RICSAC-97 A Reevaluation of the Reference Set of Full Scale Crash Tests

Transcription

1 RICSAC-97 A Reevaluain f he Reference Se f Full Scale Crash Tess Brian G. McHenry and Raymnd R. McHenry McHenry Cnsulans, Inc. Cary, NC ABSTRACT Research perfrmed in he 1970's revealed significan limiains in he available dcumenain f vehicle crush infrmain and rajecry spinu infrmain. As a resul a series f full-scale crash ess were perfrmed which became knwn as he Research Inpu fr Cmpuer Simulain f Aumbile Cllisins (RICSAC) crash ess. Previus research using he RICSAC es resuls, paricularly in relain he validain f acciden recnsrucin cmpuer prgrams, has varied widely in accepance, inerpreain and presenain f he RICSAC es resuls. This paper presens a deailed review and deciphermen in useable frm f he riginal 12 crash ess ha were perfrmed wihin he RICSAC prgram. A new mehd f analyzing accelermeer daa frm arbirary sensr psiins, n he basis f discree measures f he vehicle respnses raher han cmplee ime-hisries, is defined. A discussin f previus research which included reference he RICSAC es resuls as a measure f he validiy f recnsrucin cmpuer prgrams is included. INTRODUCTION Research perfrmed in he 1970's lcae, review, decipher and place in useable frm available experimenal daa n he srucural crush prperies f aumbiles, and n he spinu rajecries prduced by measured cllisin cndiins [1] 1 revealed significan limiains in he available dcumenain f vehicle crush infrmain and rajecry spinu infrmain. As a resul, an iniial es marix f 12 full-scale crash ess was perfrmed in 1978 which became knwn as he Research Inpu fr Cmpuer Simulain f Aumbile Cllisins (RICSAC) crash ess [2]. Fr each f he RICSAC ess, a minimum f 13 accelermeers were muned n-bard each vehicle recrd accelerain cmpnens a six seven sains. A hree lcains riaxial (XYZ) packages were insalled ("hard 1 numbers in brackes [] indicae references a end f paper muned") prvide cverage beween he frn and rear f he vehicle. The frn seer angles were measured n each vehicle by a linear srke penimeer aached he vehicle seering linkage. The ime hisry f he change in vehicle yaw, pich and rll angles and yaw rae were recrded by wdegree-f-freedm, free gyrscpes and a rae gyr. Fr vehicle rajecry measuremen, high-speed min picure cameras were lcaed verhead wih suppring cameras a eye level heigh. Als, vehicle rajecry marker sysems were designed and fabricaed by Calspan spray high pressure clred waer rails (using very small amuns f liquid avid any effecs f he waer n he radway cefficien) frm several lcains n each f he vehicles. Pls f he full-scale es accelerain ime-hisries were generaed by a cmpuer prgram which als inegraed he accelerain ime hisries (using a cmbinain f Simpsn's and Newn's 3/8 mehds) in rder bain velciy ime hisry alng each f he vehicle's hree axes. Fr ms f he repred resuls, he V values repred fr he vehicle CG were based n daa clleced by a Firewall muned riaxial (XYZ) accelermeer. The ime f apprximae separain repred in [3] was based n bservain f he ime hisries f he accelerain. The ime f separain was defined as "he pin a which bh f he invlved vehicles' accelerains reapprached 0 g's". The auhrs f [3] ned ha "he separain pin was n always bvius in cllisins in which he w vehicles spun u geher". The cmpnens f he vehicle's change in velciy ( V) were cmpued in [3] "by subracing he iniial velciy a impac frm he velciy a he ime f separain". The prcedure was dne fr he X and Y velciy cmpnens separaely. The auhrs f [3] acknwledged ha he "value fr he separain velciy was cnaminaed by he effecs f rain f he vehicles beween impac and separain". The prblem was aribued he ps-prcessing inegrain sfware used fr he ess which was primarily se up inegrae accelerain daa fr frnal cllisins wih barriers and herefre did n accun fr he rain f he vehicle (he daa reducin sfware assumed a cnsan direcin

2 csine marix). Errrs which may ccur in daa clleced frm an accelermeer n a he cener f graviy include he effecs f rain which are dependen n he magniude f he ffse f he accelermeer frm he CG and he magniudes f he vehicle's angular velciy and accelerain. Table 1 cnains bes esimaes f he acual accelermeer lcains fr he 12 RICSAC ess based n values repred by Jnes [3]. Sme f he lcains have been crreced n he basis f a deailed review f he RICSAC es reprs [2]. Table 2 cnains a lis f facrs as repred in [3] which influenced he magniude f effecs f rain n he RICSAC accelermeer daa. Ne ha any differences in he accelermeer lcains as lised in Tables 1 and 2 are due he refinemen in Table 1 f sme ambiguus dimensins repred in [2] and/r [3]. Table 1 Firewall and Rear Deck Accelermeer Lcains fr RICSAC Tess (Ne:Tes 4(S&N 6), Veh 2 Frn Deck) Table 2 Frm Jnes [3] Table 1-3 "Facrs Influencing Magniude f Effecs f Rain" n Accelermeer daa. (Ne: Cnac Durain frm Table 1-1 f [3]) BACKGROUND

