University of Southern California Dissertations and Theses

The effect of helmet liner density upon acceleration and local contact forces during bicycle helmet impacts

(USC Thesis Other)

The effect of helmet liner density upon acceleration and local contact forces during bicycle helmet impacts

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE EFFECT OF HELMET LINER DENSITY
U PO N ACCELERATION AND LOCAL
CONTACT FORCES DURING BICYCLE
HELMET IMPACTS
by
Terrance Alan Smith
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Exercise Science)
December 1997
© 1997 Terrance Alan Smith
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY O F SO U TH ER N C A L IFO R N IA
THE GRA D U A TE SCHOOL
LN IV Ç R SITY PARK
LOS AN GELES. C A L IFO R N IA « 0 0 '
This dissertation, written by
Terrance A. Smith
under the direction of his Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF P H IlO SO P H 'f
je s n or ,jr2iua:e ^tuaies
D a k :
DISSERTATION COMMITTEE
/ T t / i / ^
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
This thesis is dedicated to my wife Leslie and to my daughter Emihe for
their love, support and understanding.
1 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
This thesis began as an idea more than seven years ago and it could not have
come to fruition without the assistance of many people. Firstly, to David
Thom, a good friend and colleague who was instrumental in assisting me
with my move to Los Angeles and was always available for freeform
discussions on the issues of head protection and helmet design. Secondly, to
Hugh H. H urt, Jr., a true scholar whose insight, experience, support and
counsel has proved invaluable throughout my career at USC. Thirdly, to Dr.
Jill McNitt-Gray, for her understanding that a doctoral career is a journey of
discovery with good healthy doses of life added in to make it interesting.
I would also like to express my gratitude to Dr. Rand Wilcox for his
assistance with the statistical analysis of my data and to Barry Munkasy and
James Eagle for the advice and ideas they provided during the development
of my project.
As with all projects of this magnitude, this project could not have been
completed without the generous support of Mr. Bill Willen, Honda Motor
Company, The Protective Headgear Manufacturer’s Association (PHMA),
Bell Sports, Inc. and the Head Protection Research Laboratory.
m
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
Dedication........................................................................................................... ii
Acknowledgements........................................................................................... iii
Abstract............................................................................................................ xiii
Chapter 1: Introduction...................................................................................... 1
Problem Definition and Rationale........................................................9
Objectives of this study........................................................................11
Statement of hypothesis........................................................................11
Delimitations........................................................................................ 11
Limitations and Assumptions.............................................................. 12
Chapter 2: Review of Literature.......................................................................13
Functional Anatomy.................................................................................. 13
The Scalp............................................................................................... 13
The Skull............................................................................................... 14
The Brain Coverings.............................................................................16
The Subdural Space............................................................................... 17
The Brain.............................................................................................. 17
Mechanisms of Head Injury........................................................................18
Head Injury Due To Contact Forces...................................................19
Head Injury Due To Inertial Forces....................................................22
The Role of The Bicycle Helmet.........................................................24
Bicycle Helmet Standards.................................................................... 28
Current Issues in Head Protection Research...................................... 29
Pressure Measurement Devices............................................................ 34
Chapter 3: Methodology...................................................................................37
Experimental Instrumentation............................................................ 37
Development of Trigger and Signal Conditioning Circuitry............ 37
Transducer Response Evaluation.........................................................39
Transducer Linearity Evaluation.........................................................40
Transducer Hysteresis Evaluation.......................................................41
Transducer Repeatability Evaluation..................................................42
Transducer Response Based on Sensor Area Coverage......................42
Design of 64 Channel Transducer.......................................................43
Selection of Force Measurement Range.............................................. 43
Development of Printed Circuit Board For Signal Conditioning ....45
Software Development for Data Collection....................................... 46
Selection of Helmet Test Samples........................................................46
IV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Development of Helmet Test Matrix................................................. 47
Transducer Preparation and Calibration............................................ 50
Preparation of the Test Headform......................................................52
Impact Test Procedures........................................................................ 55
Post Test Procedures............................................................................ 56
Data Reduction.................................................................................... 57
Accelerometer Data Reduction...........................................................57
Sensor Force Data Reduction.............................................................. 58
Statistical Analysis................................................................................ 60
Chapter 4: Results.............................................................................................62
Transducer Response Evaluation......................................................... 62
Phase Response Characteristics............................................................ 63
Transducer Linearity Evaluation.........................................................64
Transducer Hysteresis Evaluation.......................................................66
Transducer Repeatability Analysis......................................................67
Sensor Output During Partial Coverage............................................ 68
Sensor Cahbration For Impact Testing.............................................. 69
Helmet Test Samples........................................................................... 72
Impact Test Results.............................................................................. 73
Peak Headform Force.......................................................................... 82
Impact D uration.................................................................................. 84
Head Injury Criteria (HIC) Results.....................................................87
Peak Sensor Force Change Results......................................................91
Correlation Analysis: Peak Headform Accleration and
Peak Sensor Force Change................................................................... 96
UniForce Transducer Distribution Profiles..................................... 101
Radial Analysis....................................................................................120
Chapter 5: Discussion.....................................................................................129
Development of the UniForce Transducer.......................................130
Sensor Response Characteristics........................................................ 130
Limitations of the Measurement System...........................................131
Linear Headform Accelerations......................................................... 133
Impact D uration.................................................................................137
Head Injury Criteria...........................................................................139
UniForce Transducer Measurements................................................141
Correlation of Peak Sensor Force Change With
Peak Headform Acceleration............................................................. 145
Distribution Analysis..........................................................................146
Current Limitations In Head Injury Research................................. 150
Chapter 6: Summary, Conclusions and Recommendations......................... 156
Conclusions.........................................................................................158
Recommendations..............................................................................160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bibliography...................................................................................................161
Appendix A: Headform Acceleration Data Sorted By Group...................168
Appendix B: Headform Acceleration Data Sorted By Drop Height............177
Appendix C: Load Distribution Profiles At Time of Peak Sensor Force
Left Front Impact Location.....................................................186
Appendix D: Load Distribution Profiles At Time of Peak Sensor Force
Right Side Impact Location .................................................... 203
Appendix E: Radial Distribution Analysis At Time of Peak Sensor Force
Left Front Impact Location.................................................... 220
Appendix F: Radial Distribution Analysis At Time of Peak Sensor Force
Right Side Impact Location..................................................... 237
V I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Number Poge
Figure 1. Trigger Logic For UniForce Sensor................................................ 38
Figure 2. UniForce Electrical Circuit..............................................................39
Figure 3. Transducer Design............................................................................ 44
Figure 4. Left Front Impact Location..............................................................48
Figure 5. Right Side Impact Location..............................................................49
Figure 6. Sensor Mounted At Left Front Impact Location............................ 53
Figure 7. Evaluation of Digital Circuit Performance..................................... 63
Figure 8. Phase Response Comparison Between Accelerometer
and UniForce Sensor........................................................................... 64
Figure 9. Linearized Response From UniForce Transducer..........................65
Figure 10. Second Order Polynomial Curve Fit of Cahbration Data........... 65
Figure 11. Boxplot of Repeatabihty D ata.......................................................68
Figure 12: UniForce Transducer Output Relative To Percentage Coverage
of Sensing Area.....................................................................................69
Figure 13: Comparison of Pre-test and Post-test Cahbration Values
For Sensor 30........................................................................................71
Figure 14: Group 1 Headform Acceleration Across Drop Heights,
Left Front Impact Location, Hemispherical Anvil............................73
Figure 15: Group 2 Headform Acceleration Across Drop Heights,
Left Front Impact Location, Hemispherical Anvil............................74
Figure 16: Group 3 Headform Acceleration Across Drop Heights,
Left Front Impact Location, Hemispherical Anvil............................74
Figure 17: Group 4 Headform Acceleration Across Drop Heights,
Left Front Impact Location, Hemispherical Anvil............................75
Figure 18: Headform Acceleration Data For Front Left Impact Location,
Hemispherical Anvil, 1.5 m Drop Height.......................................... 76
Figure 19: Headform Acceleration Data For Front Left Impact Location,
Flat Anvil, 1.5 m Drop Height............................................................77
Figure 20: Headform Acceleration Data For Right Side Impact Location,
Hemispherical Anvil, 1.5 m Drop Height.......................................... 77
Figure 21: Headform Acceleration Data For Right Side Impact Location,
Flat Anvil, 1.5 m Drop Height............................................................78
Figure 22: Scatterplot of Left Front Impact Location,
Hemispherical Anvil Impact Test Data.............................................. 98
Figure 23: Scatterplot of Left Front Impact Location,
Flat Anvil Impact Test Data................................................................98
Figure 24: Scatterplot of Right Side Impact Location,
Hemispherical Anvil Impact Test Data.............................................. 99
vu
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 25: Scatterplot of Right Side Impact Location,
Flat Anvil Impact Test Data...............................................................100
Figure 26: Scatterplot of Group 4 Right Side Impact Location,
Hemispherical Anvil Impact Test Data.............................................101
Figure 27: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at Contact...........................102
Figure 28: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 2.2 ms after Contact...... 103
Figure 29: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 5.2 ms after contact....... 103
Figure 30: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 7.8 ms after contact........ 104
Figure 31: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 10.4 ms after contact...... 104
Figure 32: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 13 ms after contact..........105
Figure 33: Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop Height Load Distribution at 15.6 ms after contact...... 105
Figure 34: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at contact............................106
Figure 35: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at 2.6ms after contact..........106
Figure 36: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at 5.2 ms after contact........ 107
Figure 37: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at 7.8 ms after contact........ 107
Figure 38: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at 10.4 ms after contact...... 108
Figure 39: Group 4 Right Side Impact Location, Flat Anvil,
2.0 m Drop Height Load Distribution at 13 ms after contact......... 108
Figure 40: Group 3 Load Distribution Profile for a 50 cm Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................109
Figure 41: Group 3 Load Distribution Profile for a 1.0 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................110
Figure 42: Group 3 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................110
Figure 43: Group 3 Load Distribution Profile for a 2.0 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................I l l
Figure 44: Group 1 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................112
Figure 45: Group 2 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................112
Figure 46: Group 3 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................113
vm
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 47: Group 4 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Hemispherical Anvil..........................113
Figure 48: Group 3 Load Distribution Profile for a 50 cm Drop Height,
Left Front Impact Location, Fiat Anvil............................................115
Figure 49: Group 3 Load Distribution Profile for a 1.0 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................115
Figure 50: Group 3 Load Distribution for a 1.5 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................116
Figure 51: Group 3 Load Distribution Profile for a 2.0 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................116
Figure 52: Group 1 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................118
Figure 53: Group 2 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Fiat A nvil............................................118
Figure 54: Group 3 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................119
Figure 55: Group 4 Load Distribution Profile for a 1.5 m Drop Height,
Left Front Impact Location, Fiat Anvil............................................119
Figure 56: Radial Analysis Quadrant Distribution For
Left Front Impact Location..............................................................121
Figure 57: Radial Analysis Quadrant Distribution For
Right Side Impact Location ..............................................................122
Figure 58: Radial Analysis Quadrant Distribution For
