98S8W23

8/2/2019 98S8W23

1/8

VEHICLE FAR-SIDE IMPACT CRASHESRichard StolinskiRaphael GrzebietaDepartment of Civil EngineeringMonash University, ClaytonAustraliaBrian FildesAccident Research CentreMonash University, ClaytonAustralia.Paper 98-S8-W-23ABSTRACT

This is a summary of a paper which first appearedin the International Journal of Crashworthiness underthe title: Side Impact Protection - Occupants in theFar-Side Seat, Vol. 3, No. 2, pp 93-122. Readers aredirected to the full paper for a more comprehensivediscussion of t he issues presented here.Much of the applied vehicle side impact occupantprotection research to date has concentrated onoccupants seated beside the struck side of vehicles.These occupants are defined as near-side occupants.Real world crash evidence however has shown thatoccupants seated on the side away from the struckside, defined as far-side occupants, are still subjectto a risk of injury. This paper examines side impactepidemiology from an injury causation perspective,and endeavours to explain evidence indicating headinjuries and seat belt related injuries constitute asignificant proportion of all far-side impact injuries.Injury mechanisms and key dynamic parametersgoverning injury severity are detailed. Computermodels simulating the dynamic motion of vehicle far-side occupants are described. Occupant kinematicsand injury parameters from the models are thencompared with real world crash case studies. Thepaper finally suggests vehicle design strategies whichmay reduce far-side injuries. Some alternativerestraint systems are proposed as potentialcountermeasures to reduce occupant injuries.INTRODUCTION

Vehicle occupants are particularly vulnerable inside crashes. Australian studies by Fildes et al [ 1,2]have revealed that side impact crashes accounted for25 percent of all injury crashes and 40 percent ofserious injury crashes where an occupant was eitherhospitalised or killed. Vehicle side impact occupantprotection research has concentrated mostly on theoccupants seated beside the struck side of the vehicle.

These occupants are defined as near-side occupants.All regulatory side impact test standards focusexclusively on this scenario and rely on an assessmentprovided by one test intended to represent one pointin the spectrum of real world near-side impactcrashes.Real world crash evidence however has shownthat far-side occupants as illustrated in Figure 1, arestill subject to a risk of injury. The study by Fildes etal noted that around 40 percent of restrained sideimpact crash casualties were far-side occupants.Relatively little research literature is available thataddresses protection of far-side occupants. The scopeof this paper thus addresses far-side occupants wherethere may be opportunities for improved occupantprotection.Before any countermeasures to problems ofvehicle far-side impact crashworthiness can beproposed however, it is first necessary to establishclear injury patterns which characterise the problem.It then becomes possible to identify the injurymechanisms and finally propose countermeasurestrategies which can address these specific injuryprocesses.

Figure 1. Far-Side Impact Crash Configurati onSIDE IMPACT STATISTICS

Ginpil et al [3], examined data relating to crashesoccurring in Australia in 1990 where at least oneoccupant fatality resulted from an impact to the sideof the vehicle. They reported that 79 percent ofoccupants in the sample were killed in near-sideimpacts compared to 36 percent of occupants killed in

1819

8/2/2019 98S8W23

2/8

far-side impacts. Further, this study reported that far-side fatalities were more likely to sustain serious headinjuries than near-side fatalities. The report detailed26 percent of far-side fatalities sustained severeinjuries only to the head and no other region,compared with only 18 percent for near-side fatalities.Mackay et al [4] using UK field accident datadiscussed the relative importance of head injuries forrestrained far-side occupants. The report highlightedthe issue of torso restraint and detailed a frequentmechanism where far-side occupants can slip out oftheir chest restraint to make head contact withstructures on the opposite side of the vehicle cabin.Based on a study sample of 51 head injury cases, 35percent were judged to have come out of their seatbelt. Mackay also reported frequent far-side occupantinjuries to chest and abdominal body regions fromexcessive seat belt loads caused by the upper torsoslipping from the shoulder belt. The reportedmechanism of the upper torso flailing about in thecabin was considered a key element in the furtherinvestigation of far-side impact occupant injury.Subsequent sections of this paper deal with acloser examination of Australian injured occupantcases. Mathematical computer modelling simulationsof these cases have been developed which attempt toreconstruct the essential mechanisms and provide abasis for the selection of potential countermeasuredesigns.CRASHED VEHICLE CASE STUDIESCrashed Vehicle File

