lbf N 0 lbf 0 N ax. Fore Peak Aeleration 88.6 g BEV 56.4 mph 90.8 km/h 0 mph 0 km/h (Note: aelerometers exeeded range.) Figure 9. Aeleration of the Honda Civi During the Repeated Impat. Figure 8. Repeated Frontal Crash Test of the with a 1984 Honda Civi. It was learned that this 15 year old ar had previously been repaired in the rear half of the ar. This repair, plus the exessive rash of 56 mph BEV (90 kph) aused the unloked doors to open during the repeated rash, whih redued the vehile stiffness. Hene, the rush vetors shown graphially in Figure 10 exeed those whih would have resulted without door openings. The redued stiffness effet is refleted in Figure 6, where the rush data follows the fore saturation trend as previously desribed, although the rush required slightly lower deformation energy relative to the other data points. The validity of the fore / defletion urve for the nd Civi test is questionable for the reasons stated above. The fore-defletion urve is not presented in this paper. 7

10 Figure 10. Honda Civi Damage Profile after nd Impat. TEST3: REAR IPACT OF 1988 DODGE GRAND CARAVAN In the third example test, a 1988 Dodge Grand Caravan was rear-impated by the at a speed of 41.4 mph (66.6 kph) (same speed as for the Civi tests beause the engine is governed). The result was a mph delta-v. This test is vastly more severe than the FVSS-301 or NCAP rear impat tests beause of the large weight of the and the higher impat speed (Table 3). In the NHTSA tests, the delta-v is roughly half the test losing speed depending upon test vehile weight, whih for this test ar would result in 16. mph delta-v in the FVSS-301 and 18.9 mph delta-v in the NCAP (6.0 or 30.4 kph). The fous of this test was examination of the post-impat dynamis of the Caravan with respet to a stationary ar loated 10 feet ahead, whih was simulated by foam ore side panels (Figure 11). In the aident under reonstrution, the Caravan was struk by a semi tratortrailer and pushed into the rear of the next ar. This produed an unusual and unexplained rear rush dynami on that ar. Table 3. Rear Impat Test; 1988 Dodge Grand Caravan with 1988 Dodge Grand Caravan US Unit SI US Unit SI ass 340 lb 1551 kg 6,968 lbf 1,3 kg Speed 0 mph 0 km/h 41.4 mph 66.6 km/h Delta-V mph 59.1 km/h mph -7.5 km/h Ave.Crush 31.7 inh 80.5 m 0 inh 0 m ax. Crush 3.3 inh 8.0 m 0 inh 0 m PDOF 180 degree 0 degree Crush Energy 173,900 ft-lbf 35,777 Nm 0 ft-lbf 0 Nm E/W 44.6 lbf N 0 lbf 0 N ax. Fore () 15,633 lbf 959,183 N 15,633 lbf 959,183 N Peak Aeleration 47.1 g 7.9 g FB BEV 39.0 mph 6.8 km/h 0 mph 0 km/h Equivalent- BEV (4000 lbf B) 53.1 mph 85.5 km/h 0 mph 0 km/h () Observed fore extremes are unertain. Figure 11. Rear Impat of a 1988 Dodge Grand Caravan by the. 8

11 Instrumentation in and data-redution proessing of the Caravan test was idential to that of the Civi tests. The fore-defletion data was again obtained from equations () and (7) and is shown in Figure 1. The observed result is a rapid rise to a peak fore followed by plasti yielding of the rear strutures at low fore followed by a rise to an approximately onstant fore of around 100,000 pounds. frame vibration and relative motion of various masses amplified extremes in the fore measurement at these high loads in the urrent struture. Hene, peak and minimum fore values are unertain. Additional testing to evaluate the size of the error is warranted. required impat speed would have been 53.1 mph (85.5 kph) to produe the same damage to the Caravan, were the replaed by a 4,000 pound barrier. The rash plot for the 1988 Caravan onsists of three data points, NHTSA test 16 on a 1988 Plymouth Voyager (same struture as Dodge Caravan), the test run at 41.4 mph (66.6 kph) and a 7 mph (11.3 kph) fixed barrier assumed interept value [Woolley 1991]. These three data points indiate that the rear struture ould be modeled by a onstant stiffness and fore saturation model (Figure 13). Figure 14 shows the extent of the residual damage to the Caravan in this rash. Figure 1. Observed Fore-Defletion for Rear Impat of a 1988 Dodge Grand Caravan (Extreme values exaggerated by aelerometer motion). Figure 13. Crash Plot for Rear Impat Tests of 1988 Plymouth Grand Voyager and Dodge Grand Caravan. The rear axle was pushed downward as it was driven forward, ausing the rear of the van to lift upward. Rear average rush was measured post-impat at 3 inhes (81 m), more than 3 times the 9.6 inhes of rush (4 m) in the 30 mph FVSS-301 NHTSA-16 test. The 9 Figure 14. Damage Profile at the Rear of a 1988 Dodge Grand Caravan. TEST 4: OFFSET /OVERRIDE FRONTAL IPACT OF A 199 ISUZU RODEO In test example 4, the front fae of the was mounted to aommodate an offset override rash into the front driver s side of a 199 Isuzu Rodeo. The amount of the overlap (0 %) and the override (4.5 inhes (6 m) ground to lower edge) was set to model a speifi aident being reonstruted. The Rodeo was set at an angle of 4 degrees to the path of the onoming. With this alignment the orner of the diretly ontated the top of the left front tire as it passed through the fender but missed the engine entirely. Figure 15 shows the pre-test setup and post-test vehiles at rest. No video was taken of this test. The purpose of the test was to obtain the aeleration of the Rodeo under these aident onditions for studies of restraint system performane via sled testing. Therefore the Rodeo was instrumented with triaxial aelerometers mounted to the base of the B pillars. Following the test, the Rodeo ompartment was onverted into a sled buk. Test results are given in Table 4. The longitudinal aeleration in this test is shown in Figure 16. Three peaks are observed. These are believed to orrespond to the interation with the top of the bumper of the Rodeo, with the tire, and with the A pillar, respetively. The tire impat gave rise to the largest aeleration without deflating the tire. Beause of the offset, the Rodeo rotated and therefore produed different left / right aelerations.

