effectiveness of the football helmet assessed by …

EFFECTIVENESS OF THE FOOTBALL HELMET ASSESSED BY

FINITE ELEMENT MODELING AND IMPACT TESTING

Liying Zhang, Dwarampudi Ramesh, King H. Yang, Albert I. King

Biomedical Engineering Department, Wayne State University

ABSTRACT

Mini-sled impact tests using a Hybrid III head-neck complex with and without a helmet were

conducted to contrast resulting head accelerations. Because this experimental configuration includes

both the linear and rotational impacts to the head, it represents the real world football impact

condition better than drop tests traditionally used to evaluate the protective effect of a helmet. It was

found that wearing a helmet significantly attenuates head linear acceleration but not angular

acceleration. Additionally, a 3-D finite element model of a football helmet simulating all of its

essential components was developed and validated to investigate its effectiveness in energy

attenuation. Results from these modeling and experimental efforts have shown that the helmet was

less effective in attenuating impact energy due to an impact to the front-boss and side of the helmet.

The validated helmet model can be used in conjunction with a validated human head model to design

helmets intelligently by determining optimal liner thickness and stiffness for omnidirectional

protection of the head, once an acceptable injury criterion is established. In addition, the helmeted

head model can be used as an inexpensive adjunct for current test procedures and for the

establishment of new procedures.

KEYWORDS: Helmet model, helmet impact testing, model validation, linear acceleration, angular

acceleration.

MILD TRAUMATIC BRAIN INJURY (MTBI) accounts for a majority of all traumatic head injuries

treated in emergency rooms, outpatient departments, and physicians’ offices in the United States. The

National Health Interview Survey (NHIS) estimated that more than 300,000 MTBIs or concussions

occur in sports or recreational activities each year. American football alone is responsible for more

than 100,000 annual brain injuries even though headgear was in use (Wilberger, 1993; Cantu, 1998).

While many of these injuries are mild, some can be quite serious, with long-term consequences.

Brain trauma exacts a huge societal cost in terms of healthcare, disability and years of productive life

lost. An estimated 2% of the US population lives with disabilities resulting from a traumatic brain

injury. Mitigation of brain injury is a major consideration in head protection. A better understanding

of the causal mechanisms of brain injury will allow us to develop better methods of prevention and

use them in the design of a safer environment.

Numerical models have been used in the automotive industry to design devices and restraint

systems to mitigate head/brain injury. However, the same technique has not been widely applied in

the helmet industry. Only a handful of lumped-parameter and finite element (FE) models of

motorcycle or bicycle helmets have been reported in the literature (Gilchrist and Mills, 1993; Yettran

et al., 1994; Brands et al., 1996) while there is virtually no FE football helmet model available. If

such a model can be developed and validated against experimental data, safety engineers can apply it

to develop better protection headgear.

On a separate but related issue, the Severity Index (SI) and maximum acceleration used in various

safety standards to regulate sports helmets are calculated from resultant linear acceleration. Although

IRCOBI Conference – Lisbon (Portugal), September 2003 27

many researchers still dispute the relative contribution of linear or angular accelerations that may

cause brain injury, research on the effects of angular acceleration was pursued more vigorously than

those of linear acceleration in recent times (Gennarelli et, 1987; Margulies and Thibault, 1992). As a

result, there is a current notion among some researchers that angular acceleration is largely

responsible for brain injury. It has been noted in the literature that the number of brain injury cases in

American football has been significantly reduced after regulatory organizations mandated the use of

the football helmet (Mertz et al., 1996; Cantu and Mueller, 2003). If indeed angular acceleration is

the primary source of brain injury, then a helmeted head should experience a lower angular

acceleration when compared to a bare head subjected to the same impact condition. Thus, the

objectives of this study were: 1) to develop a FE helmet model validated against a variety of impact

situations; and 2) to conduct a series of mini-sled head impact experiments to understand how the

resulting linear and angular head accelerations are reduced by wearing a helmet. In the future the

validated helmet model can be used in conjunction with an anatomically detailed human head model,

the Wayne State University head injury model (WSUHIM) developed by Zhang et al. (2001b), to

design helmets intelligently by determining optimal liner thickness and stiffness for omnidirectional

protection of the head, once an acceptable injury criterion is established. In addition, the helmeted

head model can be used as an inexpensive adjunct for current test procedures and for the

establishment of new procedures.