3 In 1980, Smih and Nga [4] creaed a summary f daa assciaed wih 16 full-scale crash ess (hree mre ess were perfrmed as a par f RICSAC and repred in [5] and he her was frm [6]). The auhrs "aemped achieve crrecins accelerain and velciy daa by apprximaing vehicle rainal velciy and rain f he reference frame frm i's rienain a impac, by making reference muliple accelermeer upus a differen lcains n he vehicles and by analysis f high speed phgraphy." They als included an adjusmen prcedure fr he V's insure ha he mmenum changes assciaed wih he cllisin parners were equal and ppsie (including cnsiderain f ire frces). The adjusmens were apprximains accun fr he effecs f exernal ire-frces. They fund ha "cases number 10, 12 and 15 are believed be in errr" due pr accelermeer daa fr ne f he w invlved vehicles. The deerminain f he separain velciy was n refined r presened. In 1983, Brach [7] presened a able f refinemens he repred RICSAC separain velciies n he basis ha: "Accelermeers used recrd daa n each vehicle were n lcaed a he vehicle's cener f graviy. Cnsequenly, he final velciy daa had be crreced by he angular velciy a separain and he disance f he accelermeer frm he mass cener." The repr des n include he specific equains and he values fr disances f he accelermeers frm he CG and her variables used in he adjusmen prcess (The values frm Jnes repred in Table 2 were he values prbably used). In 1985, Wley, as a par f an aack n he CRASH3 analyical prcedures [8] and as par f a presenain and esing f he "IMPAC" cmpuer prgram [9], included a review and analysis f he RICSAC es resuls as repred by Smih and Nga [4]. In addiin agreeing wih Smih & Nga ha here were prblems wih ess 10, 12 and 15, ess 4 and 6 were idenified as having eiher "a yping errr r repring errr." A "Daa Inegriy Ranking" was calculaed fr he ess based n Newn's 3 rd Law fr equal and ppsie frces and he law f cnservain f linear mmenum. The ess uilized he ps-impac velciies (he separain velciies) which were he values as repred by Jnes [3]. The esing prcedures f [8] and [9] cied many insances where he "Daa Inegriy Ranking" calculains revealed cases wih errneus mmenum gains and/r insances where he frces acing n he cllisin parners where n equal and ppsie. Als included in he repred resuls f [4] was a "velciy change" angle. The "velciy change" angle was calculaed by finding he angle wih angen f he v (Y cmpnen) velciy change divided by he u (X cmpnen) velciy change. In [8], he "velciy change" angle repred in [4] was referred as he Principal Direcin f Frce (PDOF). IN 1989, as a par f a presenain f a cmparisn f EDCRASH prgram resuls wih he RICSAC es resuls Day [10] cncludes ha N general bservains culd be made using he RICSAC daa analyzing he accuracy f he {cmpued} V by any prgram because f measuremen errrs in he RICSAC es daa. The repr als saes ha "Previus prgram evaluains which used he RICSAC daa as a means f validain fr V shuld be viewed as suspec." Als in 1989, Cheng and Guenher [11] discussed he lcalized variains in speed-change ( V) due rainal effecs as applied he speed-change experienced by he vehicle ccupans. The paper included a discussin f he general differences in measuremens prduced by he varius accelermeer lcains in he RICSAC ess. IN 1990, Day [12] as a par f a presenain f a cmparisn f he EDSMAC prgram resuls wih he RICSAC es resuls included a Table f V resuls fr RICSAC which were bained frm [1-3] and "frm persnal crrespndence frm Dr. Russell A. Smih.". Sme f he repred values are differen frm Smih and Nga [4] and may reflec refinemens by Smih and Nga repred in [13]. The repr cnclusins include he bservain: "The measured Vs were n f accepable accuracy fr use in a validain sudy because he min ransducers were placed a he vehicle's firewall, raher han he CG. This prblem migh be imprved r eliminaed by re-analyzing he riginal daa wih sfware which included a ransfrmain marix." In 1996, as par f a validain effr f he PC-CRASH cmpuer prgram, Cliff [14] used he RICSAC es resuls fr he ps-impac separain speeds as repred by Brach [7] fr cmparisn "validain" purpses. They ned ha "here appeared be sme errr in a few f he repred ps-impac speeds in [he Brach] paper when cmpared he repred pre-impac speeds and vehicle weighs." Als in 1996, Bundrf [15] presened echniques fr analysis f he CG V frm accelermeer daa a lcains her han he CG. He bserved ha "he penial fr errr is always presen if yaw velciy is high and accelermeers are disan frm he cener f graviy." As an example, he presened a deailed analysis f RICSAC Tes n. 9 (S&N Case 11). RESEARCH APPROACH The RICSAC ess [1-3] cnain he ms cmprehensive cllecin f full-scale es resuls available dae. The es reprs include bjecive infrmain n he impac speeds, vehicle weighs, dimensins, weigh disribuins, spinu rajecry and psiins f res. Sme inerpreain f he repred resuls is required, fr example, bain speed-change ( V) and separain velciies frm he accelermeer daa. Als, sme evaluains are required fr he apprximae exen f wheel drag and seer angles. The primary purpse f he research assciaed wih his paper has been develp and apply echniques fr inerpreain f he crash phase f RICSAC and her full-scale ess. Of paricular ineres are he crrec values fr he impac speedchanges and separain velciies effecive a he ceners f graviy, which can be used as an bsacle curse evaluae and/r "validae" acciden recnsrucin echniques.

