Numerical simulations of the occupant head response in an infantry vehicle under blunt impact and blast loading conditions

Original Article

Proc IMechE Part H:J Engineering in Medicine227(7) 778–787� IMechE 2013Reprints and permissions:sagepub.co.uk/journalsPermissions.navDOI: 10.1177/0954411913483430pih.sagepub.com

Numerical simulations of the occupanthead response in an infantry vehicleunder blunt impact and blast loadingconditions

Gopinath Sevagan1, Feng Zhu1, Binhui Jiang1,2 and King H Yang1

AbstractThis article presents the results of a finite element simulation on the occupant head response in an infantry vehicleunder two separated loading conditions: (1) blunt impact and (2) blast loading conditions. A Hybrid-III dummy bodyintegrated with a previously validated human head model was used as the surrogate. The biomechanical response ofthe head was studied in terms of head acceleration due to the impact by a projectile on the vehicle and intracranialpressure caused by blast wave. A series of parametric studies were conducted on the numerical model to analyzethe effect of some key parameters, such as seat configuration, impact velocity, and boundary conditions. The simula-tion results indicate that a properly designed seat and internal surface of the infantry vehicle can play a vital role inreducing the risk of head injury in the current scenarios. Comparison of the kinematic responses under the bluntimpact and blast loading conditions reveals that under the current loading conditions, the acceleration pulse in theblast scenario has much higher peak values and frequency than blunt impact case, which may reflect different headresponse characteristics.

KeywordsOccupant safety, infantry vehicle, head injury, blunt impact, blast impact, finite element analysis

Date received: 5 September 2012; accepted: 22 February 2013

Introduction

Increasing number of Warfighters suffering fromimpact- and blast-induced traumatic brain injury (TBI)has been reported in recent military conflicts.1,2 Suchinjury has caused a high rate of mortality and morbid-ity among the victims, and it is of current interest inboth clinical and academic settings. The investigationsinto the TBI are either experimental or computational.Compared to experimental studies, properly conductedcomputational studies can save time, cost, and laborsignificantly in understanding of injury mechanismsand injury thresholds. A computational model enablesresearchers to study interaction of complex parameters,such as stress transfer, energy dissipation, and wavepropagation, which are very difficult or not possible tomeasure experimentally.

Currently, most of the numerical head models avail-able for TBI research are focused on direct impact by aprojectile in the case of impact TBI3–10 and shock waveloading in the open field11,12 or shock tube13,14 in the

case of blast TBI. Such idealized studies disregarded theinfluence of environmental factors, such as buildingsand vehicles. These environmental factors, however, arevery common in the real battlefield and may have con-siderable effect on the response of Warfighters and theextent of injury.15

To overcome the limitation of neglecting environ-mental influence, effort has been made to develop finiteelement (FE) models to simulate the occupant headresponse in an infantry vehicle subjected to bluntimpact and blast loading. A comparison was madebetween these two different loading scenarios, which

1Bioengineering Center, Wayne State University, Detroit, MI, USA2The State Key Laboratory of Advanced Design and Manufacturing for

Vehicle Body, Hunan University, Changsha, China

Corresponding author:

Feng Zhu, Bioengineering Center, Wayne State University, Detroit, MI

48201, USA.

Email: [email protected]

have not been reported in the current literature. In thisstudy, a human head model was integrated into aHybrid-III crash dummy torso, which was placed in asimplified infantry vehicle model. The impact load wasproduced by a rigid projectile, and the blast wave wasbased on a planar pulse. Relatively low load levels werechosen in this study such that the energy released by theprojectile or blast does not cause penetration damageon the vehicle but considerable head response can beobserved. Modeling severe structural damage is beyondthe scope of this study. Biomechanical responses, forexample, the acceleration of the head and intracranialpressure (ICP) in the brain, were analyzed in detail.Since strain is very low in blast scenario, ICP is thewidely accepted main response variable.2 Consideringprevious publications that have shown that rotationalacceleration was not the main culprit of elevated ICP,this study does not include this loading scenario. Basedon this model, a series of parametric studies were con-ducted to discuss the effect of some key parameters,such as configurations of the seat arrangement, velocityof projectile, and the boundary conditions.

