Embed Size (px)
Citation preview
http://pil.sagepub.com/and Applications
Engineers, Part L: Journal of Materials Design Proceedings of the Institution of Mechanical
http://pil.sagepub.com/content/227/1/82The online version of this article can be found at:
DOI: 10.1177/1464420712451806
originally published online 22 June 2012 2013 227: 82Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials Design and Applications
Tso-Liang Teng, Cho-Liang Liang and Van-Hai NguyenDevelopment and validation of finite element model of helmet impact test
Published by:
http://www.sagepublications.com
On behalf of:
Institution of Mechanical Engineers
can be found at:ApplicationsProceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials Design andAdditional services and information for
http://pil.sagepub.com/cgi/alertsEmail Alerts:
http://pil.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://pil.sagepub.com/content/227/1/82.refs.htmlCitations:
What is This?
- Jun 22, 2012OnlineFirst Version of Record
- Jan 10, 2013Version of Record >>
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
Case Study
Development and validation of finiteelement model of helmet impact test
Tso-Liang Teng1, Cho-Liang Liang2 and Van-Hai Nguyen2
Abstract
Recently, many researchers have investigated helmet design with respect to helmet performance in shock absorption
tests. The finite element method contributes greatly to helmet test modeling. Finite element simulations are also
indispensable to evaluate head injury criteria. This study performs finite element analyses of helmet impact tests using
LS-DYNA software. To compensate for lack of oblique impact tests in helmet standards, this article considers two
oblique impact tests. It measures linear accelerations, rotation accelerations, and impact forces of the headform from
helmet test simulations. To verify the accuracy of the proposed method, this study compares simulation results with
experimental data. The results of this method fit the experimental results well, implying that the numerical method is a
practical approach to helmet design problems. Furthermore, the helmet test model proposed here has potential for
guiding the future development of helmet technologies.
Keywords
Bike helmet, head injury, oblique impact, headform, LS-DYNA
Date received: 18 January 2012; accepted: 11 May 2012
Introduction
With the increasing popularity of bike sports come anincreasing number of bicyclists who would suffer ser-ious injuries caused by accidents, especially head inju-ries. According to the US National Highway TrafficSafety Administration (NHTSA), approximately51,000 bicyclists were injured in traffic in 2009. Thiswas a sharp rise from 43,000 in 2007, in which bicyc-list deaths represented 2% of all traffic fatalities.Significantly, nearly all bicyclists who died (97%)were not wearing a helmet, whereas the number ofbicyclists with serious injuries, using a helmet, waslow (13%). Notably, it was even lower amongbicyclists killed (3%) in New York City.1 Obviously,wearing a bike helmet can effectively reduce the risk ofhead injury.
The bike helmet is designed to attenuate impacts tobicyclists’ skulls in falls. Generally, a bike helmet con-sists of the outer shell, liner, vents, and straps. Its shellprovides the shape of the helmet and resists punctures.The liner helps absorb the effects of the impact tokeep that force away from your head. Vent holes inthe helmet help to keep the head cool and straps keepthe helmet on the head. The parts of the helmet aresimilar in all styles, but there are different types ofhelmets depending on riding style.
Bike helmets must meet minimum standards ofconstruction and materials design. The AmericanNational Standards Institute and US Consumer
Product Safety Commission (CPSC) have createdtest methods and minimum performance criteriathat assess bike helmet capabilities to protect bicyc-lists during accidents. These standards involve the useof an instrumented headform that is dropped, wearinga helmet, onto various anvils. The speed of impact isdesigned to simulate the effect of a rider’s head fallingfrom approximate usual riding heights. The testshould satisfy certain standards or regulations; forexample, peak acceleration of the headform for theCPSC’s bicycle helmet standard should not exceed300G.2
Recently, many researchers have investigatedhelmet design with respect to helmet performance inshock absorption tests. Approach methods attend totwo main trends: experimental systems as a tool forcrash helmet evaluation and numerical simulation as aguide for new helmet design. The experimentalmethod is a scientific approach and achieves results
1Department of Mechanical Engineering, Hsiuping University of Science
and Technology, Republic of China2Department of Mechanical and Automation Engineering, Da-Yeh
University, Republic of China
Corresponding author:
Tso-Liang Teng, Department of Mechanical Engineering, Hsiuping
University of Science and Technology, Taichung City 412-80, Taiwan,
Republic of China.