4 During full-scale esing, a hard-muned accelermeer a he CG will recrd he crrec accelerain including he effecs f rain. The ime inegral f he accelermeer daa fr he X and Y direcins will prduce he crrec vehicle-fixed X and Y cmpnens f speed-changes. Hwever, in he RICSAC ess and in many her full-scale ess, he V calculains are cmplicaed by accelermeer daa fr lcains her han he vehicle cener f graviy. The firs ask f inerpreain was develp generalized analyical echniques ransfrm he speed-change infrmain frm arbirary accelermeer lcains he cener f graviy. A secndary ask was use he calculaed CG speed-change infrmain calculae he vehicle separain velciies. T crrecly use he CG speed-change infrmain calculae he separain velciies, prvisins mus be included fr he changes in he vehicle rienain. The echnique used by Jnes [3] fr calculain f he separain speeds fr he RICSAC ess is in errr, primarily due he direc subracin f he speed-change deermined by inegrain f accelermeer daa frm he iniial impac speed. The u and v speed-changes cmpued by inegrain f he accelermeer X and Y daa cann be direcly subraced frm he iniial vehicle velciies excep fr he case where n rain ccurs. A secndary surce f errr in he calculain f separain velciies repred in [3] is he assumpin ha he accelerain and velciy daa frm he firewall lcain are equivalen hse a he vehicle cener f graviy. CG-TRANSFORM T prvide a simple apprximain f he RICSAC V and separain velciies a he vehicle ceners f graviy, an analyical prcedure was develped fr ransfrmain f accelermeer daa frm accelermeers n lcaed a he vehicle cener f graviy n he basis f discree measures f he respnses raher han cmplee ime-hisries. Bundrf [15] presened a prcedure fr deerminain f he speed-change ( V) a he vehicle cener f graviy (CG) using accelermeer ime hisries frm accelermeers n lcaed a he vehicle CG cmbined wih rae gyr daa. The echnique requires eiher he availabiliy f he FM daa recrder apes r a manual digiizain f he accelermeer and yaw velciy races deermine he inegral f yaw velciy squared and, hereby, he speed-change ( V). A drawback f he prcedure is ha i can be ime cnsuming fr he case where here are n FM daa apes and/r prcessing sfware available. Als, pssible surces f errrs can ccur frm inegraing and synchrnizing randmly sampled daa (he accelerains were recrded wih riaxial accelermeers, he angles and yaw raes were recrded wih a pair f wdegree-f-freedm, free gyrscpes and a rae gyr). The CG-Transfrm analyical prcedure has inpu requiremens lised in Table 3 (fr a deailed definiin f CG- Transfrm please see Appendix 1). Table 3 Infrmain required fr CG-Transfrm analyical prcedure 1. Accelermeer Lcain n vehicle, X, Y. 2. Impac speed and heading angle. 3. Inegraed speed-change frm accelermeer daa. 4. Durain f crash pulse. 5. Separain angular (yaw) velciy and angle. T permi iniial esing and refinemen f he CG-Transfrm, he SMAC cmpuer prgram was uilized creae a se f mahemaically crrec "es case" recnsrucins. Exensins he SMAC prgram develped fr he presen research include a prvisin mnir he accelerain fr any arbirary psiin n he vehicle (e.g., Figure 1, Figure 2). CG-Transfrm was hen used ransfrm he daa frm an arbirary accelermeer lcain back he CG. Since he CG daa is a direc upu f he SMAC prgram he prcedure prvided a deailed check f CG-Transfrm. The SMAC recnsrucins used fr his purpse were generalized recnsrucins f he RICSAC es runs. The SMAC runs were based n preliminary refinemens f he riginal RICSAC SMAC recnsrucins presened by Jnes [3] and mre recenly Day [16] in validain sudies f he SMAC and EDSMAC cmpuer prgrams. The refinemens f sme f he inpus were required due he quesinable naure f sme f he SMAC inpus used in he "validain" recnsrucins (e.g., sme rque and seer values were arbirarily varied). A mre deailed discussin f he rainale fr he changes he SMAC inpus and suggesins fr refinemen f he inpus are be presened as par f a separae research prjec [17].

5 Figure 1 SMAC resulan accelerain ime hisry a CG, Rear Deck and Firewall fr RICSAC es N. 8, Vehicle N. 2 Figure 2 SMAC resulan accelerain ime-hisry a CG, Firewall and RearDeck fr RICSAC es N. 6, Vehicle N. 2