The numerical model

As mentioned above, two loading scenarios were consid-ered separately in this study, namely, blunt and blastimpacts, which are detailed in the following subsections.All the simulations were implemented using an explicitFE solver LS-DYNA 971 (LSTC, Livermore, CA, USA).

Blunt impact

In blunt impact loading conditions, effects of adding afoam cushion to the seat and changing the impact

velocity of a rigid projectile in a simplified infantryvehicle model on the acceleration response of the occu-

pant head were the main focus of this study.

Hypermesh 10.0 (Altair, Irvine, CA, USA) was used

as preprocessor to develop the FE mesh of a simpli-

fied infantry vehicle. The basic profile of the vehicle

was kept and all internal structures were neglected

since they have little influence on the overall

response. A simple seat model was fixed to one side

of infantry vehicle as shown in Figure 1. The dummy

head complex was facing the positive x- (posterior to

anterior), y- (to the left), and z- (inferior to superior)

directions, respectively. A mass element was added on

the center of gravity (CG) of the vehicle such that the

total weight was 15 ton. A cylindrical projectile with

a hemi-spherical end and the mass of approximately

10 kg was used to apply the blunt impact on left side

of the simulated tank where the seat is located. The

projectile has the length of 90mm and the radius of

88mm.The whole vehicle was modeled with 45,136 thin shell

elements, and the armor was assumed to be made of a

high-strength steel with an average thickness of 15mm.

Its material properties were described with MAT_99

(*MAT_SIMPLIFIED_JOHNSON_COOK_ORTHO-

TROPIC_DAMAGE) in LS-DYNA. Specific para-

meters are given as follows: Young’s modulus E of

207.0GPa, density r of 7830 kg/m3, yield stress sY of

1540MPa, and a principal plastic strain at failure ef of0.4. It should be noted that the lightweight infantry

vehicle in this study is not designed to resist large-

caliber armor piercing shells. It can only protect occu-

pants against machine guns and shell fragments with

lower speed.

Figure 1. Configuration of the simplified vehicle FE model and the projectile.

Sevagan et al. 779

Since one of the focuses in this study was to analyzethe human brain/skull response, a human head modelwith skull and brain assembly is necessary. A previ-ously validated human head FE model developed atWayne State University (WSU)16 was integrated into aHybrid-III dummy torso model and used as the humansurrogate. The head model has dimensions of 198mmin length, 140mm in width, and 222mm in height andthe weight of 4.6 kg. It was modeled with 23,540 ele-ments and consists of some vital components such asthe skull, brain, dura maters, falx cerebri, tentorium,and a simplified jaw. The brain tissue was modeled as aviscoelastic material with the following properties:17

short-term shear modulus G0=10.8 kPa, long-termshear modulus Gi=3.1 kPa, and time decay coefficientb=456. Bony materials were modeled as an elasticmaterial with Young’s modulus E=12GPa.

The integration of human head with dummy neck wasimplemented by removing the dummy head and connect-ing the WSU head model onto the revolute joint on theneck, as shown in Figure 2. The CG of the human headmodel was found and aligned to the location of the accel-erometer mounted on the original dummy head model tomeasure head accelerations. After integration of thehuman head and dummy torso models, the dummy wasplaced over a simplified seat and constrained with a four-point seat belt. Less than a millimeter of clearance wasgiven between dummy components and foam compo-nents to avoid initial penetration.

Using this dummy/vehicle model, the following twoseries of parametric study were conducted.

Effect of foam layer on the seat. In this section, a plasticfoam material with a thickness of 40mm was placed onthe seat as a protective structure to mitigate the kineticenergy transfer. Four different configurations are

considered here, namely, (i) both seat back and seatpan are rigid, (ii) both seat back and seat pan are cov-ered with foam, (iii) seat back is covered with foamwhile seat pan is rigid, and (iv) seat back is rigid whileseat pan is covered with foam.

The foam layer was modeled with solid elements,and the material properties were described usingMAT_63 (*MAT_CRUSHABLE_FOAM) card in LS-DYNA, which is frequently applied to model the beha-vior of energy absorbing cushions. Young’s modulus ofthe low density foam is 13MPa, and the compressivestrength is 0.4MPa. This type of foam was designed todissipate kinetic energy during plastic deformation.It has only slight deformation under the gravity ofthe dummy. The impact velocities of the projectilewere identical in all the four cases and assumed to be400m/s.