Email: [email protected]
Proc IMechE Part L:
J Materials: Design and Applications
227(1) 82–88
! IMechE 2012
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1464420712451806
pil.sagepub.com
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
closely resembling a real accident. Researchersfocus on the experimental performance of helmet cap-abilities.3–6 They evaluate drop tests based on stand-ard procedures and criteria and compare theperformance of different existing helmet designs. Forexample, McIntosh et al.7 evaluated helmet test resultsand proposed more effective performance criteria toameliorate head injury. Hui and Yu8 constructed amechanical model for bicycle helmet subjected toimpact due to free falling weight to predict thedynamic behavior of bicycle helmet under variousimpact conditions. However, the experimentalmethod for helmet analysis and design is complexand expensive. Numerical methods are of importancein a novel product development. Therefore, an alter-native method is to use numerical method for impacttest simulation and analysis. Mills and Gilchrist9 per-formed a finite element (FE) analysis of helmet obli-que impact test for assessing the linear and rotationalaccelerations of the headform and investigating theeffects of frictional parameters on the response.Blanco et al.10 investigated the crashworthiness ofthe cones liner with a FE model of a ski helmet proto-type. Impact simulation reproducing the standardsEN1077 were performed with the FE software LS-DYNA.
Rapid advances in computer technology haveenabled applied mathematicians, engineers, and scien-tists to make significant progress in solving previouslyintractable problems. The FE method contributesgreatly to helmet test modeling. As such, computersimulation is an economical and time-efficient alterna-tive to physical testing. The main advantage of thenumerical test model over the physical test model isthat it enables researchers to easily investigate theeffect of material and geometrical factors of thehelmet on head injuries. FE simulations are also indis-pensable to evaluate head injury criteria (HIC). Thisstudy performs FE analyses of helmet impact testsusing LS-DYNA software. To compensate for lackof oblique impact tests in helmet standards, this art-icle considers two oblique impact tests. It measureslinear accelerations, rotation accelerations, andimpact forces of the headform from helmet test simu-lations. To verify the accuracy of the proposedmethod, this study compares simulation results withexperimental data.4 The results of this method fit theexperimental results well, implying that the numericalmethod is a practical approach to helmet design prob-lems. Furthermore, the helmet test model proposed
here has potential for guiding the future developmentof helmet technologies.
FE modeling of helmet impact test
To confirm helmet protection capability, impact testsshould be performed based on helmet safety stand-ards. Generally, these tests involve a series of con-trolled impacts positioning the helmet on a testheadform. The helmeted headform is then droppedin guided falls onto specified test anvils. The impactsite and the impact energy must meet certain require-ments for the tests to be valid. This study builds theFE model of the helmet impact test. This helmetedheadform model consists of a headform and helmetmodels.
Headform FE model
The headform consists of three components: skin,skull, and lower head. This separation makes iteasier to adjust the mass and moment of headforminertia and also eases manufacture. The headformused for simulation testing was an approximately10mm thick plastic polyvinylchloride (PVC) scalpcovering internal aluminum casting. Total headformmass is 4.26 kg (Table 1). To design a headform with aclosed geometrical shape that fits the helmet suitably,researchers applied ISO/DIS 6220 descriptions formedium helmet size (J); this is also the suggestion ofthe CPSC’s Standard.2 The headform maintainer wasmade of aluminum casting. Figure 1 shows the head-form FE model. The PVC viscoelastic skin is modeledwith a set of 23,613 tetrahedral solid elements. Thealuminum lower head and core were modeled using12,347 and 12,549 solid elements, respectively. The10mm thick skin was modeled with the*MAT_GENERAL_VISCOELASTIC materialmodel in LS-DYNA. The 8mm thick skull and thelower head were modeled with the *MAT_ELASTICmaterial model, also using 12,347 and 12,549 solidelements, respectively. Contact between componentsof headform assembly were specified with*AUTOMATIC_SURFACE_TO_SUFACE.
Helmet FE model
Basic bicycle helmets are made from the outer shell,protective padding liner foam, comfort padding foam,and retention strap. The outer shell prevents
Table 1. Mass, inertia, and dimension of headform.
Headform Circumference (mm) Length (mm) Breath (mm) Mass (kg) Ixx (kg cm2) Iyy (kg cm2) Izz (kg cm2)
Experimental test 570 190 158 4.26 199 237 172
FE model 570 191 156 4.26 200 230 170
FE: finite element.