Side impact case studies by Fildes et al [l] detailedvehicle crash parameters together with injurydistributions for a sample of hospitalised and killedoccupants in Australian crashes. These case studieswere based on a sample of 198 side impact crashes ofpassenger vehicles. The study investigated 234injured or killed occupants and formed part of theMonash University Accident Research Centres(MUARC) crashed vehicle file collected for theAustralian Federal Office of Road Safety (FORS).The whole FORS sample compiled at the time ofwriting included 134 restrained occupants in near-sideimpacts and 52 restrained occupants in far-sideimpacts.Detailed information about the nature of eachcrash together with vehicle deformations and likelyinjury sources were recorded by MUARC researchers.Additionally, estimated vehicle delta-V values werederived using the Crash-3 computer program. Injuriesto various body regions were coded according to theAbbreviated Injury Scale (AIS), documented by theAmerican Association for Automotive Medicine, [.5].

1820

AIS is a threat-to-life scale ranging from AISl(minor severity) to AIS (non-survivable).Far-Side Case Study Sample

The sample of occupants killed or injured in far-side cases collected by MUARC was examined toestablish parameter patterns relevant to far-sideimpacts which typically lead to injury. The entrycriterion for the far-side sample was based on primaryimpact damage on the side opposite where the injuredor killed occupants were seated. To expedite thisreview, a subset of the overall MUA RC crashedvehicle file published as one page summaries wasanalysed. As the prime interest of this study is onrestrained occupants, cases where seat belts werejudged not to have been used, were excluded. Asample of 45 restrained cases classified as far-sideimpacts by MUARC was thus used. The sample ofinjured and killed occupants comprised 26 males and18 females. In the sample, 36 occupants were driversand 9 were passengers (2 in rear seats). Further, 16 ofthese far-side occupants were seated beside adjacent(near-side) occupants, whereas 29 far-side occupantswere not adjacent to any other occupant. The medianoccupant age in the far-side sample was around 30years.Case Study Vehicle Impact Data

The vehicle impact information from the casestudy sample was used to specify the significantimpact conditions applied in subsequent computermodelling simulations. Figure 2 shows the samplesdistribution of objects the car containing the subjectfar-side occupant impacted with, or was impacted by.Most cases involved impacts with other cars. Impactswith fixed objects such as trees and poles are notableas are impacts with heavier vehicles such as fourwheel drives which are included in the light truckcategory. Most impacts occurred to the mid sectionsof the subject vehicles.The sample of vehicles comprised almost equalproportions of vehicle size/weight categories. Thenumbers of small (under 900 kg), medium (900 to1200 kg) and large (over 1200 kg) passenger cars inthe sample were 14, 15 and 16 cases respectively.Figure 3 shows the distribution of vehicle mass ratios.

8/2/2019 98S8W23

3/8

Far-side25

b7 201d 15'E; 10g 5

05 +52 $2 d$Impacting Object E

Figure 2. Histogram of Impacted/ImpactingObjects, n = 45

The mass ratio in these cases was defined as themass of the far-side occupants vehicle divided by themass of the impacting vehicle. Around half of thecases in the sample have vehicle mass ratios under0.9. Figure 5 shows the distribution of the impactingobjects directional anglea with the side of thesubject occupants car as defined in Figure 4. Mostcases were classified as pure side impacts where thedirectional impact angle was around 90 degrees. Thesample is however skewed towards angles less than 90degrees where a rearward orientated velocitycomponent relative to the occupants car was presentat the time of impact.

under05 0.5to 0.89 0.910 1.1Mass Ratio

OVH1.1

Figure 3. Histogram of Impacting Vehicle MassRatios, n = 30 (subject vehicle mass/impactingvehicle mass)

Figure 6 shows the cumulative distribution of themaximum cabin intrusion for all cases in the far-sideinjury sample. Intrusion exceeded 400 mm in aroundhalf of the cases. In around 20 percent of the cases,the maximum intrusion exceeded 625 mm which isequivalent to half of the cabin width of a typical smallcar. Interestingly, 20 percent of far-side injury caseswere recorded where no intrusion was noted