Effect of impact velocity. Four initial velocities (100, 200,300, and 400m/s) were assigned to the projectile, andthe resulting head acceleration responses due to theseimpact velocities were studied. In this study, the highestimpact velocity (400m/s) was carefully selected suchthat the vehicular wall could not be fully perforated bythe projectile but considerable head acceleration couldbe generated. In the case of a full penetration, the defor-mation of the vehicle would be highly localized, and theinjury of the occupant is mainly caused by flying frag-ments within the vehicular cabin. The aim of this studyis to investigate the head response due to inertia loadingproduced by the blunt impact of projectile on the vehi-cular wall. The projectile had a fixed incident angle of90� with the side wall of the vehicle, that is, it was per-pendicular to the impact surface. In this section, boththe seat back and seat pan were protected with foamlayers.

Figure 2. Integration of a Hybrid-III dummy torso and WSU head models.WSU: Wayne State University.

780 Proc IMechE Part H: J Engineering in Medicine 227(7)

Blast impact

Shock wave reflection in the closed environment can causemore injuries than that in the open field.15 Therefore, acomparison study was conducted to investigate such effectusing the FE models described previously. In order toreduce the computational time, the vehicle model was fur-ther simplified to include only the passenger compartment.The pressure load selected is relatively low (1.1MPa) tosimulate a shock wave large enough to produce consider-able ICP without inducing large deformation or motionon the vehicle. This pressure level represents a 40 kg ofTNT (trinitrotoluene) at a standoff distance of 4 m to thefront surface of the vehicle.

The surrogate was facing to an opening throughwhich blast wave is able to propagate. The opening wasused to simulate a window, door, or gun port on thevehicle. An air domain containing the structures wasmodeled using solid elements with arbitraryLagrangian–Eulerian (ALE) formulation, throughwhich the shock wave can transmit.18 The behavior ofair was described using MAT_9 (*MAT_NULL)together with corresponding equation of state (EOS). Atriangular shock wave (peak pressure: 1.1MPa, dura-tion 2.4 ms) was directly applied on the surface of theair domain facing to the opening of the vehicle tomodel a planar blast wave in far-field explosion. Theinteraction of shock wave and head was implementedusing Lagrangian/ALE coupling algorithm, that is,*CONSTRAINED_LAGRANGE_IN_SOLID func-tion in LS-DYNA, where the force between a solidnode and a fluid node is calculated based on their dis-tance and material stiffness. In this algorithm, thesurrogate (Lagrangian part) was surrounded by thefixed air domain (Eulerian part) during the wholeblast loading procedure. In other words, the solid andfluid elements coexisted in the same space. The fluid–solid interaction (FSI) parameters used in this studywere consistent with those in the study by Zhuet al.13,14 A parametric study was conducted on thenumber of coupling points (1, 2, and 3), and it hasbeen revealed that the coupling points had little effecton the results. Therefore, to save computational cost,

one coupling point was used. It is noted that simula-tion of blast wave propagation in the air requiresmodeling air domain with a large number of elements,and fluid/coupling algorithm is very time-consuming.Considering the limited computational capability,an air domain with relative coarse mesh was used.More importantly, it has been reported that the shockwave is nearly planar in the area close to the head inthe loading scenario as selected, a relative coarsemesh should have no significant influence on theresults.14

Two boundary conditions were considered, namely,nonreflection and reflection, as illustrated in Figure3(a) and (b), respectively. In both cases, a part of theside wall with a square port facing to the blast wasmodeled such that the incident pressure waves wereidentical. In the nonreflection case, the wall behind thedummy, roof, and floor were removed to eliminatetheir wave reflection effect. For reflection conditions,however, the wave reflection behavior at these surfaceswas taken into account.

Results

Simulation results in both loading conditions arereported and analyzed in sections ‘‘Blunt impact’’ and‘‘Blast impact,’’ respectively. All the data were filteredusing Society of Automotive Engineers (SAE) filteringalgorithm at a channel frequency class of 1000 Hz.