Teng et al. 83
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
penetration of the helmet and disperses energy fromthe impact. It is usually made from some family offiber-reinforced composites (glass, carbon or kevlar),thermoplastic like polycarbonates (PCs), or acrylo-nitrile butadiene styrene. This is tough material,but it compresses when it hits anything hard. A0.5� 0.05mm thick outer shell is modeled as an elas-toplastic material. Thus, the material card *MAT24 inLS-DYNA is assigned to the PC outer shell. Thecurve of effective stress versus effective strain behaviorreferred Wenjun’s study.11
Expanded polystyrene (EPS) foams, complexmaterials with excellent energy absorption, are usedextensively for bicycle and motorcycle helmets. Theyabsorb the most energy. Protective padding liner foaminside the outer shell is made of EPS foams and is inuse for all manufacturer models. The thickness ofliner foam helmets is always optimized to protectagainst the hardest impacts. EPS foam with a densityof 86.8 kg/m3 is used for FE helmet modeling. Theliner foam is considered a highly compressible, low-density foam and is modeled using material card*MAT57 in LS-DYNA. The EPS stress–strain curvereferred Van Den Bosch’s12 study.
Comfort padding foam is the soft foam and clothlayer that sits next to the head. It helps keep the headcomfortable and fits it snugly. In some types of hel-mets, this padding can even be removed for cleaning.The comfort padding liner can have a significantimpact on the results of the drop tests and should,therefore, also be modeled. Generally, the comfortpadding of the helmet is made of polyurethane (PU)foam.9 The characteristic of this foam is soft, lowdensity, open celled, and flexible foam. The stress–strain curve of PU referred Van Den Bosch’s12
study. This foam is approximately 3mm thick.
The *MAT 57 material model can be used to simulatePU foam.
The retention system, or chin strap connecting eachside of the outer shell, is very important. It insuresthat the helmet stays attached to the head whena crash occurs. For a polyethylene terephthalate(PET) chin strap, the tensile module and tensilestrength are 4GPa and 80MPa, respectively. The*MAT 24 material model simulates the chin strap.Table 2 presents all material properties.
This study considers a new helmet model designwith medium headform size (J) and a circumferenceof 575mm. The helmet’s geometrical parameters arebased on the Snell95 Standard.13 The helmet FEmodel consists of the shell, foam liner, and strap(Figure 2). The helmet model included 20 ellipticalholes as vents, occupying 20% of the helmet area.The plastic shell is modeled with a set of 4205 shellelements and a thickness of 0.5mm. The EPS foamliner is modeled using 58,280 tetrahedral solid elem-ents; average foam liner thickness is about 28mm atfront 90�, 23mm at right 70�, and 26mm at thecrown. The chin straps use 480 shell elements withthickness of 1mm, width of 15mm, and a 3mm gapbetwen strap and headform.
Figure 1. FE model headform.
FE: finite element.
Figure 2. FE model of helmet.
FE: finite element.
Table 2. Material properties of helmet and headform.
Type of material � (kg/m3) � E (MPa)
Aluminum 2700 0.33 70,000
PVC 1150 Taken from Van Den Bosch12
PC 1200 0.37 2400
EPS foam 86 22.4
PU foam 32 0.47
PET 1400 0.44 1000
PVC: polyvinylchloride; PC: polycarbonate; EPS: expanded polystyrene;
PU: polyurethane; and PET: polyethylene terephthalate.
�: density, �: Poisson’s ratio, and E: Young’s module.
84 Proc IMechE Part L: J Materials: Design and Applications 227(1)
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
Simulation of helmet impact test
Helmet impact test
This study performs the helmet impact test simulationbased on the experimental study.4 The impactorassembly consists of helmet and headform (calledthe helmeted headform model). The headform ismounted into the helmet and maintained by a strap.The helmeted headform model drops 8mm onto analuminum plate with initial tangential and normalvelocities of 3.6 and 4.5m/s, respectively. Impacttests are performed at two sites: left 70� and front70� of the headform. The rigid aluminum plate is0.5m long, 0.15m wide, and 10mm thick.
FE model of helmet impact test
Figure 3 shows the FE of helmet impact test. Thehelmeted headform model consists of the outershell of the helmet, the foam liner, strap, headform,and the road plate. The *AUTOMATIC_SURFACE_TO_SUFACE considered in LS-DYNA for mostdynamic simulation presents the contact between theheadform and foam liner, and likewise between thefoam liner and shell. The dynamic friction coefficientbetween headform and foam liner is 0.2; between thealuminum plate and outer shell it is 0.25.9 The softconstraint option is assigned TYPE 1 to simulate con-tact of headform and foam liner, and is similarly mod-eled between the shell and foam liner. The strap isconnected to the outer shell by welds modeling.