6 Fr each f he refined RICSAC SMAC runs, he accelerain was mnired fr he Firewall (Cwl) and RearDeck lcains as specified in he RICSAC reprs. The insallain lcains fr he Firewall and RearDeck accelermeers are included in Table 1. The accelerain a any given lcain, X 1,Y 1, can be expressed fr a given ime,, as: a = u& v* ψ & (1) xx a = v& + u* ψ & (2) yy v r = resulan speed change vx = speed change in he vehicle x direcin v y = speed change in he vehicle y direcin v y V angle = aan2 1 (8) v x Where: 2 a = u& v ψ& x ψ& y && ψ 2 xx1 a1 a 1 (3) a ψ 1 & 2 = v& + u ψ& + x && ψ y (4) yy1 a1 a u = u y ψ & (5) x1 a1 v = v+ x ψ & (6) x1 a1 Where, fr a given ime inerval d: uv, uv &, & a xx, a yy ψ ψ& && ψ = frward and laeral velciy cmpnenes a cg = change in frward and laeral velciy cmpnens a cg = accelerain cmpnens a cg = vehicle heading angle = yaw velciy = yaw accelerain x a 1, y a 1 = lcain f accelermeer relaive cg u x 1, v x 1 = frward and laeral velciy cmpnens a x a 1, y a 1 a xx 1, a yy 1 = x, y accelerains a lcain x a 1, y a 1 The SMAC generaed accelerain ime hisries a he CG, Firewall and RearDeck were hen inegraed deermine he insananeus velciies and resuling speedchanges. Tw echniques were used deermine he impac speed-changes f he individual ess: Tes Equivalen and SMAC Equivalen. TEST EQUIVALENT: Fr he repred RICSAC full-scale ess [1-3] he analysis f he accelermeer daa deermine he vehicle speed-change cnsised primarily f he inegrain f he vehicle a x & a y accelermeer races deermine he ime hisries f velciies. The velciy imehisry pls were hen analyzed deermine he magniude f he changes in he individual velciy cmpnens, V x and V y. The resulan speed-change and speed-change angle were deermined by he fllwing: Where: vr = 2 vx + v 2 y (7) Vangle = speed change angle v x = speed change in he vehicle x direcin v y = speed change in he vehicle y direcin SMAC EQUIVALENT:In he riginal frm f he SMAC cmpuer prgram, he speed-change fr a given accelerain expsure was deermined by he inegrain f he resulan accelerain. Fr each ime incremen, a resulan accelerain was deermined by he vecr sum f he x and y cmpnens f he vehicle accelerains: Where: ac = 2 ax + a 2 y (9) a c = resulan accelerain, g-unis a x = accelerain in he vehicle x direcin, g-unis a y = accelerain in he vehicle y direcin, g-unis The resulan speed-change was deermined by inegrain f he resulan accelerain frm iniial cnac separain. The iniial cnac is defined as he firs insance where he resulan accelerain ges abve 1 g-uni. The pin f separain is defined as he firs ime ha he accelerain again drps back belw 1 g-uni. I was assumed ha in general handling maneuvers and vehicle spinus ha he resulan accelerain wuld nrmally be belw 1 g-uni (given ha he nminal fricin cefficien f radways is nrmally less han 1 g-uni). The Direcin f Principle Frce (DOPF) was deermined fr a given accelerain expsure in he SMAC prgram as he direcin f he accelerain a he insance f peak resulan accelerain by: a y DOPF = aan2 1 (10) a x

7 Where: DOPF = direcin f principal frce a x = accelerain in he vehicle x direcin a y = accelerain in he vehicle y direcin A summary f he seps required creae SMAC "full-scale es resuls" be used fr esing purpses f he CG-Transfrm analyical prcedure was as fllws: 1. A SMAC simulain was perfrmed fr each RICSAC es 2. Accelerain ime hisries were creaed fr he CG and he repred Firewall and Rear Deck accelermeer lcains fr each vehicle f each es. Table 1 is a lis f he repred Firewall and Rear Deck accelermeer lcains used in he SMAC ess. 3. The SMAC vehicle fixed X & Y accelerain ime-hisries were inegraed fr he CG, he firewall and he Rear Deck as fllws: 3.1. The individual speed-changes in he vehicle X and Y direcins were deermined by inegrain f he X and Y accelerain ime-hisries A Tes equivalen resulan speed-change was calculaed The SMAC equivalen resulan speedchange was als calculaed by inegrain f he ime-hisry f he resulan accelerain. Fr sme f he RICSAC ess he values fr he Tes equivalen speed-change was fund be less han he SMAC equivalen speed-change ( V) by as much as 10%-15%. The difference is due he fllwing: When we use inegrain find he velciy change prduced by an accelerain curve, we are finding he equivalen velciy ha is prduced by he variain f he accelerain ver a fixed inerval f ime. In full scale ess, he accelermeers measure he accelerain fr he X and Y direcins separaely. Therefre, he inegrain f hese accelerain ime-hisries prduce separae velciies (areas) fr he X and Y direcins. The resulan speed-change velciy frm full-scale ess is deermined by finding he square r f he sum f he squares fr he X and Y velciies. In he SMAC prgram, in addiin he separae X and Y accelerain ime-hisries, we als have available he resulan accelerain. Therefre, deermine he resulan speed-change we can inegrae direcly he resulan accelerain vs. ime curve. In a mr vehicle full-scale es, he vehicle and ccupans experience he resulan accelerain, n individually he X and Y cmpnens. The pssibiliy f differences in he resulan speed-change by hese w mehds reveals a need fr acciden recnsrucin simulain prgrams individually calculae he speedchanges in he X and Y direcins fr cmparisns wih fullscale ess, while als calculaing and repring he resulan speed-change deermined by inegrain f he resulan accelerain ver ime. Fr esing and cmparisn purpses f he CG-Transfrm analyical echnique, recnsrucins were perfrmed f he cllisin phase f he 12 RICSAC ess wih he SMAC cmpuer prgram. Table 4 cnains a summary f he RICSAC SMAC impac speed-change ( V) fr he CG, Firewall and Rear Deck, Table 5 cnains he impac cndiins, and Table 6 cnains he separain cndiins. The resuls f he SMAC recnsrucins f he RICSAC ess were used in an evaluain f he CG-Transfrm analyical echnique. The inegraed speedchange fr he Firewall and Rear Deck accelermeer lcains were used wih he her iems per Table 3 in an applicain f he CG-Transfrm analyical prcedure deermine he speed-change a he cener f graviy and he separain velciy. The resuls fr he speed-change ( V) and separain velciy fr each f he 12 RICSAC ess were calculaed. Fr all cases he speed-change and separain velciy calculaed fr he CG based n he Firewall and Rear Deck accelermeer daa was less hen 1 mph maximum errr, wih an average errr f ±0.1 MPH and a sandard deviain f 0.30 MPH. The cmparisn gave a gd indicain ha he CG-Transfrm analyical prcedure prperly apprximaes he CG speed-change and separain speed frm speed-change daa a arbirary accelermeer lcains cmbined wih discree measures f he yaw respnses. A dramaic demnsrain f he psiinal differences in accelerains and herefre he crrespnding differences in he inegraed speed-change calculains is illusraed in Figure 3 and Table 7. Figure 3 is he resulan accelerain ime-hisry fr he CG, Firewall and RearDeck fr he SMAC recnsrucin fr vehicle number 1 f RICSAC Tes N. 10. The full scale es and he SMAC simulain f RICSAC 10 each cnained a sideslap impac f he vehicles subsequen he primary impac. In Figure 3 here are w peaks in he accelerain ime hisries, paricularly a he RearDeck lcain. Table 7 cnains he inegraed speed-changes calculaed fr he primary impac, he side-slap and fr he al durain. The speed-change calculaed by cmpnens (Tes equivalen) and based n he resulan accelerain (SMAC equivalen) fr vehicle N. 1 illusraes a prblem which can ccur when inegraing individual accelerain cmpnens. Figure 4 is he X-Y accelerain cmpnens fr vehicle number 1 f RICSAC Tes N. 10. The X-accelerain fr he RearDeck lcain in Figure 4 reverses sign afer he primary impac which acs reduce and reverse he calculaed X cmpnen f speedchange fr he Tes equivalen inegrain echnique (The repred ime-hisries f he velciy cmpnens prduced by inegrain f he RICSAC accelerains cnain his anmaly, e.g., p 16-41,[2]). Deermining he speed-change