Blunt impact

Figure 4 illustrates typical high-speed impact events ata projectile velocity of 400m/s. The projectile hits theinfantry vehicle at 3 ms and then bounds back. Thevehicle starts to move and the wall deforms. The kineticenergy is transmitted to the dummy through the seatand then causes the motion of the head. Typicalacceleration–time histories of the head in three direc-tions are shown in Figure 5. The impact velocity was300m/s, and both the seat back and seat pan were pro-tected with foams. In Figure 5, it can be seen that head

Figure 3. Two boundary conditions in the blast impact simulation: (a) without reflection and (b) with reflection.

Sevagan et al. 781

x- and z-accelerations (ax and az) exhibit similar pat-terns, that is, a sharp increase in the magnitude andthen followed by a negative pulse. After the curves goback to zero, significant oscillations take place.Compared to ax and az, ay is much smaller, which indi-cates that the motion of the dummy head is mainly in

the impact direction and vertical direction, and thedegree of lateral motion is rather low.

Effect of foam layer on the seat. Figure 6 shows the peakvalues of ax, ay, az, and resultant acceleration (ar) for

Figure 4. Blunt impact events at the projectile velocity of 400 m/s.

Figure 5. Typical x-, y-, and z-acceleration traces of the head under blunt impact loading condition (impact velocity: 300 m/s; seathas the configuration (ii): both seat back and seat pan are foamed).

782 Proc IMechE Part H: J Engineering in Medicine 227(7)

the four seat configurations described in the ‘Thenumerical model’ section loaded at the projectile velo-city of 400m/s. In all four cases, again, the peak valuesof ax and az are much higher than that of ay.Configurations (i) and (iv) have similar peak accelera-tions, which are higher than the other two cases by over100%. This indicates that the foam layer at the seatback can mitigate a large amount of kinetic energydelivered by the projectile. Comparison of ar betweenconfigurations (ii) and (iii) shows that the foam layer onthe seat pan is less important due to blunt impact from

side wall since the difference in the magnitudes of peakaccelerations in these two cases is not more than 20%.

Effect of impact velocities. The peak head accelerations atfour impact velocities are given in Figure 7. Again, axand az are much higher than ay. The accelerationsincrease with impact velocity. But when increasing thevelocity from 300 to 400m/s, the increase in accelera-tion is not evident, which indicates that the structuraldeformation of the vehicle body can dissipate a largeamount of kinetic energy of the projectile.

Figure 6. Peak head accelerations of the head model at different foam placement configurations, that is, (i) both seat back and seatpan are rigid, (ii) both seat back and seat pan are foamed, (iii) seat back is foamed while seat pan is rigid, and (iv) seat back is rigidwhile seat pan is foamed.

Figure 7. Peak head x-, y-, and z-accelerations for different impact velocities.

Sevagan et al. 783

Blast impact

The ICP and motion of the head under two boundaryconditions shown in Figure 3 are described in this sec-tion. A blast wave is characterized by a supersonic sta-tic shock wave followed by a blast wind. The shockwave can induce stress wave and shear wave in thebrain then increase the magnitude of ICP. Blast wind,however, can cause the global motion of the head.Figure 8 demonstrates the blast loading events in bothboundary conditions. It can be seen that a planar wavehits the wall of the vehicle and then transmits into thevehicle through the opening. A spherical wave is gener-ated at the interface at about 1.8 ms and then propa-gates and expands the volume in the vehicle. After thewave interacts with the head at 2.5 ms, it continues tra-veling. In the case of nonreflection boundary condition,the shock wave will be mitigated gradually in the air,while in the reflection case, the wave is reflected backat the surface of the back wall and then induces the sec-ondary shock to the head.

Figure 9 demonstrates the shock wave applicationon the head model. With the high-pressure wave trans-mitted into the brain, a high-pressure zone is generatedin the coup site and then moves backward to the contre-coup site. The ICP traces at the center of the brain inboth conditions are illustrated in Figure 10. The twocurves exhibit similar modes: a positive pulse with theduration of 1 ms followed by a negative phase. Then,the pressures go back quickly and oscillations can beobserved. The peak ICP in the reflection case is higherthan that of nonreflection case by 38%. Since the totalpressure at a target is the combination of incident pres-sure and reflected pressures (caused by roof, back face,floor, and so on in the present case),19 higher pressure istransmitted into the brain in the reflection case andthen causes higher ICP. Therefore, it is suggested thatthe internal surface of a military vehicle should be pro-tected with properly designed energy absorption

materials. But the material properties, thickness, andshape of the foam claddings need to be determinedcarefully.20

Figure 11 shows the motions of the head under thesetwo conditions at approximately 9.1 ms. In the nonre-flection case, the head is pushed to move backward;while for reflection condition, due to the wave reflectionof internal wall behind the dummy, the pressures in theopposite directions are in a state of relative equilibrium.Therefore, the head motion in the loading direction isnot evident. However, because of the complex wavereflection behavior, the head has a left rotation withapproximately 25�.