METHODS

Three sets of experimental data were either obtained from or conducted at external sources and one

series of mini-sled experiments was conducted in house to develop a FE model of a large size

Riddell® VSR-4 football helmet and to assess the changes in head acceleration on a dummy head by

adding a helmet to a Hybrid III head and neck complex. The VSR-4 helmet was selected because it is

the most commonly used helmet in the National Football League (NFL). Data acquired from other

sources included those provided by Biokinetics and Associates, Ltd. (Ottawa, Canada) based on

ASTM helmet impact test protocol and those conducted at Denton ATD (Detroit, Michigan) based on

standard Hybrid III dummy head drop tests and neck calibration tests. The experiments conducted in

this study were mini-sled tests using a Hybrid III head and neck complex, with or without a VSR-4 or

a BIKE® helmet. Data obtained from the Hybrid III head calibration test and neck calibration test

were used to develop and validate a FE model of the Hybrid III head and neck complex. Additional

material properties needed for the Hybrid III head and neck model development were obtained from

compression tests using an Instron testing machine. For the VSR-4 helmet model, properties of the

comfort liner were derived from results of ASTM tests, using a reverse engineering approach. The

helmet model was then integrated with the Hybrid III head and neck model. Predictions made by the

combined model were compared with experimental results obtained from mini-sled tests. The

following sections describe different experimental and model development procedures employed in

this study.

EXPERIMENTAL METHODS:

ASTM Helmet Impact Tests: ASTM F717-89/95 is a standard test method to determine energy

attenuation characteristics of protective headgear for football recommended by the American Society

for Testing and Materials (ASTM) Committee on Standards (ASTM, 1989). Prior to the test, a

football helmet is mounted over a rigid ISO/DIS headform that can be oriented in any of the six

different positions and dropped at a specified velocity of 5.47 m/s onto a fixed metal anvil, covered by

a material with a durometer value of 38 shore “A”. Three impacts on each of the six locations,

namely crown, front, front boss, side, rear boss, and rear, are required. A triaxial accelerometer,

mounted at the center of gravity (C.G.) of the headform, is used to record the acceleration histories for

each impact location. The average maximum acceleration at each location shall not exceed 275 g.

These test results were used to tune the properties of the helmet comfort liner for the development of

the helmet model.

Hybrid III Head Drop Tests: According to the Code of Federal Regulations (Part 572), the

acceleration measured at the C.G. of a Hybrid III dummy head assembly needs to fit into a prescribed

corridor when it is dropped from a height of 376 mm onto a flat steel plate. The acceleration time


histories obtained from this type of tests were used to validate the Hybrid III head model to be

described later.

Hybrid III Neck Calibration Tests: Standard Hybrid III neck calibration impact tests were

conducted at a velocity of 7.06 m/s for the flexion test and 6.07 m/s for the extension test. During the

calibration procedure, head and neck rotations were measured and the histories of D-plane rotation

were calculated from those measurements. Here the D-plane rotation is defined as the rotation of the

base of the head relative to the pendulum. The main purpose of this set of test was to provide needed

data to validate the Hybrid III neck model, for both flexion and extension.

Mini-Sled Impact Tests: A series of mini-sled experiments was conducted to determine changes in

head accelerations when a football helmet was added to a Hybrid III head and neck complex. Bare

head impact tests were conducted first. Data obtained from these tests were used as a baseline for

comparison with helmeted head linear and angular acceleration data and to validate the Hybrid III

head-neck model. Two helmets, a Riddell® VSR-4 and a BIKE, were securely attached to the

Hybrid III head later and experiments repeated. These two helmets were selected because they were

commonly used by NFL players. Additionally, data obtained from the VSR-4 helmet test were used

to provide unique information to validate the integrated Hybrid III head and VSR-4 helmet model.

The impact tests were conducted by mounting

the head and neck complex of a Hybrid III dummy

onto a custom-designed fixture that was attached

to a pneumatically driven mini-sled (Figure 1).

This fixture was also designed to allow the head-

neck complex to be oriented in three major impact

directions (frontal, lateral, and front-boss or 45

degree oblique) to simulate typical head collisions

observed in football accidents. Impact velocities

of 5 and 7 m/s were selected for the bare head

impacts and velocities of 7 and 10 m/s were used

for the helmeted head impacts. The impact

surface was a 300-mm square foam block with a

thickness of 76 mm. It was attached firmly to a rigid barrier. The foam surface was sloped rearward

at a 30-degree angle to the vertical or Z-axis of the head coordinate system to avoid engagement of the

facial features of the Hybrid III head assembly during impact. For the impact surface, four different

types of foam material with varying densities were tested in order to simulate different impact

conditions. The selected foams were made of Expanded Polyproplyrene (EPP), Expanded

Polystyrene (EPS), GECET, and Polyurethane (Foamex). The Hybrid III head was instrumented with

a 3-2-2-2 nine-accelerometer array to measure all six components of head linear and angular

accelerations. Each impact configuration was repeated once and the averaged data were reported.

Fig. 1 - Schematic diagram of the mini-sled

test configuration.

MODEL DEVELOPMENT METHODS:

All simulations were performed using a nonlinear explicit finite element solver LS-DYNA version

960 (LSTC, Livermore, California). The following section describes relevant steps taken in the model

development process.

Football Helmet Model: Only the VSR-4 helmet model was developed in this study. The helmet

model consisted of an outer shell, energy absorbing liners, comfort liners, jaw pads, a facemask, and a

chinstrap. The geometry of the helmet was obtained by digitizing the physical helmet using a

MicroScribe® 3-D digitizer (Immersion Corporation, San Jose, California). Because the digitizer was

too small to reach some of the inner surfaces of the padding systems within the shell, components of

the padding and liner systems within the shell were detached from the shell and digitized separately.