Figure 4. Definition of Impact Geometry

Figure 5. Histogram of Directional Impact Anglea,n=4S

Figure 6. Cumulative Distribution of MaxIntrusion for All Cases, n = 45

Figure 7 shows delta-V distributions for light truckor car into car impacts. These vehicle crash delta-Vvalues were calculated by MUARC using the Crash-3computer program. It was not possible to calculateestimates of delta-V in all cases, thus Figure 7 isbased on data from 22 far-side impact cases. In thesample, 80 percent of cases have estimated delta-Vs

1821

8/2/2019 98S8W23

4/8

under 60 km/h, with half of the cases having delta-Vsestimated under 40 km/h.Based on the case study vehicle impact data,suitable test conditions to conduct subsequentdynamic simulation modelling evaluations of far-sideimpacts were established. These conditions wereconsidered representative of important segmentswithin the spectrum of real world crashes. Car to carsimulation cases were initially investigated with amedium size 1000 kg car. An impacting vehicle masshigher than the far-side occupants car was chosen,with a vehicle mass ratio of 0.9. Impact into the far-side occupant cars mid section at an angle of 90degrees was considered representative. Delta-Vsbelow 60 km/h were also considered relevant for theinvestigation.

100908070

8 60D0 50B 408 30

2010

010 20 30 40 50 60 70 80 OVW

80Delta-V km%

Figure 7. Cumulative Distribution of Delta-V forLight Truck or Car into Car Impact Cases, n = 22Case Study Occupant Injury Data

Occupant injury data from the far-side study casestudy sample was used to verify whether simulationload measurements correctly predicted occupantinjuries that would be expected in similar real-worldsituations. Details of injury severity outcomes andobserved injury loading mechanisms from the sampleof restrained far-side occupants are provided inStolinski et al [6]. The following is a summary of thefindings.The typical motion of a relatively unrestrainedupper torso pre-disposes far-side occupants to a riskof head injuries as the torso rotates across the cabininterior and the head contacts the vehicles far side orinstrument panel. Further, as the shoulder and chestslips out from under the sash part of a three point seatbelt, belt loads can be concentrated on the bodyregion impinged by the lap belt. Another mechanismobserved in high severity impacts is direct contactwith the intruding object. Where intrusionapproaches half the width of the car cabin, directcontact with an intruding object may be unavoidable.An additional important mechanism is direct contact

with adjacent occupants. It is possible that anadjacent near-side occupant may shield the far-sideoccupant from intruding objects or the vehicles far-side, however the impacts between occupantsthemselves would certainly involve some risk ofinjury.FAR-SIDE OCCUPANT/VEHICLE DYNAMICSInjury Tolerance

Questions arise as to what levels of the keydynamic measures of head velocity and lap seat beltload are critical. Kanianthra et al [7] and NHTSA [8]in the development of the US Upper Interior HeadProtection Standard - FMVSS No. 201, selected a freemotion headform impact test velocity of 15 mph (6.7m/s) for the evaluation of interior padding materials.This test speed was established on the basis of USNational Accident Sampling System (NASS) data,where it was noted to be the average occupant speedrelative to the cabin interior at which the onset ofserious injuries (between AIS 2 and AIS 3) are likelyto occur. McIntosh et al [9] conducted studies onhead impact tolerance in side impacts using cadavertests and computer simulations. Findings indicatedthat at impact velocities in excess of 6 m/s, headinjuries of an AIS 3 level were likely. McIntosh et alproposed tolerance levels of maximum headacceleration of 150 g and HIC of 7.50. The authorssuggested that even with padding, the likelihood ofbrain i njury was not reduced at these highervelocities.Leung et al [lo] proposed lap belt injury toleranceloads. Using cadaver tests, injuries were observed forlap-belt tension loads higher than 3 kN. Although thistolerance load was established with respect to frontalcollision kinematics, it is considered that a lateralimpact tolerance load may not differ too significantly.It is thought that during impact, the body rotatesaround its vertical axis so that the belt loads stillprimarily impinge on the forward portion of theabdomen. Nevertheless some fundamental research tobetter establish lateral impact tolerance loads stillneeds to be carried out.Occupant Dynamics - Sled Test Computer Model