Group 2 Left Front Impact Location, Hemispherical Anvil,
50 cm Drop Height.............................................................................123
Figure 59: Radial Analysis Quadrant Distribution For
Group 1 Left Front Impact Location, Hemispherical Anvil,
1.5 m Drop H eight.............................................................................124
Figure 60: Radial Analysis Quadrant Distribution For
Group 1 Left Front Impact Location, Fiat Anvil,
1.0 m Drop H eight.............................................................................125
Figure 61: Radial Analysis Quadrant Distribution For
Group 4 Right Side Impact Location, Hemispherical Anvil,
2.0 m Drop H eight.............................................................................126
Figure 62: Radial Analysis Quadrant Distribution For
Group 2 Right Side Impact Location, Fiat Anvil,
1.5 m Drop H eight.............................................................................127
IX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Number Page
Table 1. Head Injury Mechanisms and Clinical Outcome.............................19
Table 2: Bicycle Helmet Standards in North America...................................29
Table 3. Hysteresis Values As Reported By Transducer Manufacturer........66
Table 4: Hysteresis Values Obtained Using Experimental
Instrumentation........................... 67
Table 5: Trimmed Mean Response Ratios For Repeatabihty Analysis.........67
Table 6: Trimmed Mean Values of MSE For Each Transducer
Used During Testing............................................................................70
Table 7: Summary of Helmet Mass Analysis.................................................. 72
Table 8: 20% Trimmed Mean Peak Headform Acceleration Across Groups
and Drop Heights - Left Front Impact Location, Flat Anvil............78
Table 9: 20% Trimmed Mean Peak Headform Acceleration Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil............................................................................78
Table 10: 20 % Trimmed Mean Peak Headform Acceleration Across Groups
and Drop Heights - Right Side Impact Location, Flat A nvil............79
Table 11: 20% Trimmed Mean Peak Headform Acceleration Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil............................................................................79
Table 12: Results of Linear Contrast Analysis For Peak Headform
Acceleration - Left Front Impact Location, Flat Anvil......................81
Table 13: Results of Linear Contrast Analysis For Peak Headform
Acceleration - Left Front Impact Location, Hemispherical Anvil ....81
Table 14: Results of Linear Contrast Analysis For Peak Headform
Acceleration - Right Side Impact Location, Flat Anvil......................81
Table 15: Results of Linear Contrast Analysis For Peak Headform
Acceleration - Right Side Impact Location, Hemispherical Anvil ....82
Table 16: 20% Trimmed Mean Peak Headform Force Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil............................................................................82
Table 17: 20% Trimmed Mean Peak Headform Force Across Groups
and Drop Heights - Left Front Impact Location, Flat Anvil............83
Table 18: 20% Trimmed Mean Peak Headform Force Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil............................................................................83
Table 19: 20% Trimmed Mean Peak Headform Force Across Groups
and Drop Heights - Right Side Impact Location, Flat A nvil............ 83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 20: 20% Trimmed Mean Impact Duration Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil.............................................................................85
Table 21: 20% Trimmed Mean Impact Duration Across Groups
and Drop Heights - Left Front Impact...Location, Flat Anvil.......... 85
Table 22: 20% Trimmed Mean Impact Duration Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil.............................................................................85
Table 23: 20% Trimmed Mean Impact Duration Across Groups
and Drop Heights - Right Side Impact...Location, Flat A nvil.......... 85
Table 24: Linear Contrast Analysis of Impact Duration Between Groups
and Across Drop Heights, Left Front Impact Location,
Hemispherical Anvil.............................................................................86
Table 25: Linear Contrast Analysis of Impact Duration Between Groups
and Across Drop Heights, Left Front Impact Location,
Flat Anvil.............................................................................................. 87
Table 26: Linear Contrast Analysis of Impact Duration Between Groups
and Across Drop Heights, Right Side Impact Location,
Hemispherical Anvil.............................................................................87
Table 27: Linear Contrast Analysis of Impact Duration Between Groups
and Across Drop Heights, Right Side Impact Location,
Flat Anvil.............................................................................................. 87
Table 28: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across
Groups and Drop Heights - Left Front Impact Location,
Hemispherical Anvil.............................................................................88
Table 29: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across
Groups and Drop Heights - Left Front Impact Location,
Flat Anvil.............................................................................................. 88
Table 30: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across
Groups and Drop Heights - Right Side Impact Location,
Hemispherical Anvil.............................................................................88
Table 31: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across
Groups and Drop Heights - Right Side Impact Location,
Flat Anvil.............................................................................................. 89
Table 32: Linear Contrast Analysis of Head Injury Criteria Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil.............................................................................90
Table 33: Linear Contrast Analysis of Head Injury Criteria Across Groups
and Drop Heights - Left Front...Impact Location, Flat Anvil..........90
Table 34: Linear Contrast Analysis of Head Injury Criteria Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil.............................................................................90
Table 35: Linear Contrast Analysis of Head Injury Criteria Across Groups
and Drop Heights - Right Side Impact Location, Flat A nvil............ 91
XI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 36: 20% Trimmed Mean Peak Sensor Force Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil.............................................................................91
Table 37: 20% Trimmed Mean Peak Sensor Force Across Groups
and Drop Heights - Left Front Impact Location, Flat Anvil............ 92
Table 38: 20% Trimmed Mean Peak Sensor Force Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil.............................................................................92
Table 39: 20% Trimmed Mean Peak Sensor Force Across Groups
and Drop Heights - Right Side Impact Location, Flat A nvil............ 92
Table 40: Linear Contrast Analysis of Peak Sensor Force Across Groups
and Drop Heights - Left Front Impact Location,
Hemispherical Anvil.............................................................................95
Table 41: Linear Contrast Analysis of Peak Sensor Force Across Groups
and Drop Heights - Left Front Impact Location, Flat A nvil............ 95
Table 42: Linear Contrast Analysis of Peak Sensor Force Across Groups
and Drop Heights - Right Side Impact Location,
Hemispherical Anvil.............................................................................95
Table 43: Linear Contrast Analysis of Peak Sensor Force Across Groups
and Drop Heights - Right Side Impact Location, Flat A nvil............ 96
Table 44: r Values Obtained From Percentage Bend
Correlation Procedure..........................................................................96
XU
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
In order to address the need to monitor local contact forces during head impacts, a
custom transducer was designed to monitor local force distribution patterns on an ISO
size E magnesium headform concurrently with linear acceleration measures from an
accelerometer located at the center of gravity of the headform. The response
characteristics of the transducer were found to be predictable and acceptable given the
limitations of high speed data collection in a confined environment. During bicycle
helmet testing, the output from the transducer was also found to be sensitive to
ventilation openings and ventilation channels located on the underside of the helmet
liner.
The effect of helmet liner density upon local contact forces and headform acceleration
was evaluated using an identical bicycle helmet model fabricated in four different helmet
liner densities. The study found that peak headform acceleration and peak local contact
sensor force values were significantly lower for the low density helmet liners when
compared to the highest density of helmet liners during low to moderate energy impacts.
During the high energy impact tests against the hemispherical anvil, the lower density
helmets bottomed out, resulting in high local contact forces and high peak headform
acceleration values relative to the higher density helmets. These results suggest that a
tradeoff does exist in terms of the protection offered by low density helmets at low to
moderate energy impacts compared to the performance of higher density helmets during
the higher energy impacts.
The study also found that a poor correlation exists between peak headform acceleration
and local contact force suggesting that future head protection standards should include
evaluation of the load distribution characteristics of the helmet.
xm
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c h a p t e r 1
INTRODUCTION
In the United States, bicycle accidents cause approximately 900 deaths, 580,000
emergency room visits and 1.2 million visits to physician offices each year (Rivara et al.,
1996). The estimated economic cost of these injuries exceeds one billion dollars per year
(Buntain, 1985). Of these accidents, at least one third of all recorded injuries involve head
injury (Wood and Milne, 1988) and head and neck injuries have been found to occur in as
many as 86 percent of all cases involving fatal accidents (Fife et al., 1983).
Using a case control study to monitor the effects of head protection upon head injury,
Thompson et al. (1989) found that the use of bicycle helmets could reduce the risk of
head injury by as much as 85 percent and reduce the risk of brain injury by as much as 88
percent. More recent work by Rivara et al. (1996) involving a hospital based comparison
study of helmeted and unhelmeted accident admissions has continued to demonstrate the
effectiveness of bicycle helmets at reducing the risk of head injury.
Given the evidence of bicycle helmet effectiveness, bicycle helmet safety programs have
been introduced as an aggressive measure to counter the problem of head injuries to
unhelmeted bicycle riders. As well, several countries, cities and states have adopted
mandatory bicycle helmet laws which require all bicycle riders to wear proper protective
headgear. At the present time in the United States there are 15 states with mandatory
bicycle helmet use legislation and 55 cities, counties or localities with some form of
bicycle helmet law (Bicycle Helmet Safety Institute, 1997).
Laws requiring cyclists to wear approved helmets have been effective in reducing the
number of head injuries experienced during bicycle accidents. Since July 1, 1990 a law has
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
been in effect in Victoria, Australia which requires all bicycle riders to wear an approved
bicycle helmet while riding a bicycle. Evaluation of hospital admissions in Victoria
following the introduction of this law showed a reduction in the number of bicyclists
with head injuries ranging from 37 to 51 percent depending upon age category (Vulcan et
al., 1992). In 1989, New York State introduced a state wide law requiring all bicyclists to
wear proper head protection. Evaluation of hospital admission data between 1990 and
1995 indicates bicyclists under 14 years old have experienced a 55 percent reduction in
the number of bicycle related traumatic brain injuries. For those bicyclists over 14 years
of age, there has been a 16 percent reduction in the number of traumatic brain injuries
(Bicycle flelmet Safety Institute, 1997).
At present; however, head injuries still occur to helmeted bicycle riders. Work by Smith
et al., (1994) and McIntosh and Dowdell (1992) have found that approximately 25% of all
helmeted riders who are involved in accidents sustain some form of head injury.
Furthermore, in a study of pediatric head injuries involving bicycles, the number of
pediatric riders who sustained a head injury while wearing a bicycle helmet was found to
be as high as 76% (Grimard et al., 1995). These head injuries were found to range
between minor concussion and severe traumatic brain injury. Clearly some of these
injuries may be due to extremely high energy impacts which are in excess of the
performance limits of contemporary bicycle helmets. However, research by Smith (1997)
gathered from a group of one hundred accident involved helmets found that sixteen
riders had sustained injuries ranging from mild concussion to subarachnoid hemorrhage
yet the mean percentage crush of the helmet’s expanded polystyrene liner at the primary
impact site was only twelve percent of its total thickness. This would suggest that not all
impacts are completely depleting the energy absorbing capacity of the helmet liner and
subsequently the helmet liner is not effectively absorbing a sufficient amount of the
energy of the impact or effectively distributing the impact forces to reduce the risk of
head injury.
In order to understand how a bicycle helmet absorbs energy and protects against head
injury, it is important to understand the various mechanisms by which head injury can
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
occur. Head injuries can occur from inertial loading of the body as well as direct head
contact. Typically head injuries in bicycle accidents results from direct contact of the
head against one or more unyielding rigid surfaces (i.e. asphalt, an automobile, a tree,
etc.) (Smith et al., 1994, McIntosh and Dowdell, 1992, Smith, 1997). As a result of this
direct contact, the skull will be rapidly decelerated while the brain will continue to move
relative to the skull. The brain will contact the internal surface of the skull and then will
continue to lag behind the motion of the skull as the skull itself rebounds away from the
contact site. It is this relative motion that causes a large majority of head injuries in
bicycle accidents. In addition to the injuries caused by the relative displacements of the
structures of the skull and brain, the brain may also be injured due to local skull
deformation at the contact site. As the skull remains in contact with the unyielding
surface, it will experience local bending and localized stress concentration. This localized
deformation and stress concentration then cause the underlying structures of the brain to
experience deformation as well. If the impact forces are sufficient, the skull may
experience fracture. Different portions of the skull are more susceptible to fracture due to
their decreased thickness (Nahum et al., 1968). If the magnitude of the deformation is not
high enough to induce skull fracture, it may still be high enough to induce contusions to
the surface of the brain and surrounding structures. The edema and hemorrhaging that
arises as a result of these contusions can increase the pressure within the brain, causing
further damage.
To understand the physical relationship between the skull and brain system and external
force application, it is necessary to understand Newton’s second law of motion, which
states that the rate of change of momentum is equal to the sum of the forces producing
that momentum. Realizing that acceleration is the rate of change of velocity over time,
the definition of this relationship can be simplified to the sum of all forces acting on a
body are equal to the product of the mass of the body times the acceleration of that
body. Given a constant mass, reducing the forces acting on a body for a given impact will
reduce the magnitude of the subsequent acceleration of that body. Since acceleration is
the rate of change of velocity over time and velocity represents the rate of change of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
displacement over time, these three physical variables are intimately related. A rapid
change of displacement will be reflected by a large velocity and similarly a large velocity
change will be reflected in the form of a high acceleration. Therefore, from a protective
point of view, the role of a bicycle helmet is to minimize the acceleration levels during a
given impact by minimizing the amount of force apphed to the skull to cause the
acceleration. As mentioned earUer, these forces are appUed to the skull at the level of the
helmet/skull interface. Effective reduction of these forces, and the manner in which these
forces are applied to the skull brain system (through the use of a contemporary bicycle
helmet) will minimize the resulting brain deformation that occurs during that impact.