The digitized profiles of the helmet shell, helmet liners and facemask were used to generate surface

data, using the Pro/ENGINEER® software (Parametric Technology Corporation, Needham, MA).

Based on these surface data, a finite element mesh of the helmet was developed using the

HyperMesh pre-processor, (Version 3.1, Altair Engineering, Inc., Troy, Michigan). The headform

used in the standard ASTM test was also modeled. The same 3-D digitizing method was used to

obtain the surface of the headform and the surface data were then meshed into a rigid body headform

model.


The entire helmet model was composed of about 20,400 nodes, and 25,600 elements, not including

the headform. The outer shell of the helmet was found to have a uniform thickness of 3.9 mm. The

protective liners for frontal, side, crown, and rear regions were meshed separately to incorporate

various thickness and material definitions. The thickness of the energy absorbing liners at the crown,

front, rear and side was measured and found to be 26, 25, 13, and 13 mm, respectively. A total of

8,716 brick elements arranged in three or more layers across the thickness were generated for all

energy absorbing liners. In the actual helmet, the outer surface of the liners was firmly attached to the

inner surface of the shell. Thus, no relative motion was allowed at this interface. Snap in jaw pads

were modeled with 854 solid elements and directly connected to the nodes of the outer shell. A

thickness of 20 mm was defined for the jaw pad based on a direct measurement of the prototype. The

comfort liners at the crown, rear and side regions were modeled as solid elements with a thickness

that ranged from 6 to 10 mm. These elements were covered with a layer of shell elements of varying

thickness (0.8 to 1.8 mm). The comfort liners were meshed by a total of 4,994 elements. The steel

facemask, or faceguard was meshed with 2,590 of shell elements which were assumed rigid. The total

mass of the FE VSR-4 model was 1.24 kg (excluding the face mask), matching that of the actual

helmet. Figure 2 shows the three-dimensional FE model of a Riddell® VSR-4 football helmet.

Energy Liner

-Crown Pad

-Frontal Pad

-Jaw Pad

Comfort Liner

-Crown Pad

-Side Pad

-Rear Pad

Impact Liner

-Rear, Side Pads

(a) (b)

Fig. 2 - Helmet model overview and a half model showing the interior energy absorbing liner and

comfort liner structure.

The helmet outer shell, made of molded Kara-lit® polycarbonate produced by GE, was assumed to

have a Young’s modulus of 1.7 GPa. The material definition for the liners was either based on data

obtained from manufacturers, acquired experimentally, or using a reverse engineering method.

Energy absorbing liners of the Riddell® VSR-4 helmet were made of vinyl nitrile and polyurethane

foam which are stiff but highly compressible elastic foam materials. An elastic foam material model

was defined for the energy absorbing liner material and the properties were based on the data provided

by its manufacturer. As for the jaw pad liner, the material properties were obtained from a

compression test. Comfort liners used at the crown, rear and brow are polyurethane and polyester

foams. The material characteristic of these types of foam was modeled as an inelastic crushable foam.

Simulations of the standard ASTM football helmet tests were carried out to obtain the material

properties of the comfort liner using reverse engineering methods. To simulate the impact test, the

integrated helmet and headform model was given an initial velocity of 5.47 m/s prior to impacting a

flat rubber surface, mounted on a rigid body representing fixed metal anvil. A penalty based contact

interface definition with a coefficient of friction of 0.3 was assigned between the headform and the

inner surface of the helmet. The helmet model was positioned in six orientations to conform to the

impact locations specified by the ASTM standard. By continuously tuning the elasticity of the

comfort liner material parameters, the predicted headform acceleration response matched

experimental measurements from ASTM tests. Table 1 lists the properties of major components used

in the helmet model.

Hybrid III Dummy Head and Neck Model: A Hybrid III head and neck model was needed because

the mini-sled impact tests used a Hybrid III dummy head and neck complex. Therefore, a Hybrid III

head and neck model was also developed. An accurate solid model of the head assembly was meshed

from CT scan data of the head of a Hybrid III 50th percentile male dummy. The head model was

constructed with 4,350 solid elements with a mass of 4.54 kg (Figure 3a). The moment of the inertia


was in accordance with published data (Kaleps and Whitestone, 1988). The deformable skin was

modeled as viscoelastic solid elements. The material properties of the vinyl skin were derived from

basic vinyl material tested in compression at rates of 0.1, 0.5, and 1.0 m/s. It was found that the vinyl

skin was not remarkably rate sensitive in the range tested. Consequently, average values were used to

define the property of vinyl skin.

Table 1 - Material properties assumed for the helmet model.