12 Table 4. Frontal Offset Override Impat Test; 199 Isuzu Rodeo with 199 Isuzu Rodeo US unit SI US unit SI 370 lb 1687 kg 5,40 lb 11,449 kg Speed 0 mph 0 km/h.9 mph 36.9 km/h Overlap ~ 0% ass Top of Bumper 8.5 inh 7.4 m Override Height 4.5 inh 6. m Bottom of frame 16 inh 40.6 m Crush Width 1.6 inh 3 m Delta-V 0.0 mph 3. km/h -.94 mph -4.7 km/h Ave. Crush, rms 33.1 inh 84 m 0 inh 0 inh ax. Crush 34. inh 86.9 m 0 inh 0 m PDOF -10 degree Crush Energy 56,837 ft-lbf 77,061 Nm E/W Peak Aeleration BEV 39. lbf 0 degree N 8.9 g 1.4 mph 0 ft-lbf 0 Nm 0 lbf 0 N 34.4 km/h 0 mph 0 km/h Figure 16. Isuzu Rodeo Average Longitudinal Aeleration. The rush energy parameter, (E/w), v. residual rush is presented in Figure 17 for NHTSA frontal impat tests 1586, 1891, 313 and 406 on Isuzu Rodeos. The assumed interept of 5 mph (8 kph) is also shown [Strother 1990]. These data imply a straight line, or onstant stiffness for the frontal tests. The graph inludes the result of the offset / override rash. The average residual rush in this test required lower damage energy per unit width than the NHTSA tests. This is aused by a relatively low loal stiffness per width unit of the struture impated in this speifi aident ompared to the average full frontal stiffness. Buzeman-Jewkes et al. (1999) has previously indiated the importane of loal stiffness data for aident reonstrution appliations. Figure Isuzu Rodeo Test Setup and Post-Test Rest Positions. 10