Bicycle helmets with energy absorbing liners were first introduced in the mid 1970s and
the basic design has not changed significantly since that time. They are composed of three
fundamental components, a shell, a liner and a retention system. The helmet shell is
usually composed of some form of polyethylene derivative and acts as a barrier against
direct contact to the wearer’s head and as a barrier to prevent direct penetration. The
energy absorbing liner below the helmet shell is typically fabricated from expanded
polystyrene (EPS) and it is this structure which effectively absorbs the energy of a head
impact. Although other energy absorbing materials are available, EPS is the preferred
material for bicycle helmet liners due to its low cost and ease of manufacture and
material stabifity over a variety of conditioning environments.
EPS helmets are fabricated by taking EPS bead material and expanding the beads in a
large container (the expander) using high temperature. The high temperature in the
expander causes the pentane gas within the EPS beads to expand. The beads are allowed
to remain in the expander unit until the desired volume is reached. The beads are then
fed from the expander unit into a holding bag where they are allowed to cure for several
days. This curing process causes some of the pentane to be released from the individual
beads and this improves the final adhesion process within the mold cavity. Once the
beads are properly cured, they are fed into the helmet mold or cavity and are allowed to
further expand under high temperatures with steam. The steam process and the high
temperature causes the individual beads to expand shghtly and fuse together to form the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
final product. The helmets are then ejected from the mold and allowed to dry and return
to room temperature. Since the volume of the helmet liner is controlled by the size of the
mold, the density of the helmet is dictated by the number of beads that have been forced
into the mold cavity.
In an actual impact, a typical bicycle helmet with an EPS liner will absorb energy by
having work done upon it by the impact forces apphed to the helmet. The amount of
energy absorbed by the helmet (i.e. the net change in energy) is equivalent to the amount
of work done on the helmet. Work is defined as the product of the magnitude of the
displacement times the component of the force in the direction of the displacement. For
a given thickness and finer stiffiiess, there is a maximum amount of energy which can be
absorbed before the helmet will “bottom out”. In this case, the helmet will have absorbed
a maximum amount of impact energy and any remaining energy will be transmitted
directly to the head according to the work-energy theorem. In order to maximize impact
protection, the manufacturer could choose to increase the thickness of the finer in order
to maximize the amount of material available for deformation. This will increase the
damping characteristics of the material and increase the amount of energy that can be
absorbed during an impact. Unfortunately, this would require the manufacturer to
increase the size of the molds that are used to make the helmets and this can be quite
costly depending upon the complexity of the mold. As well, from a marketing
perspective a larger bicycle helmet is not desirable due to the fact that people do not want
to wear large, bulky helmets while riding a bicycle. As a result, bicycle helmet liners
remain at a nominal thickness of one inch (Smith, 1997).
A far more economical approach to change the energy absorbing characteristics of the
helmet finer is to adjust the helmet finer stiffiiess. By decreasing the helmet finer density
and hence the helmet finer stiffness, the manufacturer can manipulate the force-
deformation characteristics of the bicycle helmet to maximize both the load distribution
and finer deformation so that “bottoming out” only occurs during extremely high level
impacts. Recalling the work-energy theorem, if the finer deformation is maximized, then
the amount of energy absorbed by the helmet is maximized as well. Unfortunately, due
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to limitations within the thickness of the helmet liner, the liner deformation can never be
one hundred percent; therefore, there is an upper limit to the amount of energy that can
be absorbed by the helmet liner. The goal of the helmet designer is to minimize the
amount of residual energy remaining from the impact and consequently minimize the
magnitude of the forces transferred to the head.
Regardless of the design objectives of the helmet designer and the helmet manufacturer,
all helmets manufactured in N orth America must meet minimum performance standards
in order ensure that they will perform adequately under impact situations. At the present
time, there are three published bicycle helmet standards in the United States (ASTM
FI 163, Snell B95, ANSI Z90.4). A fourth bicycle helmet standard developed by the
Consumer Product Safety Commission is currently being introduced into the Federal
Register as a national standard for bicycle helmets. Each standard has specific
requirements regarding coverage, field of vision, retention system performance and
impact attenuation performance.
Although each standard may have slightly different requirements, the basic equipment to
evaluate helmet performance remains the same. The impact attenuation capabilities of
the helmet are measured by dropping a helmeted test headform onto a specific test anvil
from a specified drop height in order to cause a specific impact velocity or impact energy.
The acceleration of the test headform and helmet system is measured by an accelerometer
which is located at the center of gravity of the test headform. If the peak headform
acceleration for a given test remains below the required value (typically 300 g), then the
helmet is considered to meet the impact attenuation requirements of the test standard.
The range of impact velocities for the currently published United States test standards are
between 4.8 m/s and 6.2 m /s and this appears to be in the same range as the impact
velocities obtained during replication studies of accident involved bicycle helmets
(McIntosh and Dowdell, 1992, Smith et al., 1994). A standard assumption is made that a
helmet that performs well when tested at these impact velocities also performs well at
lower impact velocities; however, research on the energy absorbing characteristics of
sport surfaces and other athletic equipment suggest that this assumption may not be true
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Nigg, 1990). A better approach may be to evaluate a helmet’s impact attenuation
performance capabilities over a wide range of potential impact velocities.
As indicated by the test standards described above, acceleration has been used as the
standard test variable by which to evaluate risk of head injury for the past fifty years.
Standard test headforms have been consistently used to evaluate the overall acceleration
of the skull/brain system. This is based on the fundamental knowledge that brain injury
is the result of excessive motion of one part of the brain relative to another (Viano, 1990).
As mentioned above, during a head impact the scalp and skull tend to deform and this
deformation causes subsequent deformation of the underlying structures of the brain
along with the skull and scalp. If the deformation becomes too great, the skull will
fracture. If the deformation is still very high, even though there is no skull fracture, brain
injury may potentially occur due to the local contact effects. Therefore, head injuries can
be due to inertial effects (i.e. linear and angular acceleration) as well as due to local
contact injury.
Although the accelerometer can provide some insight into the deformation behavior of a
helmet, it cannot predict the relative risk of injuries due to local contact stresses. Past
research has clearly indicated that an analysis of the pressure distribution characteristics
of a given impact is imperative to monitoring head injury potential due to contact
(Patrick, 1972). In addition to the inertial response of the head during impact, it is the
magnitude and distribution of these local stresses which are also important in
understanding the mechanisms of head injury and the role that a bicycle helmet plays in
reducing these head injuries. Obviously, a well designed helmet will maximize the area
over which an impact is distributed in order to keep the local loading effects as small as
possible.
The need to monitor load distribution characteristics of a head impact has been known
for several years although the technology has not been available to accurately measure
load distribution without significantly altering the system monitored (typically a skull or
headform). In 1981 in Washington, DC, a consensus workshop on head and neck injury
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
criteria identified measurement of localized contact forces as a future research direction in
order to fully understand head injury mechanisms due to contact (National Highway
Traffic Safety Administration, 1981). This position was reiterated at a later symposium
on Head Injury Mechanisms in 1981 in New Orleans, Louisiana (Symposium Report,
1987).
At the present time, very little research has been conducted to evaluate the load
distribution characteristics of head impacts. The automotive industry has developed
several technologies to monitor facial impact load distribution which appear to be quite
effective at monitoring local contact forces during impact situations (Moulton et al.,
1984, Warner et al., 1986, Perl et al., 1989). The only localized force measurement system
developed for monitoring localized forces applied to the head during bicycle helmet
impacts has been developed by Long et al. (1989). Using a two layer pressure sensitive
photographic material (Fuji Film), the authors were able to measure the maximum
normal components of the impacts forces as they were transferred to the headform
during the impact. A red coloration appeared in direct proportion to the amount of force
apphed to the vertex of the helmeted headform and this color distribution was
subsequently cafibrated and expressed in terms of local contact force. The results
obtained from the Fuji Film indicated that several models of bicycle helmets, particularly
those with large ventilation openings, demonstrated high loads distributed over the entire
area of the liner in contact with the pressure sensitive paper. The Fuji Film also showed
higher local forces for small radius test anvils (e.g. hemispherical) when compared to a
flat test anvil. Correlations between the load cell data and the accelerometer data for each
drop test suggested that a very poor correlation existed between peak local force (i.e.
force over a 1 cm^ area) and headform acceleration. It was therefore proposed that a load
distribution test be introduced into the Australian bicycle helmet standard as a means to
evaluate bicycle helmet load distribution performance. This current test is referred to as
AS/NZS 2512.9:1996 “Methods of Testing Protective Helmets: Method 9:
Determination of Load Distribution” and is the only standard test for localized load
distribution in the world.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the Fuji Film does represent a tremendous development in the ability to
monitor localized contact forces it does have its limitations. Firstly, the film only
captures maximum load application at a specific location. It does not provide any
information regarding the time history of the apphed force. This is important in
understanding the nature of force development and transmission at the head as well as at
the head/helmet interface which has significant imphcations for head injury risk due to
contact. Secondly, the material behavior of a helmet under fracture conditions (i.e.
helmet liner fracture during impact) and the subsequent force transmission can provide
tremendous insight into effective bicycle helmet design. Lastly, the transducer is
cahbrated statically even though it is used dynamically and the amount of error
associated with this procedure is largely unknown.
Current acceleration based test instrumentation accurately monitors bicycle helmet
performance with regards to protection against inertial type injuries; however, there
remains a very real need to understand local contact forces as they pertain to head
impacts in general and more specifically to the risk of contact injury during bicycle
helmet impacts. The effect that the large and numerous ventilation holes and venting
channels found in current bicycle helmet designs have upon the load distribution
characteristics of a bicycle helmet impact are not known at this time. An effective system
which could monitor local contact forces in the time domain would certainly provide
some insight into the effect of these ventilation holes and into the manner in which
bicycle helmets with ventilation holes and channels distribute loads during impact
situations. This information could then be related directly to available tolerance literature
in order to identify potential situations where the forces approach tolerance levels or
where the load path radiates towards structures which are less tolerant to local contact
stresses.
Problem Definition and Rationale
At the present time, there is a need to better understand the nature of contact type
injuries that occur to helmeted bicycle riders. It is well accepted that one of the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fundamental design criteria for a proper helmet is that the helmet distribute the impact
forces over as wide an area as possible. The exact manner in which a helmet satisfies this
design criteria is not well known at this time. It is assumed that although the helmet is
designed to distribute the load over the surface of the head, the skull will still undergo
local bending due to the development and concentration of local stresses. As a result of
this local bending, there is a high probabiUty that the skull and the underlying structures
may experience local stresses that are sufficient to cause local trauma. Injury may include
(but not be limited to) laceration, abrasion, subgaleal hemorrhage, coup contusions and
skull fracture. Furthermore, these local contact stresses can induce large amounts of local
skull bending which can in turn result in the development of negative pressure gradients
within the brain. Remote injuries from local concentrated skull bending may include
basilar skull fracture as well as contre-coup contusions. Therefore, a helmet’s ability to
distribute load may be directly correlated to the helmet’s ability to reduce contact type
injuries.
There have been few studies that have attempted to measure local contact stresses at the
headform-helmet interface. These studies have been limited in their ability to provide
time series information regarding the load distribution characteristics of a helmeted
impact. The results of previous studies provide either maximum load information (e.g.
Fuji Film) or in the case of analog force data collection, the load measurement area is
confined to a 1 cm" area (AS/NZ 2512.9:1996). It is this methodology that is included in
the only bicycle helmet standard in the world which currently includes a peak local force
test criteria. Unfortunately, this test is performed at an energy level which is well below
that which would normally be experienced by a helmeted bicycle rider during an
accident.