8 based n he resulan accelerain per he SMAC equivalen echnique eliminaes he direcinal sensiiviy. The side-slap accelerain f RICSAC Tes 10 als reveals a prblem f deermining which speed-change and/r separain velciy shuld be repred when a sideslap ccurs. Ne ha he SMAC Equivalen resulan speed-change fr he Rear Deck calculaed frm he iniial impac hrugh he end f he sideslap (frm sec.) is dramaically higher (63 mph) han he sum f he individual speed-changes fr he primary impac (25.5 mph) and sideslap impac (17.5 mph). The difference is due he inegrain f he accelerain (> 5 g s) which ccurs a he RearDeck (due i s ffse frm he cener f graviy) during he rapid rain f he vehicle beween he primary and sideslap impac ( apprximaely 0.20 secnds). I is herein prpsed ha all es reprs include prvisins indicae a side-slap impac. The psprcessing analysis f he accelerain daa shuld als inegrae and repr he speed-change fr he primary and secndary impacs separaely. Large accelerains ha can ccur a lcains her han he cener f graviy, as demnsraed a he RearDeck fr RICSAC Tes 10 vehicle 1, shuld n be included in he speed-change calculains fr he individual vehicles since he vehicles are n in cnac during his ime perid. Hwever, he addiinal accelerains due ffse frm he cener f graviy and rainal velciy may have dramaic effecs n he magniude f ccupan expsure and cnsiderain shuld be given lcalized V calculains as suggesed in [11]. Table 4 SMAC RICSAC recnsrucins prediced V's fr CG, Firewall and Rear Deck accelermeer lcains

9 Table 5 SMAC RICSAC recnsrucins Impac Cndiins Table 6 SMAC RICSAC recnsrucins Separain Cndiins

10 Figure 3 Resulan Accelerain ime hisries fr RICSAC Tes N. 10, Veh#1 fr CG, Firewall and RearDeck Table 7 Cmparisn f Calculaed Speed-changes fr RICSAC es N.10 fr CG, Firewall and RearDeck.

11 Figure 4 Accelerain cmpnens a RearDeck Lcain fr RICSAC Tes 10, Vehicle N. 1 Anher check f he CG-Transfrm was perfrm a cmparisn f he calculaed resuls wih resuls repred by Bundrf in [15]. Bundrf included a prcedure fr ransfrmain f accelermeer infrmain frm lcains her han he vehicle cener f graviy he CG. The prcedure presened by Bundrf includes digiizing and analyzing he accelerain and angular velciy ime-hisry daa. The repred resuls by Bundrf were cmpared wih resuls calculaed wih he CG-Transfrm prcedure. The resuls f he cmparisn are cnained in Table 8. The resuls f he CG-Transfrm prcedure crrelae well wih he resuls f he Bundrf analysis prcedure. The average resul and sandard deviain are nearly idenical. Sme small differences did ccur beween he differen accelermeer lcains. The resuls f applicain f CG-Transfrm analyical prcedure he varius lcains esed demnsrae he abiliy f he CG-Transfrm prcedure accuraely calculae he speed-change a he cener f graviy fr any arbirary accelermeer lcain. Table 8 Cmparisn f Bundrf [15] wih CG-Transfrm resuls fr RICSAC Tes 9