13 Residual damage to the Rodeo front is shown in the photographs (Figure 15) and depited as displaement vetors in Figure 18. It is noted that displaement of the right side of the hood is the result of indued damage. Figure 17. Crash Plot for Isuzu Rodeo Offset Override Impat. Figure 18. Damage Profile at the Left Front of a 199 Isuzu Rodeo CONCLUSIONS assive moving barrier () rash testing for aident reonstrution purposes provides an effiient means for generating stiffness and rush energy data under unique onditions. This data supplements existing rash test data of the NHTSA database and other soures. In priniple, a wide variety of test onditions an be obtained by modifiation of the impating front fae attahed to the. The stationary test ar an be set at the appropriate angle to attain the desired prinipal diretion of fore and strike the appropriate vehile struture. The great mass of the minimizes its delta-v and rotation during impat while ausing great speed hanges in the test-ar. The tests showed that vehile delta-v in exess of NCAP tests an be suessfully performed either diretly or by applying the repeated rash-test method. Repeated tests are straightforward in their setup, and an provide energy vs. rush data at damage levels well in exess of FVSS test damage. testing is dynamially the opposite of fixed barrier testing in that a stationary test vehile is aelerated through a hange in veloity and then slides to rest over some distane whereas in fixed barrier testing the ar is deelerated to a stop from an initial impat speed. A entral advantage of the test method is the absene of a permanent testing faility, while providing the versatility of ontat area, impat loation and orientation, olliding strutures and impat speed. Low veloity hange of the allows the to be driven into ontat with the test vehile and braked to a stop after impat. The entral disadvantage is loation of a suitable test roadway with suffiient width to safely aommodate the post-impat run-out of both test vehile and. Simulations of the proposed test should be performed beforehand to evaluate the potential for unantiipated events. tests generally an not math both aident damage energy and delta-v in the same test. Therefore, instrumentation for oupant injury measures is disretionary, subjet to evaluation for eah test ondition and objetive. It is not essential to math either the damage energy or delta-v given the objetive of vehile stiffness determination via testing for an unusual rash onfiguration. The task is simpler and fewer test runs are needed if the goal is to exeed the aident damage in the test in order to provide interpolation rather than extrapolation stiffness data. Crush energy and frontal rush in the -Civi tests were ompared to orresponding NHTSA test data as validation of testing. The fore-defletion harateristi obtained from aelerometers on the and Civi in test 1 was ompared to orresponding NCAP load ell barrier data. The results indiated that foredefletion and nonlinear vehile stiffness observations an be obtained through testing for aident reonstrution purposes. A test-vehile delta-v an be obtained omparable to NCAP frontal barrier tests, and an even higher delta-v an be ahieved by a repeated impat. The seond rash repetition on the Civi produed exessive vibration of the aelerometers, indiating that the auray of peak fore measurement is diminished at aeleration levels exeeding approximately 0.7 g s with urrent struture and instrumentation. Improvement requires more rigid aelerometer mounts on the to redue vibration, and the addition of other aelerometers on major masses of the that are apable of elasti relative motion at high impat loads. 11

14 The four tests illustrate that testing allows variation of ontat area, impat loation, overlap and orientation, shape of olliding strutures and impat speed. As suh, the method an be adjusted to aident speifi onditions to obtain aident reonstrution rash test data on eah ollision partner, one vehile at a time. ACKNOWLEDGENTS The authors would like to express appreiation to ark Berry of Bowman & Brooke, Torrane, California and to Roger Glasow of Wright, Lindsey & Jennings, Little Rok, Arkansas. REFERENCES 1. [Buzeman-Jewkes, 1999] Buzeman-Jewkes, Dagmar G., Per L`vsund, and David C. Viano; Use of Repeated Cras-Tests to Determine Loal Longitudinal and Shear Stiffness of the Vehile Front with Crush, Soiety of Automotive Engineers, [Campbell, 1974] Campbell, Kenneth; Energy Basis for Collision Severity, Soiety of Automotive Engineers, [Fonda, 1990] Fonda, Albert G., Crush Energy Formulations and Single-Event Reonstrution, Soiety of Automotive Engineers, [Fossat, 1994] Fossat, E., athematial odels to Evaluate Strutural Fores in Frontal Crash Tests. Pro. 14 th International Tehnial Conferene on ESV, 94-S8-O-04, unih, Germany. 5. [NHTSA, 1999] NHTSA Crash Test Database, Selet: R&D, Crash Test Database, Dynami Query Tool. 6. [Strother, 1986] Strother, Charles E., Ronald L. Woolley, ihael B. James, and Charles Y. Warner; Crush Energy in Aident Reonstrution, Soiety of Automotive Engineers, [Strother, 1990] Strother, Charles E., Ronald L. Woolley, and ihael B. James; A Comparison Between NHTSA Crash Test Data and CRASH3 Frontal Stiffness Coeffiients, Soiety of Automotive Engineers, [Woolley,1991] Woolley, Ronald L., Charles E. Strother, and ihael B. James; Rear Stiffness Coeffiients Derived from Barrier Test Data, Soiety of Automotive Engineers, [Woolley, 1983] Woolley, Ronald L., Charles E. Strother, Gregory C. Smith, James J. oulton, and Douglas D. Allsop; Graphial Solution of Reonstrution Equations, 7 th Annual Proeedings, Assoiation for the Advanement of Automotive ediine (Formerly: Amerian Assoiation for Automotive ediine), Otober, [Wood, 1997] Wood, Dennis P., and Stephen ooney; odelling of Car Dynami Frontal Crush, Soiety of Automotive Engineers, [Varat, 1994] Varat, ihael S., Stein E. Husher, and John F. Kerkhoff; An Analysis of Trends of Vehile Frontal Impat Stiffness, Soiety of Automotive Engineers, [Zeidler, 1985] Zeidler, F., H.-H. Shreier, and R. Stadelmann; Aident Researh and Aident Reonstrution by the EES-Aident Reonstrution ethod, Soiety of Automotive Engineers, CONTACT Ronald L. Woolley, P.E., Ph.D. Woolley Engineering Researh Corp North 50 West, Provo, Utah, woolley@wer.om 1