The development of instrumentation which could capture localized forces on an impact
test headform would provide unique time series information regarding the load
distribution characteristics of helmeted impacts into a variety of different test impact
surfaces. Current helmet designs could be evaluated and the resulting localized force
information could be related directly to available tolerance criteria regarding skull
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fracture and the risk of associated head injury. Detailed test procedures could also be
developed for inclusion into current and future bicycle helmet test standards.
Objectives of this study
1. To develop a test apparatus to measure time domain force distribution
information of a typical helmeted impact.
2. To validate a test apparatus and ensure its reliabihty and repeatability to perform
task one.
3. To test and evaluate the effect that bicycle helmet liner density has upon the load
distribution characteristics of a bicycle helmet impact. Comparisons will be made
between helmet liner density and known impact tolerance criteria.
Recommendations will be provided regarding future bicycle helmet designs.
Statem ent of hypothesis
It was hypothesized that variations in bicycle helmet liner density would result in
significant differences in peak local forces between helmet liner densities when tested
under the same impact conditions. It was further hypothesized that the peak headform
acceleration response of the lower density helmets would be significantly different than
the higher density bicycle helmet liners when tested under the same conditions. It was
also hypothesized that the data would indicate that the lower density production type
bicycle helmets would provide a greater amount of load distribution (i.e. wider loading
area and lower peak force per unit area) as compared to higher density production
bicycle helmets.
Delim itations
For the purpose of this study, the following delimitations were made:
1. The test instrumentation will be developed in such a manner that the transducers
must be permanently affixed to the test headform. Consequently, the force
measurements must be taken at a specific location on the headform. The load
1 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
information that is obtained from this impact location cannot be related to other
impact sites on the headform.
2. Due to the limited number of test samples that can be obtained, only four drop
heights and two impact anvils can be used for helmet testing. These drop heights
and impact anvils cannot be considered to be inclusive of all potential impact
energies and all potential impact surfaces that may exist in real world bicycle
accident situations.
3. Forces measurement is restricted to a plane orthogonal to the surface of the test
headform.
Limitations and Assumptions
1. It is known that a rigid headform cannot simulate the natural response of a
human head. It shall be assumed that the response of the rigid magnesium
headform represents the “worst case” response of a human head under similar
loading situations.
2. The reliance upon cadaveric tolerance data shall be considered a limitation of this
study. The estabUshment of tolerance criteria based on the response of aged
cadavers must be considered to be conservative in nature relative to a healthy
adult population.
3. It shall be assumed that the helmets selected for this research are representative of
helmets that are currently available in the marketplace.
4. It shall be assumed that the two test anvils that have been selected as part of this
study are representative of the majority of impact surfaces that are observed
during bicycle accident situations.
5. The addition of force transducers to the outer surface of the headform shall be
assumed to have no adverse effect on the location of the center of gravity of the
headform or the moment of inertia about the center of mass of the headform.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c h a p t e r 2
REVIEW OF LITERATURE
Functional Anatomy
In order to appreciate the potential benefit of any form of head protection device, it is
imperative that one understand the system which is being protected, namely the skull
and its contents. Each anatomical structure has unique characteristics which, by design,
can effectively reduce or limit the forces that are applied to the head during a head
impact. This section will detail and describe individual structures which have been
extensively described elsewhere (Anderson, 1978, Johnson, 1982).
The Scalp
The scalp is the first tissue to receive trauma as a result of a direct impact to the head. It
consists of five layers, beginning with the dermal layer. This layer may contain hair,
which under certain circumstances may have some protective value; however, for the
majority of the population, the hair offers little or no protective value. Below the dermal
layer is the subcutaneous layer or superficial fascia which consists of dense fatty tissue
tightly bounded to both the skin above and to the layer below, the galea aponeurotica.
The subcutaneous layer is well vascularized; subsequently, concentrated blows frequently
cause rupture to small vessels and lead to sharply locahzed hemorrhages trapped within
this layer.
The third layer consists of the epicranial muscle which has two bellies located at the front
of the scalp (the frontal muscles) and another two bellies located at the rear of the scalp
(the occipital muscles). The galea aponeurotica acts as a flat intermediate tendon laying
between these two antero-posterior oriented muscles. The right and left occipital muscles
originate from the superior nuchal lines of the skull and the right and left frontal muscles
insert anteriorly into the skin at the eyebrow level.
1 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The fourth layer consists of loose areolar tissue which is dispersed everywhere between
the gala aponeurotica and the underlying fifth scalp layer, the periosteum of the skull.
This loose layer allows for free scalp movement and greatly facilitates shp off of
tangential blows to the head. This movement also permits a greater amount of
distribution of hemorrhagic and edematous fluids and consequently greater absorption
by the surrounding structures. As a result of the relatively weak bond between the skull
and the scalp along this layer, most scalping injuries involve a separation along this
boundary. Given the amount of energy required to induce a scalping type injury, most
injuries involving the scalp are avulsion type injuries with a scalp flap as opposed to a
complete scalping injury.
The Skull
The primary function of the skull is to support and protect the cramai contents. It is
made up of eight separate bones, four single bones (frontal, occipital, sphenoid, and
ethmoid) and two paired bones (parietal and temporal). These bones fuse together along
wavy sutures fines which increase the overall stiffness of the joints between the bones.
The general shape of the skull is spheroid which permits the greatest protection from
external impacts with the minimum amount of mass.
The structure of each bone of the skull consists of three layers. The outer layer, known
as the outer table and the inner layer, the inner table or lamina are composed of dense
bone. The middle layer between the inner and outer table is known as the diploe layer. It
consists of spongy bone which is largely responsible for hematopoeisis, or the production
of red blood cells (Anderson, 1978).
The mechanical characteristics of the skull are largely dependent upon the relative
thickness of each of these three layers. Different parts of the skull have different thickness
and consequently different tolerances to injury. For example, the frontal bones and the
occipital bones are quite thick and have quite a high threshold against fracture (Nahum et
al., 1968). Conversely, the temporoparietal bone is quite thin and has a much lower
tolerance against skull fracture and local contact stresses (Nahum et al., 1968).
1 4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
When the inner surface of the skull is viewed from a superior aspect, there appear to be
three different compartments or depressions on the floor of the skull, which support and
compartmentalize the brain. The most anterior depression on the floor of the skull is the
anterior fossa. This portion of the skull is formed by the orbital plate of the frontal bone
which also acts as the roof of the orbital cavity. Traveling medially, the anterior fossa is
formed by the cribiform plate of the ethmoid bone. The anterior fossa houses the frontal
lobes of the brain.
The middle cranial fossa is divided into right and left depressions by the body of the
sphenoid bone. The middle cranial fossa is shaped somewhat like a butterfly with its
wings spread; where the fossa for the hypophysis represents the body of the butterfly and
the lateral parts of the fossa represent the wings. It is in these lateral fossa where the
temporal lobes of the brain he. The vertical riser which separates the middle fossa and the
anterior fossa (and the orbital cavity) is comprised of the greater wing of the sphenoid
bone. Physically, the orbital plate does not meet the lesser wing of the sphenoid bone;
consequently, a space known as the superior orbital fissure, exists between them. This
fissure allows for neural communication between the middle cranial fossa and the orbital
cavity.
The posterior fossa houses the cerebellum, pons and medulla and geographically is the
most inferior of all three fossa. Its floor is formed by the occipital bone with the foramen
magnum located just anterior to the center of the floor. The bones which separate the
posterior fossa and the middle fossa are the posterior surfaces of the petrous portion of
the right and left temporal bones. In the median area, the separation is formed by the
basical portion of the occipital bone and the posterior surface of the body of the sphenoid
bone which includes the dorsum sellae.
These fossa resemble three steps, each with a tread and riser, which descend anterior to
posterior and act to further separate the brain, thereby reducing the risk of brain pitch,
roll or yaw; brain motions which could easily occur if the internal surface of the skull
were a complete and perfect sphere.
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The Brain Coverings
Immediately adjacent to the skull is the dura mater which exists throughout the entire
internal surface of the cranial cavity. It is composed of two layers, the periosteal layer
which adheres to the skull and the meningeal layer. In most places across the skull, these
two layers are fused and are not separable by dissection; however, at the superior sagittal
sinus and the lateral durai venous sinuses, these two layers are separated. These sinuses
receive all the venous drainage from the brain.
At three points within the skull cavity, the meningeal layer is drawn inward in a double
fold configuration to create the falx cerebri, the tentorium cerebeUi, and the falx
cerebeUi. These falx and tentorial structures act to further separate the brain into
compartments for protection against rolling and rotation. The falx cerebri acts as a
divider between the two cerebral hemispheres. It’s lower border is immediately adjacent
to the corpus callosum. The tentorium cerebeUi is a fold of dura mater that forms a
partition between the posterior cranial fossa which houses the cerebeUum and the
posterior part of the supratentorial brain. The smaUer falx cerebeUi projects along the
midline in the floor of the posterior cranial fossa and acts as the partition between the
right and left posterior portions of the cerebeUum.
The external surface of the dura mater is covered with meningeal arteries which lie
between the dura mater and the bone and branch to both. During impact situations, the
dura mater may become stripped away from the bone; subsequently, the many smaU
arteries may become ruptured and cause further tearing of the durai layer from the skuU.
This accumulation of blood above the dura is known as an extradural or epidural
hematoma (Unterhamscheidt and SeUier, 1966).
At several locations along the dura, the meningeal arteries become quite large and may in
fact become embedded in the bone. In the event of skuU fracture, these embedded arteries
can easUy become severed and cause a rapid accumulation of extradural blood. The most
prevalent location for the extradural hematomas appears to be in the middle cranial fossa
as a result of a sudden severing of the middle meningeal artery (Johnson, 1982).
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The Subdural Space
Below the durai layer is the arachnoid layer which is located on the inner aspect of the
dura mater but does not adhere to the dura except at certain sites, such as at arachnoid
granulations and the locations where the veins pierce the dura to enter the durai venous
sinuses. The pia mater is a thin layer of connective tissue which adheres to the surface of
the brain and spinal cord; therefore, all arteries and veins which enter the brain must pass
through the pia mater. Between the pia mater and the arachnoid layer is the
subarachnoid space. This region is not actually a space, but rather is a large meshwork of
arachnoid cells which cover the region between the pia mater and the arachnoid layer. In
addition to these cells, the subarachnoid space is filled with cerebrospinal fluid which
further acts to protect the brain during rapid movement.
The venous blood return through the subarachnoid space is carried out by the
parasagittal bridging veins which travel from the neural tissue through the subarachnoid
space, through the arachnoid layer and into the durai sinuses. As they pass from the
arachnoid to the dura mater they cross the subdural space. If the brain experiences large
angular displacements, these bridging veins will become stretched and may potentially
tear in the vicinity of the subdural space, thereby causing a subdural hematoma
(Gennarelli and Thibault, 1982).
The Brain
As mentioned earlier, the primary function of the skull is to protect the brain tissue. It is
composed of two primary hemispheres (left and right) which are separated by means of
the falx cerebri. The two hemispheres are connected via the corpus callosum which is a
large band of commisural fibers. Large displacement of the brain may cause contusions to
the medial surface of the hemispheres as they come in direct contact with the falx cerebri.
Small focal lesions may also be found along the corpus callosum as a result of this
displacement.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The cerebellum is separated from the two upper hemispheres of the brain by the
tentorium cerebeUi. Immediately adjacent to the cerebeUum is the brain stem which is
made up of the midbrain, the pons and the meduUa oblongata. Only the meduUa
oblongata descends down through the foramen magnum.
Although the cerebrum is surrounded by cerebrospinal fluid and somewhat buffered
from the effects of inertial loading, the brain can stiU experience contusions as a result of
direct contact against the cranial waU or the durai folds. Regions such as the inferior
poles of the frontal and temporal lobes of the brain are particularly susceptible to
contusion as a result of direct contact against either the anterior or middle fossa
structures. Similarly, the inferior surface of the cerebeUum may be contused against the
floor of the posterior cranial fossa.
Mechanisms of H ead Injury
The manner in which forces appUed to the head actuaUy cause head injury has been of
interest for centuries. Although the general mechanisms responsible for head injury have
been known for many years, they have only recently been related back to the primary
injuries that are commonly observed in clinical situations.