A model of a restrained occupant in a far-sideimpact sled test situation was developed using theTN0 MADYMO occupant/vehicle crash simulationcomputer program. As previously noted, themechanism of occupants slipping out of their seatbelts is crucial to the realistic modelling of far-sideoccupant motion. It was therefore necessary to use afinite element seat belt model capable of simulating1822

8/2/2019 98S8W23

5/8

seat belt slip. The occupant in the sled test wassimulated using a multi-body 50th percentile HybridIII ellipsoid model from the MADYMO dummydatabase [ 111. The finite element seat belt wasavailable as a MADYMO application and is detailedin the MADYMO Applications Manual [ 121. Theseat structure was simulated using plane elements withdummy-seat and dummy-belt contact interactioncharacteristics established experimentally by TN0[121.As a validation exercise, initial modelling wasdone to replicate a series of sled tests in a GeneralMotors Research Laboratories program by Horsch etal [ 131. Sled test response data from both Hybrid IIdummies and cadavers was published by Horschmaking comparisons possible with the computermodel output data. As in the sled test program, a 100ms, log square accelerationpulse was applied in theMADYMO model.Figure 8 shows the computer model simulationkinematics that compared favourably with a sled testrun using an unembalmed cadaver [6]. A comparisonof responses from the Horsch test series [ 131 with thecomputer model simulation responses is summarisedin Table 1. Head acceleration levels from thedynamic motion itself are well below published injurytolerance levels. However if the head were to contactan interior cabin structure at a velocity exceeding 6m/s, as observed in the simulation, injury would beexpected. It is noted that measured maximum lap beltloads are above the abdominal injury tolerance levelssuggested by Leung.Occupant Dynamics - Car Test Computer Model

A sled test scenario is somewhat limited as realvehicle lateral impacts also involve varying levels ofrotational acceleration. A full car model alsoprovides the opportunity to model more realisticacceleration pulses developed during the interactionof two colliding vehicles. Further, it was considereduseful in subsequent studies, to develop the capabilityto model interactions between the far-side occupantand adjacent near-side occupants in the car. It wastherefore decided to assembl e a full car and occupantsimulation model to further study far-side occupantdynamics in full vehicle side impact situations.This model was developed [6] by integrating thesled test model, described in the previous section witha car side impact model originally developed tosimulate an ECE R95 [14,15] side impact test at 50km/h. Centre console and floor transmission tunnelsurfaces were added to the model for far-side sideoccupant contact interaction as these structures arepresent in most cars and are considered important tomodelling correct occupant dynamics.

Table 1.Comparison of Sled Test and Computer Model

Peak Responses

Load (kN)*Outboard Lap BeltLoad (kN)** 2.3 - 4.3 4.7

* Inboard Lap Belt Load - (adjacent dummysright side)**Outboard Lap Belt Load - (adjacent dummys leftside)The full car and occupant computer model was runin a number of test cases and the results are detailed inTable 2. The model at this stage has only been runwith the far-side occupant in the vehicle. A 1000 kgcar with a 1111 kg impacting barrier (mass ratio =0.9) was used. The subject car containing the far-sideoccupant was impacted mid section by the mobilebarrier at 90 degrees.It was observed that for far-side occupants where

the upper torso is relatively unrestrained, headvelocities were closely related to the car delta-Vs.The simulation results show that at an impact speed of40 km/h, neither the head velocity nor the lap beltload exceeded previously established tolerancevalues. At 50 kmlh, the head velocity exceeded thetolerance value. At 60 km/h, both the head velocityand the lap belt loads exceeded the tolerance values.The 60 km/h test speed resulted in only a 31 km/hdelta-V of the impacted car. Delta-Vs up to this levelin the MUARC far-side injury sample, as detailed inFigure 7, accounted for less than half of the injuredoccupants. Accordingly, the majority of injured casesoccurred beyond this level of delta-V. This isconsistent with the simulation findings that injurytolerance levels are certainly exceeded beyond thislevel of crash severity.

1823

8/2/2019 98S8W23

6/8

100 ms 120 ms

200 msFigure 8. Dynamic Motion in Far-Sided Impact Sled Test from MADYMO Simulation of Hybrid III DummyDuring log Side Impact Pulse.