12 EVALUATION OF THE "RAW" RICSAC DATA The ask f evaluaing and ransfrming he accelermeer daa fr he RICSAC full scale ess was underaken deermine he CG speed-change and separain velciy fr each es. A few pssible prblem areas which migh affec he crrelain are: Surce f Raw Daa. Accelerain Time-hisry Oscillains. Excessive Cllisin Durain. SOURCE OF RAW DATA: The raw V values as repred in [3] were used fr he cmparisn. A brief visual check f he ime-hisry pls vs. he repred values was perfrmed check fr any majr differences. The reasn he her repred refinemens f he values culd n be used was ha he her repred values cnained adjusmens and crrecins ransfrm he resuls frm he Firewall accelermeer lcains he CG, he very ask ur CG-Transfrm is perfrm. ACCELERATION TIME-HISTORY OSCILLATIONS: Sme f he accelerain ime races cnained in he RICSAC reprs [2] appear have excessive scillains (e.g., RICSAC Tess.Veh: 2.1,2.2,5.1,5.2,8.2,11) and sme have resulan accelerains exceeding 40 g's (e.g., RICSAC Tes.Veh: 10.1,12.1,7.2). These resuls may eiher reflec a pssible prblem wih he accelermeers r wih he chice in cuff frequency filer. Sample scillgraph daa frm Mayr and Naab [18] cnained in [19] illusraes he differences beween unfilered and filered accelerain daa fr a 20 mph laeral impac. Peak accelerains vary frm apprximaely 100 g's fr unfilered daa 40 g's fr 50 Hz cuff frequency 28 g's fr 25 Hz cuff frequency. Als a dramaic reducin in he amun f scillain ccurs wih increased filering. EXCESSIVE COLLISION DURATION: The cllisin cnac durain repred by Jnes (see Table 2) appears be excessive. Fr example, Bundrf [15] deermined he pulse durain fr RICSAC Tes 9 be apprximaely secnds whereas Jnes repred he durain be secnds. Generally fr car--car cllisins, a nrmal range f pulse durain is A review f he ime-hisry accelermeer races and he inegraed velciy races cnained in he riginal RICSAC es reprs als appears indicae ha he durains f he pulses repred by Jnes are excessive. A cmparisn f he SMAC recnsrucin pulse durains indicaes ha an average adjusmen f he durain f apprximaely secnds wuld be apprpriae. Fr he purpses f his repred research, an apprximae adjusmen f was used (he average f ). RESULTS The inpus fr he repred raw RICSAC daa [2, 3] which were used wih he CG-Transfrm analyical echnique are cnained in Table 9. A cmparisn f he CG-Transfrm calculaed resuls wih he speed-change a he CG as prediced by he SMAC prgram is cnained in Table 10. A cmparisn f he CG-Transfrm calculaed separain velciies wih he separain velciies prediced by he SMAC prgram is cnained in Table 11. Given he pssible variains and prblems which can ccur in full-scale es measuremens he crrelain f he CG-Transfrm calculaed resuls and he SMAC resuls fr ms f he RICSAC ess appears very gd. In all cases he cied crrelain f CG-Transfrm wih he mahemaically crrec SMAC resuls is wihin ±10% fr V and wihin ±13% fr he separain velciy fr a leas ne f he vehicles. This gives a gd indicain ha prperly inerpreed resuls f all he ess are in cmplee agreemen wih Newn's Laws. Sme areas f difference can be explained as fllws: Fr he V values frm he full-scale ess, differences in he individual cmpnens may be aribuable pssible accelermeer calibrain errrs. Errrs als can be caused by crsscupling prblems wih gyrs which can ccur due vehicle rll and pich angles. Larger errrs may be prduced by prblems wih he individual accelermeers eiher in he crush zne r n cmpnens pssibly affeced by he impac cnfigurain. Fr he separain velciy, differences in he individual cmpnens and resulan are prduced by any f he fllwing pssibiliies: pssible errrs in deerminain f he ime f separain in he full-scale ess pssible errrs in he repred rienain f he vehicle a he ime f separain pssible synchrnizain prblems beween he riaxle accelermeers and he gyrs Table 9 CG-Transfrm inpus fr he 12 RICSAC full-scale ess raw daa

13 Table 10 Cmparisn f V a he CG based n CG-Transfrm calculaed resuls and SMAC recnsrucins fr he 12 RICSAC full-scale ess

14 Table 11 Cmparisn f Separain velciies based n CG-Transfrm calculaed resuls and SMAC recnsrucins fr he 12 RICSAC full-scale ess CONCLUSIONS 1. The RICSAC daa, when inerpreed wih he CG-Transfrm prcedure, are reasnably accurae and are suiable fr heir inended purpse f esing he validiy f recnsrucin echniques. 2. Previusly repred findings f grss errrs and vilains f Newn s laws in he repred RICSAC daa are errneus. 3. The SMAC prgram has been demnsraed crrelae well wih prperly analyzed full-scale es resuls RECOMMENDATIONS 1. Fuure validain cmparisns wih he RICSAC fullscale ess shuld als give cnsiderain uilizing he mahemaically crrec SMAC resuls presened herein. Use f he SMAC prgram resuls fr cmparisns avids he variabiliy ha nrmally ccurs in full-scale ess. Wihu direc measures f he variabiliy f experimenal respnses, by means f repeaed runs f ess, he accuracy f analyical predicins cann be prperly evaluaed. 2. Fuure full-scale ess shuld uilize a ime-hisry simulain prgram like SMAC r a hree-dimensinal equivalen permi a check f resuls and cnversin prcedures fr daa frm accelermeers n lcaed a he CG. 3. Any inegrain analysis f accelermeer daa shuld include cnsiderain fr pssible direcinal changes f individual cmpnens. This can be accmplished by direc analysis f he resulan accelerain avid he pssibiliy f a reducin in he repred values fr individual cmpnen velciies due accelerain direcinal reversals. 4. Full-scale ess shuld include prvisins fr separaely repring and analyzing he ccurrence f side-slap ypes f cllisins 5. Full-scale es repring and subsequen analysis shuld be in a crdinae sysem in cnfrmiy wih sandard SAE J670e REFERENCES 1. McHenry, R. R., Lynch, J. P., Segal, D. J., "Research Inpu fr Cmpuer Simulain f Aumbile Cllisins", Calspan Repr ZQ-6057-V-1, Cnrac DOT-HS , Jun Shemaker, N. E., "Research Inpu fr Cmpuer Simulain f Aumbile Cllisins - Saged Cllisins, Vl. II & Vl. III", Calspan Repr ZQ-6057-V-4 & V-5, Cnrac DOT-HS , December Jnes, I. S., Baum, A. S., "Research Inpu fr Cmpuer Simulain f Aumbile Cllisins - Vlume IV - Saged Cllisin Recnsrucins", Calspan Repr ZQ V-6, Cnrac DOT-HS , December Smih, R. A., Nga, J. T., "Examples f Saged Cllisins in Acciden Recnsrucin", presened a ASME Winer Meeing, Nv 1980 and cnained in ASME publicain "Highway Cllisin Recnsrucin" 5. James, M. E., Rss, H. E., Whiingn, C., "Imprvemen f Acciden Simulain Mdel and Imprvemen f Narrw Objec Acciden Recnsrucin", Texas