Head injury mechanisms can be classified under the broad and general categories of head
injury due contact forces and head injury due to inertial forces. The exact mechanisms of
head injury and the potential outcome as a result of these loading situations are described
in Table 1 (GennareUi, 1987).
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CONTACT INTURIES OUTCOME
Contact With Integument Laceration, Abrasion, subgaleal hematoma/hemorrhage
Skull Deformation Local Effects:
Skull fracture (Linear, Depressed, ComminutecQ
Suture Separation, Ping Pong Fracture (occurs usually in small children)
Coup Contusion (Contusion Directly Beneath Impact Site)
Remote Effects:
Vault Fracture, Basilar Skull Fracture
Stress Waves Coup-Contre-Coup Contusions
Extracerebral Hematoma/ Hemorrhage
INERTIAL INTURIES OUTCOME
Surface Strains Bridging Vein Rupture (Subdural Hematoma)
Epidural Hematoma, Subdural Hematoma
Subarachnoid Hemorrhage, Hematoma
Deep Strains Concussion Syndrome
Diffuse Axonal Injury (DAI)
Table 1. Head Injury Mechanisms and Clinical Outcome (Gennarelli, 1987)
Head Injury Due To Contact Forces
Contact injuries are those injuries that are caused whenever an object strikes the head
directly or when the head strikes an object. As a result of the contact, local skull bending
occurs and the resulting skull deformation causes a compressive strain on the outer table
of bone and a tensile strain on the inner table. Since bone is weaker in tension than in
compression (Gurdjian et al., 1947), these tensile loads cause a fracture to begin at the
inner table and propagate along the path of least resistance away from the impact site.
This material property of skull bone was first discovered by Gurdjian et al. (1945) using a
strain sensitive material called “Stresscoat” (Magnaflux Corporation) The technique for
strain evaluation involves appUcation of the strain sensitive material directly to the skull
prior to impact. The properties of the “Stresscoat” material are such that it will crack in
response to tensile strain occurring to the material upon which it is sprayed. Gurdjian
and Lissner (1945) found that the stresscoat cracks first appeared on the internal surfaces
of the skull and based on this finding, the authors concluded that linear fractures of the
skull arose from failure of the bone from tensile stresses created while bending the bone.
In a later study (1946) they discovered that the amount of absorbed energy necessary for
threshold deformation of the skull differed in various regions of the skull. In the
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
midfrontal region, blows of 1.58 to 2.03 Joules of energy (14 to 18 in-lb) caused
deformation similar to those obtained with the 0.90 J of energy (8 in-lb) applied to the
mid-occiput. The amount of energy dissipated in deformation and fracture of the skull
was considered by the authors to be small. They also noted that the shape and velocity of
the object being struck influenced the strain patterns and provided a directional response.
They also found that the path of the strain produced in the skull depended on the shape,
contour, and thickness of the bone in the area of impact. Deformation patterns were
more common in the weaker buttress regions of the skull, particularly in the frontal,
foramen magnum and temporoparietal regions; regions which contain many sutures and
therefore are more likely to sustain deformation. Their work also found a great range of
variation in the amount of energy required to fracture the skull simply due to the shape
and thickness of the skull and scalp. They concluded that the severity of the skull
fracture (linear, comminuted or depressec^ was a function of the size of the impact area
(or size of the impacting object), the local skull thickness and the rate at which the load
was applied to the skull.
In an effort to better define the tolerance threshold for different portions of the skull,
Nahum et al. (1968) conducted tests to the frontal, temporoparietal and zygomatic
regions of the head using a 1 square inch impactor and found the zygomatic region to
have the lowest threshold for fracture (890 N). The force required to produce clinically
significant fractures (with an effective contact area of 1 square inch) for the
temporoparietal region was found to be 2446 N and 4893 N for the frontal region. The
authors also conducted a pressure distribution calculation which showed a wide range of
local pressures compared to the average pressure over the area of load application. They
reasoned that the contour and rigidity of the head at the zone of contact could account
for local contact pressures which were as high as 200% over the average contact pressure
value. They also noted that the timing sequence of loading over the area of the impactor
could cause as much as a 30% increase in the average pressure over the measured area as
compared to a load cell which was placed in series with the impactor. These authors also
attempted to compare the localized force measurement at the surface of the skull (using
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the mass of the head and the acceleration of the head obtained at the time of impact) and
the driving force applied by the impactor. In some instances there was very good
agreement (i.e. high correlation) between the driving force measure taken from the
impactor and the calculated force value while in other instances the forces measured by
the impactor were found be 50% of the force value calculated using the accelerometer
mounted on the head. When the impacts were directed towards a metal block, the
agreement was within a few percent.
It must be mentioned that there has been no published research that has found a direct
correlation between the production of a skull fracture and the amount of cerebral
damage. Skull fracture may occur with or without brain damage or concussion, and
death may result without skull fracture. Consequently, merely preventing skull fracture
does not insure complete safety or protection of the brain.
Although skull fracture is clearly an endpoint in terms of skull fracture mechanics, in
many cases the bones of the skull will undergo a significant amount of deformation prior
to fracture (Nusholtz et al., 1984). As a result of this deformation, contusions directly
below the impact site may occur as a result of compressive forces apphed to the tissue
directly beneath the impact surface. Furthermore, due to the inertia of the brain
suspended inside the cranial cavity, it will tend to lag behind the motion of the skull.
Consequently, this lag in brain motion toward the impact site may cause tensile strains in
the area opposite the point of impact (contre-coup contusion) in addition to the local
deformation effects (coup contusion).
In addition to brain injury due to relative motion, it has been hypothesized that brain
injury may also occur as a result of pressure waves that propagate through the brain
tissue. Using animal models, Denny-Brown and Russell (1941) demonstrated the presence
of positive pressures at the site of the skull impact and the existence of negative pressures
at the opposite side of the skull. These findings suggested the presence of a large linear
pressure gradient within the skull immediately following impact. Gurdjian et al., (1958)
suggested that these pressure gradients may be the cause of damage that is found in the
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vicinity of the foramen magnum and brain stem. Later work by Thomas et al., (1967)
found that when the impact acceleration is sufficiently large, the pressure gradient no
longer varies linearly from anterior to posterior. Instead, the area of negative pressure
opposite the impact site becomes markedly larger and consequently may produce greater
damage.
To further complicate the complex response of the skull and brain under impact, Evans
et al., (1958) theorized that the longer the time during which the energy is absorbed by
the skull and brain or the longer the deceleration duration, the greater the magnitude of
the energy that can be absorbed or the peak deceleration that can be safely tolerated. The
authors also suggested that the rate of loading is also an important factor in the
absorption of impact energy, i.e., an impact can be better tolerated if the impact energy is
absorbed slowly (slow race of work) rather than absorbed rapidly (fast rate of work). The
authors also identified biological factors that likely affect the outcome of a given head
impact to include the thickness of the scalp and the skull, the age of the individual, the
state of health and possibly sex and race.
Head Injury Due To Inertial Forces
In the early 1960’s as automobile safety research began to flourish, head injury due to
contact forces or the application of linear forces was accepted by many experts as the sole
mechanism of head injury in automobile accidents. Consequently much work was done
to try to understand these linear acceleration effects. The potential existence of inertial
loading as a source of head injury was not proposed until Ommaya and Yamell (1969)
reported several cases of restrained automobile passengers who had sustained serious head
trauma yet they had not experienced any head contact. In other words, the application of
an arresting force to their chest caused relative head motion (both linear and angular)
between the head and the neck and the thorax as well as the skull and brain. This finding
of head injury without head contact was in direct contradiction to earlier work which
suggested that in order for head injury to occur, there must be head contact (and the
subsequent application of linear forces).
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to properly study this mechanism of inertial loading, Ommaya and Gennarelli
(1974) fabricated a form fitting helmet for applying uniform acceleration pulses to a
group of rhesus monkeys. One group was subjected to pure translational acceleration and
a second group to a combined loading situation of both translational and angular
acceleration. The form fitting helmet eliminated any contact effects. Their findings were
that those monkeys subjected to the combined loading condition experienced diffuse
head injuries while those monkeys that were subjected to translational acceleration
experienced focal or localized head injuries. Furthermore, all of the monkeys who were
subjected to the combined translational and angular acceleration effects experienced
cerebral concussion while none of the monkeys subjected to pure translational
acceleration effects experienced concussion. This finding clearly indicated that angular
acceleration effects represented a significant head injury mechanism that previously had
not been considered.
As noted in the table above, inertial injuries that occur primarily as a result of angular
acceleration effects can be categorized as both surface strain events and as deep strain
events. This conforms to the centripetal theory of head injury proposed by Ommaya
(1973) that states as the levels of angular acceleration increase, the level of strain within
the brain will increase and the level of neural damage will go deeper into the brain. The
“final” insult to the brain during inertial type loading is that the mesencephalic portion of
the brain experiences direct strain effects. At this level, significant neurological disruption
of the reticular activating system occurs along with the damage to other neurological
control systems. Clinically, this is diagnosed as concussion syndrome or in more severe
cases diffuse axonal injury (DAI).
Due to the unique structure of the brain and its surroundings, this centripetal theory
cannot be thought of as an “onion peel” theory of head injury. As indicated above, the
brain has surrounding compartmentalized structures including the falx and the anterior,
medial, and posterior fossa to limit brain motion. For any given force application, the
level and location of injuries will be largely dictated by the anatomical structures that are
stressed during an accident. Recent work on pediatric closed head injury cases by Levin
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
et al., (1997) has found a relationship between the depth of the brain lesion and the level
of functional outcome; thereby lending support for this theory of head injury.
Research by Gennarelli and Thibault (1982) has also illustrated the structural significance
of certain portions of the brain relative to the risk of head injury. Since it is known that
the brain will lag behind the skull when subjected to rotational acceleration, there is a
considerable amount of relative brain motion both within the brain tissue and between
the brain and the skull. This relative motion places a tremendous strain on the local
microscopic structures of the brain as well as the bridging veins which penetrate the dura
and supply blood to the brain. Rupture of these bridging veins has been found to be the
most common mechanism for subdural hematoma following inertial loading of the skull
and brain (Gennarelli and Thibault, 1982).
The Role of The Bicycle Helmet
As our understanding of head injuries and head injury mechanisms has increased, so too
has our understanding of head protection and head protection technologies. The original
principles of head protection, first proposed by Charles F. Lombard in 1953 (Lombard
and Advani, 1966) were actually fundamental principles of protective systems. The first
principle was to create a safe barrier or inviolate space between the object being protected
(namely the brain) and the externally applied forces. Secondly, the protective system
should be designed to minimize the forces that are transmitted to the object being
protected by absorbing as much impact energy as possible. The third principle was to
design a protective system that remained in place throughout the entire impact sequence
(thereby effectively attenuating impact energy throughout the entire impact sequence).
This technology was first applied to motorsport and aircrew helmets and it remains a
fundamental axiom of head protection to this day.
In accordance with these principles of protective systems, the bicycle helmet acts as a
barrier between the skull and brain and the impact surface. It is designed to distribute the
load over a large area and effectively increase the impact duration (compared to an
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
unhelmeted condition) by having work done upon it by the external impact forces.
Investigations of real world motorcycle accidents have clearly shown the different injury
profiles of helmeted and unhelmeted riders; therefore, a similar difference can be assumed
to exist between helmeted and unhelmeted bicycle riders (Hurt and Thom, 1992).
In order to measure the effectiveness of the protective systems developed using the
guidelines described above, early researchers in head injury mechanisms applied the
theorem that the risk of head injury was a function of both acceleration amplitude and
time. The Wayne State Tolerance Curve (WSTC) developed by Lissner et al. (1960)
remains as one of the few guidelines for injury that illustrates this relationship. Their
research on humans and animal subjects clearly showed a relationship between average
acceleration (over the pulse duration) and the total pulse duration. They observed that
the threshold for brain injury (as identified by a loss of consciousness) was very high if
the impact duration was short (5 ms or less) and very low if the impact duration was long
greater than 15 to 20 ms). Since this was an acceleration based criteria, protective systems
designers made every effort to ensure that the average acceleration values obtained from
their test subjects were below the threshold limits as suggested by the WSTC. They often
accomplished this by using effective padding material which dramatically increased the
impact duration and distributed the impact over a large area, thereby reducing the
average acceleration experienced by the test subject. This same fundamental technique is
applied today in the design of bicycle helmets. The expanded polystryene (EPS) Uner is
designed to maximize the impact duration and distribute the impact forces over as large
an area as possible in an effort to reduce the magnitude of the impact forces applied to the
skull and brain.