To validate a FE helmeted headform model, thisstudy draws some comparisons between simulationand experimental results. Mills and Gilchrist4 experi-ment shows response results such as liner acceleration,rotational acceleration, deformation of liner foam,and forces on the road. This study also suggestssome simulation response results which are useful toevaluate standard requirements like energy on head-form and HIC. HIC summarizes the relationshipbetween linear acceleration, impact duration, andthe onset of skull fractures. Equation (1) describesHIC.14
HIC ¼Max1
t2 � t1
Z t2
t1
aðtÞdt
� �2:5ðt2 � t1Þ ð1Þ
where a(t) is defined as resultant acceleration at theheadform center of gravity (CG); t1 and t2 (with con-ditions �t ¼ t2 � t1415ms) are two time instants
Time [ms]0 2 4 6 8 10 12 14
Lin
ear
acce
lera
tion
[G]
0
20
40
60
80
100
120
140
160
Exp. impactSim. left impactSim. front impact
Figure 4. Resultant linear acceleration on the impact test of front and left 70� sites.
Figure 3. FE model of helmeted head drop test.
FE: finite element.
Teng et al. 85
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
Time [ms]
0 2 4 6 8 10 12 14
Rot
atio
nal a
ccel
erat
ion
[kra
d/s2 ]
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Y-Rot-accel. frontZ-Rot-accel. front
Figure 6. Rotational acceleration on the impact test of front 70� site.
Time [ms]0 2 4 6 8 10 12 14
Rot
atio
nal a
ccel
erat
ion
[kra
d/s2
]
-2
0
2
4
6
Y-Rot-accel. Exp.Z-Rot-accel. Exp.Y-Rot-accel. Sim.Z-Rot-accel. Sim.
Figure 5. Rotational acceleration on the impact test of left 70� site.
Table 3. Comparison between simulation and experiment tests by oblique impacts.
Helmet
Impact
side
Maximum
linear
acceleration (G)
Maximum
z-rotation
acceleration
(k rad/s2)
Maximum
y-rotation
acceleration
(k rad/s2)
Maximum
Fn (kN)
Maximum
Ft (kN)
Maximum
liner
crush (mm) HIC value
Experiment Left 70� 129 4.0 4.9 6.0 1.4 15 720
FE model 138 4.1 4.5 6.3 1.25 16 841.7
Experiment Front 70� 117 <1.0 and >�1.0 5.5 0.5
FE model 118 0.4 1.0 6.3 1.0 17 715.9
HIC: head injury criteria; FE: finite element.
86 Proc IMechE Part L: J Materials: Design and Applications 227(1)
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
during the impact. The HIC predicts head injury incase of impact from flat objects and considers bothmain types of brain injuries: consciousness and linearskull fractures. A HIC value of 1000 is the safetythreshold for the human head.
Results and discussions
Figure 4 shows the resultant CG linear acceleration ofthe headform on the impact test at front and left 70�
sites. The shape and linear acceleration peak of theheadform relate significantly to helmet assessment cri-teria. As the figure indicates, the simulation lastsslightly longer than the experiment, and the resultantsimulation curves were consistent with those obtained
experimentally. Peak value of the simulation result islower than 10% disparity compared with experimentresults (Table 3). It is evident that impact durationtime yields a very similar response between simulationand experiment. The HIC value on the left 70� siteand front 70� site are 841.7 and 715.9, respectively.Accordingly, the HIC values do not exceed the safetythreshold (HIC values of 1000).
Figures 5 and 6, respectively, show rotationalacceleration at the headform’s CG on the left andfront 70� impact test sites. This study measures they- and z-axis components of rotational acceleration.The y- and z-axis rotational acceleration for impacton the 70� left site (Figure 5) have a maximum valueof 4.5 and 4.0 k rad/s2, respectively. Also, the
Time [ms]0 2 4 6 8 10 12 14
Forc
e on
the
road
sur
face
[kN
]
0
2
4
6
8
Fn Sim. left impactFt Sim. left impactFn Sim. front impactFt Sim. front impact
Figure 8. Force components Fn and Ft on the impact test of front and left 70� sites.
Time [ms]
0 2 4 6 8 10 12 14
Forc
e on
the
road
sur
face
[kN
]
0
2
4
6
8
Fn Sim. left impactFt Sim. left impactFn Exp. left impactFt Exp. left impact
Figure 7. Force components Fn and Ft on the impact test of left 70� site.
Teng et al. 87
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
simulation’s headform rotational accelerations have alower value than the experimental results (Table 3).The resultant linear headform acceleration of thenumerical model resembles experimental results well.However, its maximum value is 10% lower than theone measured in the experiment.