Substructure Material Model ρ (kg/mm3) E (MPa) ν

Shell Elastic 1.36E-06 1.7E+03 0.40

Energy Brow Elastic Foam 2.8E-07 8E-01 0.01

Liner Crown 1.2E-07 8E-02 0.01

Side 2.02E-07 5E-01 0.01

Rear 2.02E-07 5E-01 0.01

Comfort Crown Inelastic 3.0E-08 4.48E-02

Liner Side Crushable foam 7.0E-08 3.4E-01

Rear 7.0E-08 3.4E-01

Jaw Liner Elasto-plastic 1.2E-07 1.0E-01 0.01

Chin Strap Elasto-plastic 4.19E-06 2.4E+03 0.4

The geometry of the neck assembly was taken from the manufacture’s drawings. It is composed of

four butyl rubber segments sandwiched between five aluminum discs. The aluminum plates were

modeled as rigid solid elements. Each rubber segment has a partial depth horizontal slit on the

anterior side with a different depth at each level to yield the prescribed neck shape during bending.

These slits were explicitly modeled and surface-to-surface contact was defined for each slit. Elements

adjacent to the slit were covered with shell elements and were integrated using a full integration

scheme to avoid hourglassing due to excessive deformation. The rest of rubber intervertebral discs

was modeled as a viscoelastic solid, and a reduced integration scheme with hourglass treatment was

used.

The occipital condyle was represented by a revolute joint. The entire neck assembly was

constructed with 4,190 elements (Figure 3b) having a mass of 1.54 kg. Figure 3c shows the integrated

head-neck model. Butyl rubber material properties were obtained by compressing a neck rubber

segment at rates of 0.05, 0.1, 0.5, and 1.0 m/s to a maximum compression of 20% and a load of 8,500

N. The decay constant for the neck rubber was estimated to be 0.05 s-1. The properties used for the

neck model are listed in Table 2. The rotational stiffness of the occipital condyle joint was calculated

by dividing the moment by the angle of the occipital condyle measured during neck pendulum tests.

It was found that the joint stiffness determined using this method was lower than those used by

Seemann et al. (1986), Wismans and Hermans (1988), Deng (1989) and Yang and Le (1992).

(a) Head model (b) Neck model (c) Head and neck assembly

Fig. 3 - Hybrid III head and neck model.

Table 2 – Material properties assigned for the neck model. Structure Material Model Density (kg/mm3) E (MPa) υ

Neck Rubber Viscoelastic 1.05E-06 1.18E+01 0.48

Aluminum* Rigid 2.70E-06 6.89E+04 0.38

*: Although the aluminum plate was defined as rigid body, the Young’s modulus was still needed for contact

detection.


MODEL VALIDATION PROCEDURES:

Head-Neck Model Validation: The head model was validated against data obtained from head drop

tests while the head-neck model was validated against neck pendulum test procedures required by the

Code of Federal Regulations (Part 572). To simplify the head-neck model simulation, the measured

pendulum acceleration profile from the test was used as input to the model without the modeling of

the Hexcel block (honeycomb) used in the test. Neck D-plane rotation time histories predicted by

model from the simulation of the flexion and extension tests were compared to the acquired

experimental data. Additionally, the head-neck model was validated against data obtained from bare

head to foam block impact using a mini-sled. The foam material properties used in the model were

derived by dropping a weight onto various foams. Only the frontal impact simulations were

performed because impacts in the other directions were not conducted for unhelmeted head-neck

complex on the mini-sled (Figure 4a). Two different foam materials (EPP 4.2 and 3.0) were

simulated for the bare head impact. The sled with its payload was given an initial impact velocity of 7

m/s. Impact force and linear and rotational accelerations predicted by the model were compared with

experimental measurements.

Helmet Model Validation: The helmet model, developed using results from the ASTM tests, was

validated against data obtained from testing of a helmeted head-neck complex mounted on the mini-

sled (Figures 4b, c and d). The moving sled plate was represented by a small block with a mass equal

to that of actual sled plate. The nodes on the back surface of the foam block where the rigid barrier

was located were defined as rigid. To simulate helmeted head impacts, the Riddell helmet model,

identical to that used in the mini-sled tests, was integrated with the validated Hybrid III head-neck

model. A sliding interface with a coefficient of friction of 0.7 was used to represent the interaction

between the head and the helmet. Three impact directions, namely the frontal, lateral and front-boss,

were simulated. The same procedures used for the bare head impact tests were also used for this

series of helmeted head impact validations.

Foam

Block

Sled Plate

(a) (b) (c) (d)

Fig. 4 - Simulation of the test set up of a Hybrid III head-neck complex impacting a foam material: (a)

Unhelmeted frontal impact; (b) Helmeted frontal impact; (c) Helmeted front-boss impact; (d)

Helmeted lateral impact.

RESULTS

MODEL PERFORMANCE SIMULATING THE ASTM IMPACT TESTS:

Figure 5 shows a comparison of the resultant acceleration-time traces of the headform between

three experimental measurements and model predictions at crown, frontal, side, and rear locations.