Under impact situations, EPS liners and human skulls and brains behave according to
Newton’s laws of motion. For example, Newton’s first law of motion states that a body
will remain at rest until an external force acts upon it causing it to move. Similarly, a
body in motion will remain in motion until an external force acts upon that body to
alter its motion. During an impact situation, a body is suddenly accelerated or
decelerated as a result of a collision with another body. Assuming rigid body mechanics.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
we know that the sum of all forces in this collision will be equal to the product of the
mass of the body and the acceleration (or deceleration) of the rigid body. In other words,
we know that ZF = ma.
We also know that immediately prior to the impact coUision, the rigid body possesses
translational kinetic energy (KE =1/2 mv^, rotational kinetic energy (RKE = 1/2 Ico^
and potential energy (PE = mgh). According to theorem, energy can neither be created
or destroyed; therefore, during a free fall helmeted head impact situation, the energy that
the helmet/skull/brain system has prior to impact (i.e. kinetic energy) must be somehow
transferred into a safer, less destructive form of energy in order to reduce the potential
for injury. This occurs in the form of work done on the helmet according to the work
energy theorem which states:
Work = A Kinetic Energy
Work is the product of force and displacement (Fd) where F is equal to the force of
impact in the direction of the deformation, d equals the magnitude of the deformation or
displacement that a helmet undergoes. The right side of the equation above equals the
change in kinetic energy of the object being worked upon. As the amount of work done
by the impact forces acting upon helmet liner increases, there will be a net increase in the
amount of energy absorbed by the helmet liner (i.e. a net change in kinetic energy). As
the amount of energy absorbed by the helmet liner increases due to an increase in the
amount of work done by the impact forces, the amount of energy available for transfer
directly to the head (in the form of work done by the impact forces acting upon the skull
and brain) is reduced. Depending upon the deformation capacity of the helmet liner,
almost all of the kinetic energy of a given impact can be effectively absorbed through the
destruction of the helmet liner. In other words, for a given amount of work, if the
displacement is maximized for a given helmet liner, then the resulting impact forces
acting in the direction of the displacement will be minimized. Recalling that ZF=ma, it is
possible then to hypothesize that for a given head mass, it may be possible to minimize
the acceleration of the head during a given impact situation by max im izin g the
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
displacement and maximizing the amount of energy absorbed by the helmet liner.
Unfortunately, due to material and physical limitations, helmet liner displacement can
never be one hundred percent; therefore, there is a limit to the amount of work which
can be done to the helmet liner during an impact situation.
There is a great deal of evidence available to indicate that helmets are effective at
absorbing impact energy and consequently reducing the forces that are imparted to the
head during bicycle accidents. A study by Dorsch et al. (1984) found that bicychsts who
struck their heads while wearing a helmet sustained less severe injuries that those who
were not wearing helmets. Evaluation of accident involved helmets by WiUiams (1991)
and Smith et al. (1994) have also found that helmets play a significant role in preventing
or attenuating the level of injury that would have been sustained by the rider had he or
she not been wearing a bicycle helmet. Epidemiological studies which compare helmeted
and unhelmeted populations have shown that helmets can reduce head injury by as much
as 85% and brain injury by 88% (Thompson et al., 1989).
Although there are clear benefits to wearing a helmet, some research has called into
question the density of current production bicycle helmets (Corner et al., 1987, Mills and
Gilcrhist, 1991). Specifically, it has been suggested that due to the current requirements
of bicycle helmet standards, many manufacturers have been forced to produce helmets
that are far stiffer than they need to be in order to effectively absorb the energy of a given
impact (Mills and Gilchrist, 1991). It has also been suggested that current production
bicycle helmet liners are too stiff for small children (Comer et al., 1987, Lane, 1986). If
helmet liners were in fact too stiff then the helmet liner would likely have a minimum
required velocity in order to initiate crush and energy absorption. If this minimum crush
velocity was not achieved during the impact, then the liner would experience very little
deformation with little work done upon the helmet; therefore, a significant amount of
force could potentially be transmitted through the helmet to the skull. This deformation
would subsequently cause local deformations in the region of contact and greatly increase
the probabihty of local brain contusions. Furthermore, this rapid deformation could
potentially initiate the development of pressure gradients within the brain which could
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cause further damage (Nusholtz et al., 1984). Data from one hundred accident involved
bicycle helmets gathered by the author found the mean helmet liner crush to be 12 %,
ranging from 0.8% to 38.7%; suggesting that the above situation may be true (Smith,
1997).
Given this potential scenario of stiff bicycle helmet liners, it would seem appropriate to
evaluate the effect of bicycle helmet liner stiffness and the maimer in which a helmet
distributes and absorbs the impact forces being apphed to the external surface of the
helmet. Obviously, from our knowledge of protective systems and energy dissipation, it
would be expected that the less stiff helmets would distribute the impact forces over a
wide area, thereby limiting the localized forces. Consequently, the local deformations of
the skull would be kept to a minimum. Using motorcycle helmets of varying liner
densities, Hopes and Chinn (1989) found the lowest headform acceleration values and
head injury criteria (HIC) values when testing the lowest density of helmet liner (25
kg/m^. This strongly suggests that helmet liner density can be manipulated in order to
minimize acceleration and HIC values and thereby reduce the potential for head injury.
Bicycle Helmet Standards
In order to ensure that any bicycle helmet design performs as an effective head protection
device, standards have been developed to provide minimum performance criteria for
coverage, impact attenuation and retention system performance. The amount of coverage
is specified in order to ensure that the helmet covers an appropriate area of the head
without significantly compromising the function of the helmet (i.e. it doesn't impinge on
the neck). The establishment of a test line on or above the coverage line also specifies that
region of the helmet that may be subjected to impact attenuation testing.
The impact attenuation tests are conducted using a rigid test headform which is fitted
with a uniaxial accelerometer and dropped in guided freefall onto an impact anvil. The
accelerometer records the peak headform acceleration and this value is used as a pass/fail
criterion for the impact attenuation performance of a helmet. Most contemporary
standards limit the maximum allowable headform acceleration to 300 g (ANSI Z90.4,
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Snell B90, CPSC, ASTM F1447). The lower the peak headform acceleration value, the
less transmitted force to the skull and brain. It has been su^ested by several authors that
this number may be too high for the general population (McIntosh et al., 1995, Mills and
Gilchrist, 1991) as well as for small children (Lane, 1986). To this end, the Canadian
Standards Association (CSA) have specified a lower allowable peak headform acceleration
for the general public and the Consumer Product Safety Commission have specified a
lower allowable peak headform acceleration limit for helmets designed for children
under 5 years of age. Table 2 summarizes the impact energy characteristics and maximum
allowable headform accelerations for each of the current bicycle helmet standards in
North America.
Standard Impact
Surfaces
Impact Energy Impact
Attenuation
Criteria
ASTM F1447 Flat,
Hemispherica
1 , Kerbstone
Flat: 6.2 m/s
Hemi/Kerb: 4.8 m/s
No peak
acceleration
above 300 g
Snell B95 Flat,
Hemispherica
1
Flat: 100 J
Hemispherical: 65 J
No peak
acceleration
above 300 g
CPSC - Draft
1996
Flat,
Hemispherica
1 , Kerbstone
Flat: 6.2 m/s
Hemi/ Kerb: 4.8 m/s
No peak
acceleration
above 300 g
CPSC - Below 5
years
Flat,
Hemispherica
1
Flat: 6.2 m/s
Hemi/Kerb: 4.8 m/s
No peak
acceleration
above 250 g
CSA D113.2 Flat, Cylinder Flat: 80 J
Cylinder: 55 J
80 J Impacts:
No peak above
250g
55 J Impacts:
No peak above
200g
Table 2. Bicycle Helmet Standards in N orth America
Current Issues in Head Protection Research
The actual force that is produced during an impact situation depends upon many factors
besides the amount of deformation that occurs to the helmet liner. It also depends upon
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the strength of the material and the area of the material that is being loaded (Newman,
1993). It has been suggested that stiff materials are not ideal for bicycle helmets in that
they require a minimum impact velocity to initiate crushing (Mills and Gilchrist, 1991).
The actual behavior of the spherical helmet liner is confounded by an increasing area as
the helmet liner velocity comes to zero. It is further comphcated by the fact that helmet
liners do experience a certain amount of rebound following impact; therefore, the actual
amount of liner deformation that occurs during impact is extremely difficult to measure.
These behavioral characteristics make computation of the actual amount of work done to
the helmet liner very difficult although simple Kelvin type models have shown excellent
agreement with actual test data (Mills and Gilchrist, 1991). The benefits of this modeling
approach are such that theoretical designs can easily be tested and compared to the
performance of existing bicycle helmet designs. Recently, more advanced fimte element
models have been applied to motorcycle flat anvil impacts with some limited success
(Wismans et al., 1996).
A similar modeling approach has been adopted by the automotive industry in an effort
to understand the complex mechanisms of head injury in automotive accidents. The
development and use of high speed computers has facilitated the use of finite element
models of the head as an effective tool for monitoring strain within the brain structures
(Bandak, 1996, Zhou et al., 1994, Ueno et al., 1996). It is hoped that the development of
these models will allow for a more complex analysis of the internal response of the brain
and brain tissue under impact situations. However, at the present time, there is a great
need for fundamental vahdation of these models to clearly understand the manner in
which the forces are applied to the head. Past cadaver work was largely restricted to skull
mounted transducers which monitored gross body motion. Only recently has there been
an interest in understanding local forces due to impact trauma and most of this research
has been directed at understanding the manner in which forces are apphed to the face
during steering wheel impacts (Allsop 1993, Warner et al., 1986, Perl et al. 1989). No
research to date has been conducted to physically evaluate the load distribution pattern of
forces which are apphed directly to the skull and the manner in which these forces are
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transmitted to the brain. Such research could be defined as the primary input fimction to
a head injury model and once this input fimction is well defined then the response of the
brain could easily be monitored using the finite element models.
As mentioned above, certain types of loading to the head produce a localized mechanical
response of the skull and brain system (local skull bending and contact injury). This
locahzation is a consequence of the inabihty of the loaded region to effectively disperse
the input energy that is seen at the helmet/head interface after the helmet has absorbed a
maximum amount of energy through defonnation. This localized load phenomenon is
particularly important in bicycle helmet design since one of the primary functions of a
bicycle helmet is to distribute the impact forces over as wide a range as possible.
Early work by Aldman (1984) proposed a test method for evaluating a helmet’s abihty to
distribute load by placing a helmet fitted with a load cell on the surface of a
hemispherical head model and conducting impact tests using a 25mm radius spherical
impactor. The mass of the impactor was 5kg and was released from a drop height of .75
m while the load over a 1 cm^ area was measured on the surface of the headform. Based
on the work of Nahum et al. (1968), the authors proposed a lower force limit of 1 kN for
the frontal part of the head and a .4 kN load limit for the temporal region of the head as
measured over the 1 cm’ area. This would equate to pressure values of 10 mPA for the
frontal region and 4 mPA for the temporal region. These values were based on the
cadaver work of Nahum et al. (1968) and were thought to be acceptable threshold Umits
for skull fracture. To date, no proposed threshold hmits for brain injury due to localized
loading exist.
Using the identical impact anvils and test headforms as used by Aldman (1984), Long et
al. (1989) incorporated Fuji Film as a means of monitoring peak loads transmitted
through the helmet to the headform under impact situations. Fuji Film is a photographic
emulsion system that utilizes two separate layers of film, an active layer and a transfer
layer. These two layers are sandwiched together and the system is placed onto the
headform prior to attachment of the helmet. When the impact occurs, force transmission
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
causes contact between the two layers which in turn causes activation of pigment dyes on
the active layer proportional to the amount of force applied. As the force increases, the
intensity of the color will increase. This intensity can then be calibrated using
photodensitometry against a known reference load in order that a peak load and load
area can be determined.