Table 2.Peak Responses of Far-Side Occupant in the Car Test Computer ModelImpact Speed Head Velocity Lap Belt Load Car Acceleration Car Delta-V(km/h (m/s)* pi)** (g) *** (km/h, m/s>40 5.3 1.4 10.5 20.4,5.7

50 7.3 2.3 12.6 25.9, 7.260 9.1 3.5 14.6 31.1, 8.6

* head velocity relative to the car interior, ** outboard lap belt load, *** lateral acceleration measured at the carscentre of gravity

1824

8/2/2019 98S8W23

7/8

FAR-SIDE INJURY COUNTERMEASURESSeat Belt Pretensioners

If better upper torso lateral restraint could beprovided, the level of torso swing could be reduced,causing either fewer head impacts to occur with thecabin interior or at least at lower impact velocities.Seat belt pretensioning of the sash belt that isactivated during lateral impacts may be effective indelaying or reducing the extent of seat belt slip.Reduction of sash belt slip would also be expectedto reduce the concentration of lap belt loading bymaintaining more belt loading on the upper torso.Lateral Wing Seat Bolsters

Another countermeasure could be the provision ofhighly defined lateral wing torso restraints that arebuilt into seat structures. This design configuration iscommon in motor sport. Together with a full harness,this design provides a high level of lateral occupantrestraint for racing drivers. It appears likely howeverthat comfort issues could l ead to serious difficulties inconsumer acceptance of this configuration. Carefuldesign of inboard lateral bolsters would be needed toensure their acceptability in passenger vehicleapplications.As such designs provide more torso restraint,further research is needed to establish whetherunacceptable i nertial loads are transferred to theunrestrained head/neck of far-side occupants. Thehuman neck model developed by de Jager [16] maybe useful in this regard. It will also be important toensure that thoracic loads on far-side occupants do notexceed tolerance levels.Reverse Geometry Seat Belts

Another potentially effective countermeasureagainst far-side head impacts worth considering maybe the provision of reverse geometry integrated seatbelts. This design configuration has been implementedin the rear seats of so me BM W passenger vehicles.Some tests have shown however that belt/neck loadscould be excessive with this seat belt configuration.Horsch et al [13], who conducted cadaver sled tests,indicated that serious C5-C6-C7 cervical spineinjuries could occur. Conversely, Kallieris et al [ 171,using cadavers in full vehicle side impact tests,concluded that application of reverse geometry seatbelts did not cause neck injuries beyond A IS level 1.It appears therefore that more research is still neededinto the biomechanical effects of reverse geometryseat belts and potential benefits for far-side occupants.Reverse geometry integrated seat belts would however

incur cost and weight penalties for vehicles becausemore complex seat and re-inforced floor structureswould need to be designed.Inflatable Restraint Systems

Alternative solutions, involving inflatable airbagrestraint systems using existing technology may bepossible. This way, existing consumer expectations ofcomfort and space would not need to becompromised. A basic problem with inflatablesystems s the need to provide surfaces or supports foran airbag to react against whilst restraining anoccupant. It is possible that an inflatable curtaindevices originally intended for protection of near-sideoccupants could also provide protection for far-sideoccupants by providing a buffer against head impactswith far-side structures.CONCLUSIONS

In far-side impact situations, a key mechanismleading to injury is the upper torso slipping out of theshoulder belt. As a result, excessive head velocitiescan develop in the car together with excessive lap beltloads. This process is considered responsible formany of the head and abdominal injuries observed infar-side impact occupant injury case studies. Headvelocities relative to the car interior exceeding 6 m/sand lap belt loads higher than 3 kN have been shownto be critical values, above which, head or abdominalinjuries are likely. Other important loadingmechani sms include direct contact with i ntrudingobjects and contact with adjacent occupants.The simulation studies described in this paper haveshown that with impact velocities at or above 60 km/h(with delta-Vs exceeding 31 km/h), critical values ofhead velocity relative to the car interior or lap beltload can be exceeded in some crash situations. This isconsistent with real world case studies, where mostinjury cases in the sample were noted to occur abovethis level of crash severity.Some alternative restraint systems have beenproposed as potential countermeasures to reduce far-side impact injuries. These include seat beltpretensioners which are activated by lateralacceleration, lateral wing seat bolsters, reversegeometry seat belts and inflatable devices. Furtherresearch is recommended to investigate the usefulnessof these countermeasures, particularly to ensure thatneck injury tolerance loads are not exceeded.