The predicted peak resultant accelerations at the C.G. of the headform for crown, frontal, side and rear

impacts were 112, 162, 198, and 189 g, respectively. They were within 4% of the average

experimental head acceleration response for the corresponding impact conditions. In addition, the

corresponding impact durations from model simulation were quite comparable with those observed

experimentally. Overall, the comparison results indicated that the model was capable of capturing

head resultant translational acceleration-time traces, in the absence of any head rotation, for the

specified impact directions.

SIMULATION OF THE HYBRID III HEAD AND NECK CALIBRATION TESTS:


Figure 6a shows the resultant head acceleration predicted by the model compared to experimental

measurement during a Hybrid III head drop test. The HIC value calculated by the model was 776 in

comparison with 758 obtained from experimental data. Model results were within 5% of the

experimental values. The D-plane rotation histories predicted by the model are compared with

experimental data, as shown in Figures 6b and 6c. The model predictions agreed well with

experimental data and were well within the range of peak rotations specified for this test (error bar).

Also, the opening of the slits simulated during extension and the compressive contact of the slits

during flexion resembled the phenomena seen experimentally (Figure 7).

Accele

ration (

g)

A

ccele

ration (

g)

Fig. 5 - Comparison of head acceleration between model results and experimental data from ASTM

football helmet drop tests: crown, frontal, side and rear impact.

Head Drop Resultant Acceleration

0

40

80

120

160

200

240

280

320

0 1 2 3 4 5 6Time (ms)

Acce

lera

tio

n (

g)

Model

Test

Neck Pendulum Flexion

-60

-40

-20

0

20

40

60

80

0 0.05 0.1 0.15 0.2

Time (s)

D-P

lan

e R

ota

tio

n

(Deg

rees)

Experiment

Model

Neck Pendulum Extension

-40

-20

0

20

40

60

80

100

120

0 0.05 0.1 0.15 0.2

Time (s)

D-P

lan

e R

ota

tio

n

(Deg

rees)

Experiment

Model

(a) (b) (c)

Fig. 6 – Validation of the head and neck model against the standard Hybrid III dummy head drop test

and head-neck pendulum test: (a) Head resultant acceleration; (b) D-plane rotation in flexion; (c) D-

plane rotation in extension.

Fig. 7 – Neck flexion and extension test simulation.


SIMULATION OF THE MINI-SLED IMPACT TESTS:

Figure 8 shows model validation results for a bare head impact against EPP 3.0 pcf (pounds/cubic

foot) foam at 7 m/s impact velocity. The difference between the model and test data is summarized in

Table 3. Overall, the contact force magnitude predicted by the model matched well with experimental

data. The maximum difference in translational acceleration of the head between model prediction and

experimental data was 10.5%. The duration of the contact force and acceleration was slightly shorter

as predicted by the model. The maximum difference in peak rotational acceleration was 8.4%.

For helmeted head-neck mini-sled impact simulations, the model responses followed the

experimental data with varying degrees of correlation, both in magnitude and in trend. Figure 9

shows a comparison of impact force and head acceleration histories between model results and those

obtained experimentally for frontal impact against EPP 4.2 pcf foam at an impact velocity of 7 m/s.

The predicted impact force and translational acceleration were about 9.8% and 5.7% higher than those

measured experimentally. However, the rotational acceleration predicted by the model tended to be

lower than that of the test data with a difference of 7.4%. For both the front boss and side impact

simulations, resultant translational and rotational accelerations predicted by the model were within 7%

of the experimental results (Figures 10 and 11). The maximum difference in peak force between

model predictions and test data was no more than 10%. However, the model was not capable of

capturing the second peak in the force time histories seen experimentally (Figures 9, 10 and 11).

Frontal Impact Without Helmet

-6000

-4000

-2000

0

0 5 10 15 20

Time (ms)

Fo

rce

(N

)

Model

Test

Frontal Impact Without Helmet

-1600

-1200

-800

-400

0

400

0 10 20 30 4

Time (ms)

Acc

ele

rati

on

(m

/s2)

0

Model

Test

Frontal Impact without Helmet

-10000

-6000

-2000

2000

6000

0 5 10 15 20

Time (ms)

Accele

rati

on

(ra

d/s

2) Model

Test

Fig. 8 - Comparison of force and linear and rotational head acceleration between model results and

experimental data for a bare head impact at 7 m/s against EEP foam with a density of 3.0 pcf.

Table 3 – Comparison of the model predictions and experimental data for impact with and without

helmet in mini-sled test. Impact Condition Peak Response Experiment Simulation % diff.