Using a 3 kg impactor and three difference energy levels (29.4 J, 50.0 J, 73.6 J) and several
different impact anvils, Long et al. (1989) evaluated the load distribution capabihties of a
series of contemporary bicycle helmets. They noted that two of the helmet models
possessed large ventilation openings and exhibited very high loads over the entire area of
liner contact as measured by the Fuji Film. This was significant in that although the
helmets did induce high local loads relative to similar helmet models, they still passed the
contemporary standards to which they had been designed. An analysis of load
distribution versus anvil type revealed that skull loading was more dependent upon the
contact area of the helmet with the headform than on the construction of the helmet. As
expected, the more aggressive anvils (i.e. smaller radii) produced lower peak headform
acceleration values with greater penetration into the helmet as compared to the flat anvil
tests which resulted in overall higher peak headform acceleration values. When compared
to the skull loads, the authors found that there was a good correlation between skull load
and peak headform acceleration values for the flat anvil test series. However, they found
that the more aggressive anvil tests resulted in high skull loads with low peak headform
acceleration values. It was suggested that these results illustrate the shortcomings of an
acceleration based injury criteria as a means to assess a helmets abihty to limit skull
loading under certain impact situations.
Based upon the results of the impact tests described above, the authors proposed an
injury tolerance level of 25 kgF/ cm^ (245 N / cm^ as a pass fail criteria for measuring
local loads on the test headform when impacted at a 29.4 J energy level using an edge
anvil. This proposed test criteria was then modified such that it could be used as a
method for determining load distribution properties of bicycle helmets tested to the
Austrahan and New Zealand Bicycle Helmet Standards (AS/NZS 2512.9:1996). The
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
adapted test involves placing the helmet on a hemispherical test headform which is fitted
with a load cell capable of measuring force over a 100 cm^ circular area. A 3 kg mass is
then dropped in a guided freefall onto the helmet assembly from a height of 1 meter and
the loading as measured by the force transducer is not allowed to exceed SOON.
Although the current systems for measurement of localized loading appear to provide
some insight into localized skull loading during helmet impacts, both systems have
fundamental limitations. The single load cell only provides information regarding load
distribution over a very small area and little is known about the total distribution
pattern. Furthermore, due to the complex contour matching between the helmet and the
headform test device, it may be possible that the actual peak force may not occur at the
centroid of the impact site but rather, it may actually occur in an adjacent region which
is not measured by the load cell.
A clear disadvantage of the Fuji Film measurement system is that it provides only peak
loading information and no knowledge can be gained with regard to when that load was
actually applied to the head. Given the fact that the system being measured is not rigid,
the assumption that peak local force occurs at the moment of peak headform acceleration
may not be correct. Furthermore, since only peak values are recorded by the Fuji Film
over the entire headform surface, the timing of these localized peak values is not known.
It is likely that the manner in which these localized forces are applied to the head
temporally is an important factor in determining the relative risk of skull fracture, skull
bending, and brain injury. Similarly, the local force transmission pathways likely play a
role in the ability to predict the potential for head injury due to high local contact forces
in certain regions. Lastly, helmet liners often experience fracturing as a means of
absorbing additional impact energy and the localized loading patterns during helmet
fracturing and the subsequent changes in the risk of head injury due to high local contact
forces is not known at this time.
In order to advance the understanding of load distribution during helmeted impacts,
these limitations must be overcome. An appropriate measurement system must have the
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ability to collect information over a large area and it must be capable of providing
temporal information over the entire duration of the impact sequence. In this way, it
may be possible to better understand the complex behavior of the helmet and headform
system under impact loading and to possibly gain insight into the even more complex
behavior of the helmet, skull and brain system.
Pressure Measurement Devices
The methodological problems encountered when attempting to monitor local loads and
pressures have been well defined previously by researchers conducting footwear research
(Cavanagh and Ae, 1980). The problems faced in this area of research are very similar to
the ones found in helmet research. Namely, the area to be monitored is very small and
contains complex curvatures; the interface between the two systems (shoe and foot versus
helmet and headform) requires a very thin transducer system capable of collecting a large
amount of information and lastly, the system must be rehable and repeatable in order to
conduct comparative scientific analyses.
At the present time, there are two approaches to the collection of local contact forces in
footwear. The first approach is the use of discrete miniature load cells which are apphed
to the shoe insole to monitor discrete local forces (Hennig and Milani, 1995). Although
these transducers provide useful discrete information, assumptions must be made with
respect to the total load distribution pattern within the shoe complex. As well, the low
shear resistance of these transducers makes them susceptible to movement and
consequently may alter their output over repeated trials (Cavanagh and Ae, 1991).
The second methodological approach is to collect local contact forces using an insole
measurement system. Currently, there are two systems which are commercially
available. One system is manufactured by Novel (Novel Gmbh, Munich Germany) and
is known as the EMED system while a second system is manufactured by Tekscan
(Tekscan Inc., Boston, MA). Both systems utilize thin flexible insole systems which are
capable of monitoring local forces over a very large area (the insole of a shoe).
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The EMED system applies capacitance transducer technology as a means of measuring
local forces at the shoe/foot interface. The system is housed within a flexible foam
substrate; therefore, it has the capacity to conform over irregular surfaces. Within this
foam substrate, a system of 4000 individual sensors are located in a single transducer the
size of a shoe insole. As force is applied to each sensor, there is a commensurate change in
the capacitance of that sensor and this change in capacitance is then converted to a
voltage and calibrated to a specific force or pressure value. Software then displays the
overall force distribution pattern.
The Tekscan system applies resistive ink technology to monitor local forces at the
shoe/foot interface. A matrix of sensing strips are laid perpendicular to each other over
the entire surface of the transducer. At the intersection of each of the “sensing tracks”,
there exists a sensor capable of measuring the forces seen at that location. The sensor is
constructed with two layers of substrate (polyester film) with a layer of silver conductive
material applied to each layer of film. A proprietary pressure sensitive material is then
applied over the sensing area and an adhesive is then apphed outside the sensing area to
join the two layers of substrate. As force is exerted on the sensor, the two layers of
pressure sensitive material compress together and cause a change in resistance output. As
the force increases, the resistance decreases and as the force decreases the resistance
increases. A small op-amp circuit is then used to convert the resistance to a voltage
output which is measured by the A /D system. Using a known cafibration constant and
available software, this output voltage can then be converted to a load or pressure value
and expressed in terms of the overall distribution pattern.
Both of these systems are currently in use in clinical and research applications. Recent
work by McPoil et al. (1995) has suggested that the system developed by Tekscan Inc. has
some serious limitations. Using both bench based tests and human subject tests, the
Tekscan system was found to have a large amount of error which was not constant over
changing applied load levels. The reliability of the Tekscan transducer was also found to
be less than the rehabifity of the EMED transducer when tested with human subjects
performing normal walking tasks in a laboratory environment. It must be mentioned
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that this research was largely funded by the manufacturers of the EMED system;
therefore, the findings should be viewed as somewhat biased. It would be advisable that
this research be corroborated by an independent research facility.
In summary, the current body of knowledge indicates that head injuries can occur due to
both contact injury as well as inertial movement of the brain relative to the other
structures of the brain and skull. Although the technology exists to monitor the inertial
effects of impact forces being applied to the skull, very little is known regarding the
effects of contact forces applied to the skull. In particular, very httle is known regarding
the magnitude of the local contact forces and the manner in which these contact forces
are distributed over the skull. As a result, it becomes difficult to clearly characterize the
subsequent local response of the skull and brain and truly understand head injury as a
result of relative displacement between local structures. Furthermore, the lack of
technology to monitor local contact forces also makes it difficult to understand how head
protection devices function under impact conditions.
If such a system could be developed, then it would be possible to provide some insight
into how contact forces are distributed over the surface of the head and this could be used
to evaluate bicycle helmet designs in order to understand how they could be made to be
more effective at reducing local contact forces. The output from such a transducer also
may be useful for the future development of finite element models which monitor local
contact stresses.
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapt e r 3
M ETHODOLOGY
Experimental Instrumentation
All impacts were conducted with an ISO Size E test headform which was fitted with a
custom load sensing transducer designed and manufactured exclusively for this project by
Force Imaging Technologies of Chicago, Illinois. Prior to any actual testing, it was
necessary to design and develop control software and signal conditioning electronics for a
64 channel A /D collection system. Furthermore, it was necessary to evaluate the
performance of this load sensor prior to any actual impact testing in order to confirm its
operational characteristics. Preliminary evaluation of the transducer was accomphshed by
testing a single sensor which was fabricated to the same force range as the test transducer.
The amplification circuit and the recording instrumentation were identical to that used
during the actual test procedures.
Development of Trigger and Signal Conditioning Circuitry
In order to properly supply voltage to the UniForce sensor immediately prior to impact,
a custom designed circuit was developed to supply voltage to the UniForce sensor and to
amplify the signal for collection by the analog to digital collection system. Figure 1
illustrates the logic circuit for the circuit design. The limiting factor for the sensor was
the fact that the maximum duration over which the supply voltage could be suppUed to
the sensor was 100 ms. If the supply voltage duration was greater than 100 ms, then the
sensor would be damaged. As a result of this limitation, a system was required such that
the driving voltage signal would be initiated immediately prior to the impact sequence.
At all other times, the voltage signal supplied to the sensor had to remain at 0 volts. This
was accomphshed by the use of a single pole double throw (SPDT) relay which was
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
anchored to the system ground; therefore, until an appropriate trigger signal was received
at the relay, the voltage supply to the sensor remained at zero volts.
External Synch Putse
From GHI VS300
• SVoltOnvmg
Voltage
Grounded Signal
(OVolti)
SPOT Relay
(Normaly Grounded)
Generated By
Lab View V I
Am BOARD
Data Colection
UNIFORCE
SENSOR
Synch Trigger
Figure 1. Trigger Logic For UniForce Sensor
The trigger signal to the relay originated from the synchronization pulse of a GHI VS300
velocity gate which was located a sufficient distance above the impact surface that the
actual impact event occurred after the sensor voltage had been applied for approximately
10 to 15 ms. The signal originating from the GHI velocimeter was fed into a sensing gate
located on a National Instruments AT-MIO-64E A /D board which sensed the rising
synchronization pulse. Once the synchronization pulse was detected, a custom Lab View
program generated a square wave pulse with an amplitude of 5.0 volts and a duration of
80 ms. The generated square wave pulse was then fed to one side of the input pin on the
SPDT relay. The receipt of this trigger signal by the relay then actuated the relay switch
which caused a -5 volt signal to be directed to the sensor for the duration of the generated
square wave pulse (80 ms). The -5 volt signal was generated by feeding a -15 volt signal
from an external power supply through a -5 volt voltage regulator (Motorola 7805).
Upon termination of the computer generated square wave pulse, the relay switch
returned back to its grounded position.
Following activation of the relay switch, the -5 volt signal was fed into one side of the
UniForce sensor immediately prior to impact. Upon impact, the active sensor would
provide a change in resistance output which was proportional to the amount of force
which was being apphed to the sensor. The output from the sensor was then fed into an
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
op-amp circuit fTexas Instruments Quad Op-amp TL084CN) which converted the
change in resistance to a change in voltage and then subsequently amplified the signal for
data collection by the A /D board. Figure 2 illustrates the circuit that was used in this
application.
UniForce Sensor
■ 5 Volt Driving
V oltage
100 pF
20K-Ohms
TL084CN Quad Op Amp
O utput
V oltage
Figure 2: UniForce Electrical Circuit
Transducer Response Evaluation
The phasic response characteristics of the UniForce sensor were evaluated by matching
the response of the sensor against the response characteristics of an Endevco type 2215
piezoelectric accelerometer which was located at the center of gravity of a test headform
mounted on a twin wire drop frame apparatus. The headform was rotated to expose the
right side of the test headform to the Modular Elastomer Programmer (ME?) contact
surface. Once the contact point was identified, the sensor was affixed to the headform
using two sided tape. Visual inspection confirmed that the impact site located on the
external surface of the headform was coincident with the sensitive axis of the
accelerometer.