To study impact force on a road surface, this studymeasures normal force Fn and tangential force Ft onthe top plate. Figure 7 shows Fn and Ft versus timetrace on the impact test at the left 70� site. Fn and Ft
have a single peak value of 6.25 and 1.25 kN, respect-ively. The force on the road surface is quite close incomparisons between simulation and experimenttests. However, the tangential simulation forcereduces noise, with an unique maximum value ineach curve. Figure 8 compares forces between twoimpact sites; peak value of normal force and tangen-tial force are quite close and their curve shape is verysimilar.
The simulation yielded the best results in directfront impact. In direct left impact the results areacceptable. It is clear (Table 3) that the values oflinear acceleration, rotational acceleration compo-nents, and force components are fairly equal in thetwo impact sites. The FE model of a helmeted head-form corresponds to reality.
Conclusion
The helmet impact test model used LS-DYNA, and itsconstruction was based on experimental data. Thisstudy indicates that the development and validationof helmet impact test simulations correlate reasonablywith experimental data. It proposes a simulationmethod and demonstrates the feasibility of its appli-cation. The proposed model conforms to the simula-tion requirements of a helmet impact test.Numerically, measuring helmet safety in drop testsis useful. The study validates crashworthiness analysisof the helmeted headform model. The numericalmodel predicts helmet safety during impacts, reducesthe duration of research and design cycles, and cutsexperimental costs. The accuracy of the helmetedheadform model renders it a valuable approach tohelmet drop test simulations. The proposed modelcan examine the behavior of bicyclists and analyzeinjuries during fall accidents. Moreover, this simu-lated model serves as a valuable tool to assist thefuture development of safety helmet technologies.
Funding
This research received no specific grant from any fundingagency in the public, commercial, or not-for-profit sectors.
References
1. US Bicycle Helmet Safety Institute. Bicyclists and other
cyclists, traffic safety facts-2009 data, NHTSA’sNational Center for Statistics and Analysis, 2011.http://www.bhsi.org/stats.htm (2008, accessed 1 June
2012).2. Consumer Product Safety Commission. 16 CFR Part
1203:Safety standard for bicycle helmet. FederalRegister, vol. 63, No. 46, 1998. Washington, DC: The
US Consumer Product Safety Commission.3. Di Lando L, Sala G and Olivieri D. Deformation mech-
anisms and energy absorption of polystyrene foams for
protective helmets. Polym Test 2002; 21(2): 217–228.4. Mills NJ and Gilchrist A. Oblique impact testing of
bicycle helmets. Int J Impact Eng 2008; 35(9):
1075–1086.5. Liu DS, Chang CY, Fan CM, et al. Influence of envir-
onmental factors on energy absorption degradation of
polystyrene foam in protective helmets. Eng Fail Anal2003; 10(5): 581–591.
6. Caserta GD, Iannucci L and Galvanetto U. Shockabsorption performance of a motorbike helmet with
honeycomb reinforced liner. Compos Struct 2011;93(11): 2748–2759.
7. McIntosh AS, Kallieris D, Mattern R, et al. An evalu-
ation of pedal cycle helmet performance requirements.In: Proceedings of the 39th staff car crash conference,San Diego, CA, 8–10 November 1995, SAE technical
paper no. 952713, pp.111–120. Warrendale, PA: SAEInternational.
8. Hui SK and Yu TX. Modeling of the effectiveness ofbicycle helmet under impact. Int J Mech Sci 2002; 4:
1081–1100.9. Mills NJ and Gilchrist A. Finite-element analysis of
bicycle helmet oblique impacts. Int J Impact Eng
2008; 35: 1087–1101.10. Blanco DH, Cernicchi A and Galvanetto U. FE
Modeling of innovative helmet liners. In: Proceedings
of the 11th international LS-Dyna users conference,Dearborn, MI, USA, 6–8 June 2010, pp.9-1–9-12.
11. Wenjun HU, Zhang F, Tian C, et al. Dynamic stress–
strain respond and yield behavior of polycarbonate.Chin J Mater Res 2007; 21(4): 439–443.
12. Van Den Bosch HLA. Crash helmet testing and designspecifications. PhD Thesis, Technische Universiteit
Eindhoven, 2006.13. Snell Memorial Foundation. 1995 Standard for protect-
ive headgear for use with bicycles. 1995. North
Highlands, CA: Snell Memorial Foundation, http://www.smf.org/standards/b/b95std (1995, accessed 2012).
14. European Enhanced Vehicle-safety Committee.
Improved test methods to evaluate pedestrian protectionafforded by passenger cars. EEVC/WG 17 report, 1998,updated 2002. Washington, DC: European Enhanced
Vehicle-safety Committee.
88 Proc IMechE Part L: J Materials: Design and Applications 227(1)
at Bartin Universitesi on November 14, 2014pil.sagepub.comDownloaded from