15 Transprain Insiue, DOT-HS , DOT-HS & DOT-HS , April Shaw, L. M., "Cunermeasures fr Side Impac", Cnrac DOT-HS Brach, R. M., "Impac Analysis f Tw-Vehicle Cllisins", SAE Paper Wley, R.L., Warner, C.Y., Tagg, M. D.,"Inaccuracies in he CRASH3 Prgram", SAE paper Wley, R.L, "The "IMPAC" Cmpuer Prgram fr Acciden Recnsrucin", SAE paper Day, T.D., Hargens, R.L.,"Furher Validain f he EDCRASH Using he RICSAC Saged Cllisins", SAE paper Cheng, P.A., Guenher, D. A., "Effecs f Change in Angular Velciy f a Vehicle n he Change in Velciy Experienced by an Occupan during a Crash Envirnmen and he Lcalized Dela-V Cncep", SAE Paper Day., T. D., Hargens, R. L., "Furher Validain f EDSMAC Using he RICSAC Saged Cllisins", SAE Paper N Smih, R.A., Nga, J. T., "The Accuracy and Sensiiviy f Crash", DOT-HS , March Cliff, E. C., Mngmery, D. T., "Validain f PC-Crash - A Mmenum-Based Acciden Recnsrucin Prgram", SAE Paper Bundrf, R.T., "Analysis and Calculain f Dela-V frm Crash Tes Daa", SAE Paper Day., T.D., Hargens, R.L.,"Furher Validain f EDSMAC Using he RICSAC Saged Cllisins", SAE Paper N A review f he rajecry aspecs f he RICSAC ess are expeced be cmpleed in early 1997 and will hen be published. 18. Mayr, R. P., Naab, K. N., "Basic Research in Aumbile Crashwrhiness - Tesing and Evaluain f Mdificains fr Side Impac", Crnell Aernauical Labrary, CAL Repr N. YB-2684-V-4, Ocber 1969) 19. McHenry R. R., Miller, P. M., "Aumbile Srucural Crashwrhiness", SAE Paper CONTACT INFORMATION Quesins r cmmens n he paper are welcmed and can be addressed he auhrs by: mchenry@inerpah.cm Psal Service Mail: 103 Brady Cur, Suie 200 WWW: Cary, NC USA hp:// APPENDIX 1: CG-TRANSFORM PROCEDURE In he fllwing, a prcedure is defined whereby accelermeer measuremens aken a lcains away frm he CG are ransfrmed in impac speed-changes a he cener f graviy and in separain velciies n he basis f discree measures f he vehicle respnses raher han cmplee imehisries. The prcedure, eniled CG-Transfrm, is validaed by means f applicains SMAC runs in which he upus include accelerain cmpnens a seleced lcains n he simulaed vehicles. DEFINITION OF SYMBOLS OXY = O X Y = ψ = Vehicle-fixed crdinae sysem wih rigin a he cener f graviy. Space-fixed crdinae sysem. Heading angle f vehicle relaive space-fixed sysem. u, v = Cmpnens f velciy in he direcins f he vehicle-fixed X and Y crdinaes respecively, inches/secnd. u, v = Cmpnens f velciy in he direcins f he space-fixed X and Y crdinaes respecively, inches/secnd. x1, y1 = Crdinaes f accelermeer lcain in vehicle-fixed sysem, inches. ax1, ay1 = a x, a y = u, v = Cmpnens f accelerain a pin 1 relaive he vehicle-fixed sysem, inches/sec 2. Cmpnens f accelerain a he cener f graviy relaive he vehicle-fixed sysem, inches/sec 2. Changes in velciy in he direcins f he vehicle-fixed X and Y crdinaes respecively, inches/sec. In he fllwing pages, a d ver a symbl is used indicae a ime derivaive. pin 1 The subleer 1 indicaes he value f a variable a