Without Force (N) -5596 -5400 -3.5

helmet Frontal Linear Accel. (m/s2) -1130 -1250 +10.5

Rotational Accel. (rad/s2) -8666 -7940 -8.4

Force (N) -4745 -5210 +9.8

Frontal Linear Accel. . (m/s2) 911 963 +5.7

Rotational Accel. (rad/s2) 8010 7420 -7.4

With Force (N) -5037 -5450 +8.2

helmet Front-Boss Linear Accel. . (m/s2) 982 988 +0.6

Rotational Accel. (rad/s2) 8274 8390 +1.5

Force (N) -5059 -5538 +9.5

Side Linear Accel. . (m/s2) -1190 -1124 -5.6

Rotational Accel. (rad/s2) 10532 11300 +7.2


Frontal Impact with Helmet

-6000

-4000

-2000

0

0 5 10 15

Time (ms)

Fo

rce

(N

)

Model

Test


-200

200

600

1000

0 5 10 15 20

Time (ms)

Ac

ce

lera

tio

n (

m/s

2)

Model

Test1


-4000

-2000

0

2000

4000

6000

8000

0 5 10 15

Time (ms)

Acc

ele

rati

on

(ra

d/s

2)

Model

Test

Fig. 9 - Comparison of force, linear and rotational head acceleration between model responses and

experimental data for a helmeted frontal impact at 7 m/s velocity against EPP (4.2 pcf) foam.

Frontal Boss Impact with Helmet

-7000

-5000

-3000

-1000

1000

0 5 10 15 20

Time (ms)

Fo

rce

(N

)

Model

Test


0

200

400

600

800

1000

1200

0 2 4 6 8 10 12 14 16 18 20Time (ms)

Ac

ce

lera

tio

n (

m/s

2)

Test

Model


0

2000

4000

6000

8000

10000

0 5 10 15 20

Time (ms)

Ac

ce

lera

tio

n(r

ad

/s/s

)

Model

Test


experimental data for a helmeted front-boss impact at 7 m/s velocity against EPP (4.2 pcf) foam.

Side Impact with Helmet

-7000

-5000

-3000

-1000

1000

0 5 10 15

Time (ms)

Fo

rce

(N

)

20

Model

Test


-1200

-800

-400

0

400

0 5 10 15 20 25

Time (ms)

Acc

ele

rati

on

(m

/s2)

Model

Test


0

2000

4000

6000

8000

10000

12000

0 5 10 15 20 25 30

Time (ms)

Acce

lera

tion

(ra

d/s

2)

Model

Test


experimental data for a helmeted side impact at 7 m/s velocity against EPP (4.2 pcf) foam.

MINI-SLED EXPERIMENTAL RESULTS:

Table 4 summarizes the translational and rotational accelerations and HIC obtained from helmeted

mini-sled experiments at a speed of 7 m/s. Each impact test was repeated once and the averaged data

were reported. Impacts from the side and front-boss directions produced higher translational head

accelerations compared to those seen in frontal impact at the same impact velocity. Impacts to the

frontal and the frontal boss regions produced nearly the same amount of rotational acceleration

whereas impact to the side of the helmet resulted in the lowest rotational acceleration. The HIC value

was the highest for front-boss impacts followed by side impacts.

Table 4 - Head resultant acceleration and HIC values for different impact locations Riddell Helmet

7 m/s

Frontal

Impact

Front-Boss

Impact

Side

Impact

Mean T. Accel. (g) 89 97 103

Mean R. Accel. (rad/s2) 7790 7772 6504

HIC15 382 436 398

Figure 12 shows a comparison of linear and rotational head accelerations with and without either

the BIKE® or Riddell® helmet for frontal impacts at 7 m/s into a variety of foams. Results clearly

showed that the translational head acceleration decreased by as much as 29.5% for the Riddell®

helmet and 22.1% for the BIKE® helmet, in comparison with unhelmeted impact data. On the other


hand, changes in the peak rotational acceleration due to the use of both helmets were small and

inconsistent with the average reduction of only 5.6%.

Frontal Impact , 7 m/s

0

20

40

60

80

100

120

140

160

180E

PP

1.5

EP

P 2

.0

EP

P 3

.0

EP

P 4

.2

EP

S 1

.5

EP

S 3

.0

EP

S 4

.0

GE

CE

T 4

.0

FO

AM

EX

4.0

Tra

nsla

tio

nal A

ccele

rati

on

(g

)

Without Helmet

With BIKE Helmet

With Riddell Helmet

Frontal Impact , 7 m/s

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

EP

P 1

.5

EP

P 2

.0

EP

P 3

.0

EP

P 4

.2

EP

S 1

.5

EP

S 3

.0

EP

S 4

.0

GE

CE

T

4.0

FO

AM

EX

4.0

Ro

tati

on

al

Ac

ele

rati

on

(ra

d/s

2) Without Helmet

With BIKE Helmet

With Riddell Helmet

Fig. 12 - Comparison of acceleration measured at the head C.G. without a helmet and with a BIKE

and Riddell helmet for impacts conducted at 7 m/s.

DISCUSSION

A 3D FE model of a VSR-4 football helmet has been developed and validated against mini-sled

impact tests during which both linear and rotational acceleration were involved. In addition, a Hybrid

III head model and a Hybrid III head-neck model was developed and validated against data obtained

from the standard head drop test and head-neck pendulum test, respectively. Overall, the computer

models developed were capable of duplicating the dynamic response of the system seen

experimentally. However, the correlation of model to the experimental data varied with different

impact conditions.