Discussion

A comparative study was conducted on the bluntimpact and blast loading conditions to reveal the differ-ence in their kinematic responses. In both cases, theimpulses per unit area transmitted into the brain CG(i.e. integration of ICP over time) were kept identicaland equal to 0.32 Pa�s, which could be consideredcaused by a projectile at 300m/s in the blunt impactscenario and an overpressure with the magnitude of 3bar and duration of 0.7 ms in the blast scenario. In thisstudy, the analyses were concentrated on the compari-son of response–time histories in two different loadingscenarios and the influence of the load–time wave formon the response–time curve. Since energy is not relatedto loading time duration, identical impulse was appliedon both modes for comparison purpose. The seat wasprotected with foam layers in both loading scenarios,and reflection boundary conditions were used in theblast case. Figure 12(a) and (b) illustrates the compari-son of time histories of ax and az, respectively, near theCG of the head. In Figure 12(a), it can be seen that theacceleration response under blast loading has a muchhigher frequency. The first pulse in the blasting case

Figure 8. Blast loading events in nonreflection and reflection conditions.

784 Proc IMechE Part H: J Engineering in Medicine 227(7)

has a duration of 2 ms while that in the blunt impactcase is 6.3 ms. For the acceleration in z-direction, the

peak values in both loading conditions are much moresimilar, as shown in Figure 12(b). However, again, the

Figure 9. Blast wave transmission into the head model.

Figure 10. Intracranial pressures at the center of gravity of the head model under reflection and nonreflection conditions.

Sevagan et al. 785

pulse in the blunt impact has much longer durationthan in the blast scenario.

There are some limitations in this study. The currentcomputational analysis was semiquantitative with theemphasis on ‘‘trends’’, rather than on actual values.

Due to the lack of real-world experimental data, nocomparison with testing results was conducted, and thehead model was validated only at simple loading con-figurations. The focus of the analysis was placed on thebiomechanical response without injury considered. The

Figure 11. Comparison of head motions under reflection and nonreflection conditions at approximately 9.1 ms.

Figure 12. Comparison of linear accelerations near the CG of the head under blunt and blast impact conditions: (a) ax and (b) az.

786 Proc IMechE Part H: J Engineering in Medicine 227(7)

infantry vehicle model was highly simplified and onlybasic geometry was kept. The mechanical properties ofthe armor are unknown, and the material parameterstaken from open literature were used. These issues canweaken the accuracy of the simulations. In addition, awider range of loading levels will be studied in thefuture work when more real-world data becomeavailable.

Concluding remarks

FE models have been developed to simulate the occu-pant head response in an infantry vehicle under bluntimpact and blast loading conditions. The geometry ofthe vehicle was significantly simplified, and the materialproperties were taken from the literature. To study thebiomechanical response in the brain, a validated humanhead model was integrated into a Hybrid-III dummytorso, which was seated in the simulated military vehi-cle. In the blunt impact loading condition, a projectilehits the vehicle and the response of the head wasrecorded. Parametric studies were conducted on themodel to analyze some key parameters such as seatconfiguration and impact velocity on the accelerationresponse. It has been revealed that the foam layer onthe seat back has a significant contribution in energydissipation, and a large amount of kinetic energy of theprojectile was absorbed by the vehicle structure, partic-ularly at higher impact velocity. For blast impact, anair domain with ALE formulation was built to allowshock wave transmission. The planar shock wave cantravel into the vehicle through an opening. If wavereflection at the internal surfaces is considered, a sec-ondary shock on the head can induce higher ICP thanin the nonreflection condition. Comparison of the kine-matic responses under the blunt impact and blast load-ing conditions reveals that under the current loadingconditions, the acceleration pulse in the blast scenariohas much higher peak values and frequency than bluntimpact case, which may reflect different head responsecharacteristics.