16 Frm (15) and (16), he cmpnens f accelerain relaive he vehicle-fixed sysem a he cener f graviy f he vehicle (i.e., x 1 = y 1 = 0) may be expressed: a u& vψ & (17) x = ay = v& + uψ & (18) Thus, he V cmpnens a he cener f graviy relaive he vehicle-fixed sysem, including rainal effecs, may be bained by inegrain f (15) and (16): a xd = a x d + y && ψd + x ψ & d (19) 2 a d = a d x && ψd + y ψ & d (20) y y Figure 5 Velciy cmpnens f a Pin n a vehicle T bain he cmpnens in he vehicle-fixed sysem, f he separain velciy (i.e., he velciy a he end f a perid f accelerain) i is necessary inegrae (17) and (18) in he fllwing manner: In Figure 5, velciy cmpnens f pin 1 relaive he vehicle-fixed crdinae sysem are defined by: u1 = u y1ψ & (11) v1 = v+ x1ψ & (12) u = u + f ud & = u + a xd + v & 0 ψ d (21) v = v + f vd & = v + ayd uψ & d (22) Frm he preceding, i may be seen ha he fllwing inegrains are required crrec readings bained frm accelermeers a pin 1: The velciy cmpnens f pin P relaive he space-fixed crdinae sysem may be expressed as: u 1 = u 1 cs ψ - v 1 sin ψ (13) v 1 = u 1 sin ψ + v 1 cs ψ (14) Subsiuin f (11) and (12) in (13) and (14) and differeniain f he resuling equains yields he fllwing cmpnens f accelerain relaive he vehicle-fixed sysem: ψd && & ψ 2 d vψ& d uψ& d (23) (24) (25) (26) 2 a = u& vψ& y && ψ xψ x1 1 1 & & 2 1 (15) a = v& + uψ& + x && ψ yψ (16) y1 1 Reasnable apprximains f he values f he inegrals lised as equains (23) hrugh (26) can be bained by he use f analyical funcins fied es daa: && ψ = Q 1 sin ω - Q 2 sin 3ω (27)

17 u = u f - ( u) cs ω (28) v = v f - ( v) cs ω (29) Inegrain f equains (23) hrugh (26) using he funcinal relainships defined by (27), (28) and (29) and applicain f he resuls equains (19), (20), (21) and (22) yields he fllwing relainships: π ω = (30) 2 ( ) y y1 1 1 V y = ad= a d x ψ& + ya (35) 1 π Q Q B = ψ 1 2 f 2 ω 2 & ω 9ω (36) 1 π Q1 C = f ω 4 ω ψ & (37) D= axd + Bv0 (38) Q 1 3ω π ψ& = f + 4 ψ & ω ψ (31) 2 3 Q 2 1 [ Q &] = 3 ω ψ (32) y 0 E = a d Bu delu = u = (39) D+ ( B+ C) E 2 (40) 1+ ( B+ C) 1 Q Q A = π 2 1 1π ψ & f + 2ψ& f ω 2 ω 4ω (33) Q2 Q2π + 2ψ& f 9ω 4ω x x1 1 1 V x = ad= a d+ y ψ& + xa (34) delv = v = ( E ( B+ C) D) 2 ( 1 + ( B+ C) ) (41) u f = u 0 + u (42) v f = v 0 + v (43)

18 REFERENCES 19 "Aumbile Srucural Crashwrhiness", McHenry R.R., Miller, P.M., SAE Paper "Research Inpu fr Cmpuer Simulain f Aumbile Cllisins", McHenry, R.R., Lynch, J.P., Segal, D.J., Calspan Repr ZQ-6057-V-1, Cnrac DOT-HS , Jun "Research Inpu fr Cmpuer Simulain f Aumbile Clliisins - Saged Cllisins, Vl II & Vl III", Shemaker, N.E., Calspan Repr ZQ-6057-V-4 & V-5, Cnrac DOT-HS , December "Research Inpu fr Cmpuer Simulain f Aumbile Cllisins - Vlume IV - Saged Cllisin Recnsrucins", Jnes, I.S., Baum, A.S., Calspan Repr ZQ-6057-V-6, Cnrac DOT-HS , December "Examples f Saged Cllisins in Acciden Recnsrucin", Smih, R.A., Nga, J.T., presened a ASME Winer Meeing, Nv 1980 and cnained in ASME publicain "Highway Cllisin Recnsrucin" 5 "Imprvemen f Acciden Simulain Mdel and Imprvemen f Narrw Objec Acciden Recnsrucin", James, M.E., Rss, H.E., Whiingn, C. Texas Transprain Insiue, DOT-HS , DOT-HS & DOT-HS ,April "Cunermeasures fr Side Impac", Shaw, L.M., Cnrac DOT-HS "Impac Analysis f Tw-Vehicle Cllisins", Brach, R.M., SAE Paper "Inaccuracies in he CRASH3 Prgram", Wley, R.L., Warner, C.Y., Tagg, M.D., SAE paper "The "IMPAC" Cmpuer Prgram fr Acciden Recnsrucin", Wley, R.L, SAE paper "Furher Validain f he EDCRASH Using he RICSAC Saged Cllisins", Day, T.D., Hargens, R.L., SAE paper "Effecs f Change in Angular Velciy f a Vehicle n he Change in Velciy Experienced by an Occupan during a Crash Envirnmen and he Lcalized Dela-V Cncep", Cheng, P.A., Guenher, D.A. SAE Paper "Furher Validain f EDSMAC Using he RICSAC Saged Cllisins", Day., T.D., Hargens, R.L., SAE Paper N "The Accuracy and Sensiiviy f Crash", Smih, R.A., Nga, J.T., DOT-HS , March "Validain f PC-Crash - A Mmenum-Based Acciden Recnsrucin Prgram", Cliff, E.C., Mngmery, D.T., SAE Paper "Analysis and Calculain f Dela-V frm Crash Tes Daa", Bundrf, R.T., SAE Paper "Furher Validain f EDSMAC Using he RICSAC Saged Cllisins", Day., T.D., Hargens, R.L., 17 A review f he rajecry aspecs f he RICSAC ess are expeced be cmpleed in early 1997 and will be published a he 1998 SAE Inernainal Cngress and Expsiin. 18 "Basic Research in Aumbile Crashwrhiness - Tesing and Evaluain f Mdificains fr Side Impac", Mayr, R.P., Naab, K.N.,Crnell Aernauical Labrary, CAL Repr N. YB-2684-V-4, Ocber 1969)