For bare head frontal mini-sled impact simulations, slightly larger discrepancies in rotational

acceleration, especially during the acceleration period, were observed between the model and

experimental results. This was probably related to our inability to identify the precise point of contact

between the forehead and the foam surface during the actual test. A small variation in the point of

contact with respect to the C.G. of the head could significantly affect the magnitude of the computed

rotational acceleration. Another source of discrepancy is the existence of the small clearance between

the sled plate and the sled rail that could induce unexpected vertical motions during impact as

observed in the high-speed video. Improved stability of the mini-sled should reduce the variations.

As for the validation of helmeted head impacts, rotational acceleration in frontal impact was

generally under-predicted even though the predicted contact force and linear acceleration from model

were higher than those seen experimentally. One reason for the discrepancy could be due to the

sliding definition and the assigned friction coefficient between the head and helmet in the model. Part

of the rotational energy could be dissipated due to relative sliding occurring between the head and

helmet in the model and it might not be the case in the actual test. Notice that experimentally

obtained force response generally exhibited a second peak, while the model was not capable of

predicting this trend. High-speed video footage taken during the mini-sled tests was reviewed to

decide the potential source of discrepancy. It was found that there was a slight upward motion of the

sled due to a minute clearance between the sled and rail, especially for impacts involving stiffer

materials (e.g. EPP 4.2). This could induce a second contact between the helmet and foam block

experimentally which could not be simulated by the model. In addition to a better characterization of

foam material properties, fine-tuning of the helmet model may be needed to further improve model

predictions.

In terms of the liner material used in the current VSR-4 helmet, the model indicated that

compression of the liner was 90% or higher during side and rear impacts. This may imply that a

different foam is needed to maximize the energy attenuation capacity in these areas. In addition, more

deformation or shell bending was predicted by the model in a side and rear impact compared to other

impact sites. These observations indicate that there is room for improving the current helmet design

in terms of structural and material optimizations.



Three ASTM tests were conducted for each of the six impact locations while linear accelerations

were measured at the C.G. of the ISO/DIS headform. The resultant linear head acceleration ranged

from 116 to 210 g’s with the highest response occurring during a lateral impact. According to the

performance criterion currently specified in the ASTM standard, the helmet passed the tests.

However, these data clearly indicate that the current helmet design does not offer equal protection to

the head for impacts coming from different directions. Analysis of data obtained from the mini-sled

impact tests, a different experimental approach conducted in this study, also demonstrated that the

head experienced the highest response in front-boss (oblique 45 degree) impacts in comparison with

other directions in terms of peak linear acceleration or HIC values. Therefore, if brain injury could be

mitigated by reducing translational head acceleration, both the drop test and mini-sled test results

suggest that the head received less protection from the current helmet if impacts were delivered to the

front-boss region and to the side of the helmet.

In all existing helmet standards, the assessment of helmet energy attenuation performance utilizes a

single criterion regardless of impact location. An analysis of MTBI data obtained from the NFL

database revealed that 40% (10 out of 25) concussion cases were due to an impact located in the 45-

90 degree quadrant of the helmet, 36% were the result of impacts in the 0-45 degree quadrant and

18% in the 90-150 degree quadrant (Zhang et al., 2003). In a previous study using a FE human head

model, it was shown that the human head had a decreased tolerance to lateral impact in comparison

with an impact from the frontal direction (Zhang et al., 2001b). Animal models of concussion and

diffuse axonal injury studies (Hodgson et al., 1983; Gennarelli et al., 1987) also showed that animals

sustained a more severe form of brain injury from lateral impacts than from other impact directions.

Low tolerance of the head to lateral impact in comparison with frontal impact was also observed in

cadaver tests conducted by Tarriere (1985). Data obtained from field accident, animal and cadaver

experimental studies, mathematical modeling, along with that obtained from this study all suggest that

directional sensitivity of the head/brain need to be taken into account in helmet design.

No clear trend can be established when comparing rotational accelerations for bare and helmeted

heads during mini-sled testing of the Hybrid III head and neck complex. For the BIKE® helmet, four

of the nine foams tested resulted in reduced angular accelerations while seven of the nine foams tested

resulted in reduced angular accelerations when a Riddell® helmet was used. On the other hand, linear

accelerations were significantly reduced for both helmets, under all test conditions. The average

reduction in rotational acceleration was only 5.6% compared to an average reduction of 26% in linear

acceleration when a helmet is worn. Thus, it can be concluded that both helmets tested are effective

in reducing the head translational acceleration due to the padding but had little or no effect on

rotational acceleration. This phenomenon makes physical sense. Padding in a helmet increases the

contact area with the head and thus reduces the peak force and linear acceleration. However, the

magnitude of an impact-induced angular acceleration is a function of the location of impact with

respect to the C.G. of the head-neck system, the rotational moment of inertia, and magnitude of the

contact force. Because these parameters are very similar for both the bare and helmeted head test

configurations, it is understandable that the two helmets studied are not effective in reducing

rotational acceleration.