Funding

This research is partially supported by GeneralDynamics Land Systems and partially supported bythe Bioengineering Center, Wayne State University.

Acknowledgement

The financial supports are gratefully acknowledged.

Conflict of Interest

The authors declare that there is no conflict of interest.

References

1. Okie S. Traumatic brain injury in the war zone. N Engl J

Med 2005; 352: 2043–2047.

2. Taber KH, Warden DL and Hurley RA. Blast-related

traumatic brain injury: what is known? J Neuropsychiatry

Clin Neurosci 2006; 18(2): 141–145.3. Claessens M, Sauren F and Wismans J. Modeling of the

human head under impact conditions: a parametric

study. In: 41st Stapp car crash conference, Lake Buena

Vista, FL, 12 November 1997, SAE paper 973338, 1997.4. Kang H, Willinger R, Diaw RM, et al. Validation of a

3D anatomic human head model and replication of head

impact in motorcycle accident by finite element model-

ing. In: 41st Stapp car crash conference, Lake Buena

Vista, FL, 12 November 1997, SAE paper 973339, 1997.5. Zhang L, Yang KH, Dwarampudi R, et al. Recent

advances in brain injury research: a new human head

model development and validation. Stapp Car Crash J

2001; 45: 369–394.

6. Zong Z, Lee HP and Lu C. A three-dimensional human

head finite element model and power flow in a human head

subject to impact loading. J Biomech 2006; 39(2): 284–292.7. Kleiven S. Evaluation of head injury criteria using a finite

element model validated against experiments on localized

brain motion, intracerebral acceleration, and intracranial

pressure. Int J Crashworthines 2006; 11(1): 65–79.8. Ho J and Kleiven S. Can sulci protect the brain from

traumatic injury? J Biomech 2006; 42: 2074–2080.9. Cloots RJH, van Dommelen JAW, Nyberg T, et al.

Micromechanics of diffuse axonal injury: influence of

axonal orientation and anisotropy. Biomech Model

Mechanobiol 2011; 10: 413–422.10. Wright RM and Ramesh KT. An axonal strain injury cri-

terion for traumatic brain injury. Biomech Model

Mechanobiol 2011; 11: 245–260.11. Taylor PA and Ford CC. Simulation of blast-induced

early-time intracranial wave physics leading to traumatic

brain injury. J Biomech Eng 2009; 131: 1–11.12. Charfi MS, Karami G and Ziejewski M. Biomechanical

assessment of brain dynamic responses due to blast pres-

sure waves. Ann Biomed Eng 2010; 38(2): 490–504.13. Zhu F, Mao H, Dal Cengio Leonardi A, et al. Develop-

ment of an FE model of the rat head subjected to air

shock loading. Stapp Car Crash J 2010; 54: 211–225.14. Zhu F, Skelton P, Chou CC, et al. Biomedical responses

of a pig head under blast loading: a computational simu-

lation. Int J Numer Method Biomed Eng, 2012; 13: 25–29.15. Bauman RA, Ling G, Tong L, et al. An introductory

characterization of combat-casualty-care relevant swine

model of closed head injury resulting from exposure to

explosive blast. J Neurotrauma 2009; 26: 841–860.16. Ruan JS, Khalil T and King AI. Human head dynamic

response to side impact by finite element modeling. J Bio-

mech Eng 1991; 113(3): 276–283.17. Zhu F, Jin X, Guan F, et al. Identifying the properties of

ultra-soft materials using a new methodology of com-

bined specimen-specific finite element model and optimi-

zation techniques. Mater Design 2009; 31: 4704–4712.18. Souli M. LS-DYNA advanced course in ALE and fluid/

structure coupling. Livermore, CA: Livermore Software

Technology Co., 2004.19. ConWep (Conventional Weapons Effects program).

Vicksburg, MS: US Army Engineer, Waterways Experi-

mental Station, 1991.20. Zhu F, Chou CC and Yang KH. Shock enhancement

effect of lightweight composite structures and materials.

Compos Part B: Eng 2011; 42: 1201–1211.

Sevagan et al. 787