182.5

8/2/2019 98S8W23

8/8

Further research work is also recommended toexamine:l larger sample sizes of real world far-side injurycases;l abdominal injury mechanisms from the lap beltunder lateral impact loading;l injury mechanisms during adjacent occupantinteractions and contact with intruding structures;l issues of eccentric and angled vehicle impact

geometry;l effects of pole and tree impacts.

Consideration should be given to the inclusion of afar-side occupant dummy in future side impact designrule test standards.Finally, it has been shown that opportunities existfor the development of new and innovative restraintsystems.ACKNOWLEDGMENTS

The authors wish to thank Stefan Nordin andRobert Judd of Autoliv Australia P/L for advice onrestraint system technologies that were discussed inthis paper and Dr. Clive Chirwa, editor of I.J. Crashfor al lowing the paper to be published in the ESVproceeding in summary form.REFERENCES1. B Fildes, J Lane, J Lenard, P Vulcan, PassengerCars and Occupant Injury: Side Impact Crashes,Australian Federal Ofice of Road Safety, Report No.CR134, 1994.2. B Fildes, K Digges, D Carr, D Dyte, P Vulcan,Side Impact Regulation Benefits, AustralianFederal Office of Road Safety, Report No. CR154,1995.3. S Ginpil, R G Attewell, A Jonas, CrashesResulting in Car Occupant Fatalities: Side Impacts,Australian Federal Ojjke of Road Safety, Report No.OR15, 1995.4. G M Mackay, J Hill, S Parkin, J A R Munns,Restrained Occupants on the Non-Struck Side inLateral Collisions, Proceedings of 13th ESVConference, Paris, France, 1991.5. American Association for Automotive Medicine,The Abbreviated Injury Scale: 1990 Revision ,Illinois.

6. R Stolinski, R Grzebieta, B Fil des, Side ImpactProtection - Occupants in the Far-Side Seat,International Journal of Crashworthiness, Vol.3No. 2, pp 93-122, 1998.7. J N Kanianthra, W Fan, G Rains, Upper InteriorHead Impact Protection of Occupants in Real WorldCrashes, Proceedings of 15th ESV Conference,Melbourne Australia, 1996.8. NHTSA, FMVSS No. 201 - Upper Interior HeadProtection - Final Economic Assessment, NHTSAOffice of Regulatory Analysis, 1995.9. A S McIntosh, N L Svensson, D Kallieris, RMattern, G Krabbel, K Ikels, Head Impact Tolerancein Side Impacts, Proceedings of 15th ESVConference, Melbourne Australia, 1996.10. Y C Leung, C Tarriere, D Lestrelin, C Got, FGuillon, A Patel, Submarining Injuries of ThreePoint Belted Occupants in Frontal Collisions -Description, Mechanisms and Protection, SAE PaperNo 821158.11. TNO, MADYMO Database Manual - Version5.2, TN0 Road-Vehicles Research Institute, 1996.12. TNO, MADYMO Applications Manual - Version5.2, TN0 Road-Vehicles Research Institute, 1996.13. J Horsch, D Schneider, C Kroell, Response ofBelt Restrained Subjects in Simulated LateralImpact, SAE Paper No 791005.14. P J A de Coo, E G Janssen, A P Goudswaard, JWismans, M Rashidy, Simulation Model for VehiclePerformance Improvement in Lateral Collisions,Proceedings of 13th ESV Conference, Paris, France,1991.15. ECE Regulation No.95, Uniform Provisionsconcerning the Approval of Vehicles with regard tothe Protection of the Occupants in the event of aLateral Collision. October 1995.16. M de Jager, Mathematical Head-Neck Models forAcceleration Impacts, Eindhoven University ofTechnology, PhD Thesis, 1996.17. D Kallieris, G Schmidt, Neck Response andInjury Assessment Using Cadavers and the US-SIDfor Far-Side Lateral Impacts of Rear Seat Occupantswith Inboard Anchored Shoulder Belts, Proceedingsof 34th Stapp Car Crash Conference, Paper No902313.

1826