CONCLUSIONS

1. A three-dimensional FE model of a football helmet has been developed. To the best of our

knowledge, this is the first FE model that simulates all the essential components of a football

helmet. Additionally, a Hybrid III head-neck complex was also developed and validated.

2. The helmet-head-neck model was validated against experimental data obtained from mini-sled

tests subjected to both linear and angular motions.

3. The structural and material characteristics of the helmet tested were not optimized to protect

head against impacts from front-boss and side of helmet.

4. Under a limited number of impact conditions, the two football helmets tested significantly

attenuated the head translational acceleration but were less effective in reducing rotational

acceleration.

5. Current helmet designs need to be improved to offer 360-degree protection.


ACKNOWLEDGMENTS

This research was supported in part by a grant from National Football League Charities for the

study of Mild Traumatic Brain Injury. The opinions expressed here are those of the authors and do

not necessarily reflect those of the National Football League.

REFERENCES

American Society for Testing and Materials Standard, ASTM Standard F717, 1989.

Brands, D. Thunissen, J. and Wismans, J., Finite element modeling of a helmet, Proc. International

Conf. on New Frontiers in Biomechanical Engineering, 1997, pp. 289-294.

Cantu, R.C., Athletic head injury. Clinics in Sports Medicine, 17(1), 1998, pp. 31-41,

Cantu, R.C. and Muller, F.O. Brain injury-related fatalities in American Football,

Deng, Y.C., Anthropomorphic dummy neck modeling and injury considerations, Accid. Anal. & Prev.

21, 1989, pp. 85-100.

Gilchrist, A. and Mills N.J., Deformation analysis for motorcycle helmets, Proc. IRCOBI Conference,

1993, pp. 269-281.

Gennarelli, T.A., Thibault, L.E., Tomei, G., Wiser, R., Graham, D., and Adams, J., Directional

dependence of axonal brain injury due to centroidal and non-centroidal acceleration, Proc. 31st

Stapp Car Crash Conference, SAE Paper No. 872197, 1987.

Hodgson, V.R., Thomas, L.M., and Khalil, T.B., The role of impact location in reversible cerebral

concussion, Proc. 27th Stapp Car Crash Conference, SAE Paper No. 831618, 1983.

Kaleps, I. and Whitestone, J., Hybrid III geometrical and inertial Properties, SAE Paper No. 880638,

1988.

Margulies, S.S. and Thibault, L.E., A proposed tolerance criterion for diffuse axonal injury in man, J.

Biomechanics, 25, 1992, pp. 917-923.

Mertz, H.J., Prasad, P. Nusholtz, G., Head injury risk assessment for forehead impacts; SAE Paper

No. 960099, 1996

Thurman, D.J, Brance, C.M., and Sniezek, J.E., The epidemiology of sports-related traumatic brain

injuries in the United States: Recent developments, J. Head Trauma Rehabil. Apr., 13 (2), 1998,

pp. 1-8.

Seemann, M.R. Muzzy IIII, W. H., and Lustick, L.S., Comparison of human and Hybrid III head and

neck dynamic response, Proc. 30th Stapp Car Crash Conference, SAE paper No. 861892, 1986.

Tarriere, C., Depressurized cadaver head impacts—some findings, Biomechanics Symposium on

Head Injury Prevention Past and Present Research, Wayne State University, 1985, pp. 93-144.

Wilberger, J.E, Minor head injuries in American football, Sports Med. 15: 5, 1993.

Wismans, J. and Herman, J.H.A., MADYMO 3D simulation of Hybrid III dummy sled tests, SAE

Paper No. 880645, 1988.

Yang, K.H., and Le, J., Finite element modeling of Hybrid III head-neck complex, SAE Paper No.

922526, 1992.

Yettrarn A.L., Godfrey, N.P.M., and Chinn, B.P., Materials for motorcycle crash helmets-a finite

element parametric study, Plastics, Rubber and Composites Processing 22, 1994, pp. 215-221.

Zhang, L., Yang, K.H., and King, A.I, Comparison of brain responses between frontal and lateral

impacts by finite element modeling, J. Neurotrauma 18 (1), 2001a, pp. 21-30.

Zhang, L., Yang, K.H., Dwarampudi, R., Omori, K., Li, T., Chang, K., Hardy, W.N., Khalil, T.B.,

King, A.I., Recent advances in brain injury research: A new human head model development and

validation. Stapp Car Crash Journal, 45, 2001b, pp. 369-393.

Zhang, L. Yang, K.H., King, A.I, and Viano, D.C., A new biomechanical predicator for mild

traumatic brain injury – A preliminary finding, Proc. ASME Summer Bioengineering Conference,

2003, pp.137-138.

effectiveness of the football helmet assessed by …

Documents