The headform dropframe assembly was raised to a free fall height of 1.0 m and released
using a pneumatic actuating system. A second GHI VS300 timing gate was located
immediately above the impact surface and the synchronization pulse from this unit was
used to measure the impact velocity immediately prior to contact.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The accelerometer data and the force data were captured using a custom Lab View
program which monitored the accelerometer channel for a rising slope and an increase in
the accelerometer voltage above .50 volts. Once this trigger level had been attained, the
program collected 500 samples at a rate of 10,000 Hz. The program was written in such a
manner that it was capable of collecting 100 samples prior to the trigger; therefore, pre
impact information was also captured.
A total of thirty trials were captured for this evaluation. Upon completion of all testing,
the output from the accelerometer and the output from the sensor were fed into SPSS
(SPSS Inc, Chicago, Illinois) which computed the cross correlation r^ of the two signals.
Transducer Linearity Evaluation
The linearity of the UniForce sensor was evaluated by determining the voltage output of
the sensor relative to a known external force. The known external force was applied to
the active sensing area of sensor using a small force applicator that was attached to an
arbor press. The arbor press was fitted with a rack and pinion assembly which allowed
for the application of varying amounts of force depending upon how much force was
applied to the arm assembly. The diameter of the force applicator was identical to the
diameter of the sensing area of the sensor (6.35 mm, .25 inch). A Sensotec Model 31
compression/tension load cell was placed in series with the arbor press to monitor the
compressive forces applied to the sensor. Prior to data collection, the load cell was placed
horizontally on a flat surface and the output was set to zero volts.
As there was no impact event to monitor for the driving voltage trigger, a timing circuit
was used to capture calibration data trials; consequently, the testing sequence was such
that the program was activated to collect data, a force was manually applied to the sensor
while a countdown began to initiate trigger of the driving voltage for the sensor. Upon
activation of the sensor, a custom Lab View program was used to capture the output from
the load cell and the output from the sensor at a rate of 10,000 H z for a duration of 150
ms (1500 samples). The program was designed to continuously monitor the voltage drive
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
side of the sensor until a -5 volt signal was observed (indicating initiation of the voltage
supply to the sensor). Upon observation of the -5 volts signal the program was also
designed to capture 250 samples prior to the trigger event; therefore, the entire
calibration/voltage supply sequence was collected.
Upon completion of the data capture sequence, the load cell voltage was converted to a
force value in Newtons using the calibration factor supplied by the manufacturer. An
iterative routine was then used to find that point in the data collection sequence wherein
the driving voltage went to -5 volts. The data frame in which this event occurred was
identified and the program then proceeded to count an additional 25 frames beyond this
point. This was done to ensure that there was no possibility of any significant voltage
spikes or oscillations which could affect interpretation of the sensor output. The output
from the load cell and the output from the sensor were then averaged over the next 750
points. The output from these mathematical computations was then stored to a data file
for later use in the sensor cafibration procedures.
This procedure was repeated over the entire load range of the sensor so that a linearity
analysis could be conducted over the complete range of the sensor. The output from both
the load cell and the sensor were then input into a MiniTab spreadsheet (MiniTab Inc.,
State College, PA) and evaluated using a linear regression analysis. This statistical
procedure provided the Pearson product moment correlation as well as the information
regarding the residuals that remain as a result of the linearizing procedure. Following a
review of the output from the linear regression analysis, additional curve fit procedures
were utilized using Lab View software. These procedures are described below.
Transducer Hysteresis Evaluation
Hysteresis evaluation was conducted by using the same cafibration equipment as noted
above. A total of 10 sequential loading sequences were collected for this portion of the
analysis. Five of the trials were incrementally increasing in terms of the amount of force
applied to the sensor (loading trials) while five of the trails were incrementally decreasing
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in terms of the amount of force applied to the sensor (unloading trials). In all trials, the
force applied to the sensor was controlled by adjusting the amount of force applied to the
armature of the arbor press.
The average load observed over the duration of the applied voltage was measured using
the LabView software described above. At each increment in load, the average sensor
output values during the loading phase were compared to the average sensor output
values during the unloading phase according to the equation:
I (OiOput During Loading-Output During Unloacüng) ,
Sensor Hysteresis = | ----------------------------- | x 100
This value represented the transducer hysteresis at each individual load increment.
Transducer Repeatability Evaluation
The abihty of the sensor to provide repeatable output was evaluated by calibrating the
same sensor on three separate days using the same load. A total of ten trials were
collected on each test day and the average load applied and the average sensor output
were stored to file. Upon completion of data collection, these values were subjected to a
dependent groups analysis comparing trimmed means and Yuen’s method to test the
hypothesis that the trimmed means were equal. A trimmed means analysis was chosen
over a classical mean analysis based on the abihty of the trimmed mean to handle outliers
within the data. The procedure involves ranking the test scores and removing 20% of the
largest and 20% of the smallest values and recalculating the mean. The value obtained is
referred to as a 20% trimmed mean. Upon completion of the trimmed mean calculations,
the dependent groups analysis was conducted using the SPLUS procedure pairdepb
(Wilcox, 1997).
Transducer Response Based on Sensor Area Coverage
Given the fact that the sensing area of each individual sensor was relatively small (31.67
mm^, it became important to understand the effect that partial loading had upon the
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
response characteristics of the transducer. In order to accurately monitor this, a known
load (nominally 225 N) was apphed over one quarter, one half, three quarters and the
whole sensor sequentially. The corresponding transducer output was monitored using
the same programs as described above. The voltage output of the sensor was then
expressed as a ratio of the magnitude of the load and the sensor output (Newtons/volt). It
was expected that this information would provide some insight into the response of the
sensors when the test helmet made contact with only a small portion of the sensor.
Design of 64 Channel Transducer
Selection of Force Measurement Range
Upon completion of the preliminary evaluation tests, a final sensor design pattern was
developed for fabrication by Force Imaging Technologies of Chicago Illinois. The
pattern was designed specifically for this project and was designed to specifically conform
to the contours of an ISO size E test headform. The 65 channel transducer pattern is
illustrated in Figure 3. It consisted of 65 individual sensors located on an increasing radii
to a maximum of radius of 63.5 mm (2.5 inches). The sensor has symmetrical sensors
located in each quadrant; therefore, it was possible to measure load distribution
symmetry during the impact sequence.
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SECI10N3
SECHON2
• #
SECTON4
SECnON 1
SECTIONS
SECTIONS
SECTIONS
SECTION?
Figure 3. Transducer Design (X denotes inactive sensor)
Two specific transducer ranges were produced for this testing program. Ten low range
sensors (0 to 220 N) were fabricated to monitor local headform contact forces during
“typical” helmet impacts against flat and hemispherical anvils. This range was selected
based on data obtained from individual 220 N and 440 N production sensors which were
mounted to the lateral surface of the ISO size E headform. A group of commercially
available bicycle helmets were then placed on the test headform and impacted against
both flat and hemispherical impact anvils from a variety of drop heights. As well, several
samples of bicycle helmets with very low density expanded polystyrene liners were also
impacted against the flat and hemispherical impact anvils using the same instrumentation.
The results from this pilot work indicated that under certain circumstances, very high
local forces (i.e. in excess of the sensing range of the 440 N transducer) could be observed
by the transducers. Consequently, ten high range sensors (0 to 2200 N) were fabricated to
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
monitor local headform impact forces during tests which were known to cause high
levels of localized loading (e.g. low density helmet liner with a 2.0 m drop height onto a
hemispherical anvil). Given a 10 volt maximum measurable output by the A /D system,
the resolution of these high range sensors was considered poor relative to the lower range
sensors (22 N /volt versus 220 N/volt). Consequently, these high ranges sensors were
only used for a limited number of tests.
Development of Printed Circuit Board For Signal Conditioning
Upon completion of the preliminary bicycle helmet impacts and development of the
sensor pattern, a custom electronic circuit was built to provide driving voltage to each
individual sensor, amplify the signal of each sensor and interface the output signal with
the AT-MIO-64E A /D board. This custom circuit was designed using Protel shareware
printed circuit board software which allowed for the design of a unique 64 channel
system for this application.
The printed circuit board was fabricated by transferring the two sides of the printed
circuit board diagram (upper and lower) to two mylar screens (Press n’Peel Blue,
Techniks, Inc. New Jersey). These screens were then each ironed onto one side of a two
sided copper sheet using locator holes to ensure that the upper and lower circuit patterns
would match. The printed circuit board was then placed in a solution of ferric chloride
for 60 minutes in order to etch the printed circuit board. Upon completion of the
etching process, the remaining ferric chloride was rinsed away, leaving the conductive
tracks which were protected by the mylar screen.
The mylar screening material was then removed using steel wool and holes were drilled
into the printed circuit board to allow for placement of all circuit components. Solder
connections were then made between the upper and lower circuits on the printed circuit
board. Once this process was completed, all components were placed on the printed
circuit board and soldered in place. Continuity checks were made throughout the process
in order to prevent any possibility of excessive noise or signal loss.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Software Development for Data Collection
Prier to testing the 64 channel transducers, a custom software program was developed
using LabView. Using the logic circuit illustrated in Figure 1, the system was expanded
to handle 64 channels of data. This included one accelerometer or force channel and 63
channels of data from the UniForce transducer. The system control software was written
in an identical manner as described above. The software monitored the accelerometer
channel for a rising slope which went above a threshold level of .40 volts (roughly
equivalent to 40 g). Upon receipt of this trigger, the software then collected 300 samples
per channel at a rate of 7500 Hz for each channel. The software also captured 100
pretrigger samples to result in a total of 400 data points for each channel at a rate of 7500
Hz per channel. The captured data was then displayed on a single graph which illustrated
all channels. A second custom LabView program was then written to display the voltage
output for each individual channel.
Selection of Helmet Test Samples
The selection of the helmet model for this test series was based on its current availability
and the abihty of the manufacturer to produce identical units in four different expanded
polystyrene densities. The model selected was a common generic bicycle helmet which is
currently being mass produced for sale in North America at a retail price of under $30.
The helmet consisted of an expanded polystyrene (EPS) liner which was fitted with an
internal support ring and a 0.5 mm outer shell fabricated from PETG (glycol modified
polyethylene terephthalate). The outer shell was placed on the upper surface of the liner
and secured in place using 10 mm wide tape. Due to cost considerations and production
restrictions, the helmets were not fitted with comfort pads or retention systems. It was
felt that the lack of fit pads would not have a significant influence upon the impact
behavior of the helmets tested.
A total of 50 helmets at each of four different densities were fabricated exclusively for
this project and selected for impact testing and load distribution measurement. The
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
density range was from 32 kg/m^ to 128 kg/m^ in 32 kg/m^ increments (i.e. 32 kg/m^, 64
kg/m^, 96 kg/m^, 128 kg/m^. This range of densities was selected in order to explore the
effect that bicycle helmet liner density had upon the load distribution characteristics of a
given model of bicycle helmet. Research by the author has found that current production
bicycle helmet liner densities range from 64 kg/m^ to as high 112 kg/m^; therefore, it was
felt that this range of helmet liner densities could adequately represent the continuum of
possible bicycle helmet liner densities which could be fabricated from a single mold.
Each helmet was produced from one of twelve individual helmet cavities. Upon receipt
of the helmet test samples, each helmet was weighed using an electronic scale. This value
was then coupled with the average volume of the twelve cavities used to fabricate these
helmets in order to determine an exact density value for each helmet used in this study.
Development of Helmet Test Matrix
Two separate impact locations were selected for load distribution analysis. These
locations were the left front of the headform and the right side of the headform. These
locations were selected based on the research of Smith et al. (1994) and McIntosh and
Dowdell (1992) which suggested that these locations represented impact sites which were
frequently impacted in bicycle accidents. In addition to this, the selection of the front
impact location allowed for the analysis of the potential effect that ventilation holes
would have on the load distribution characteristics of a bicycle helmet impact.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
f
Figure 4. Left Front Impact Location (circle denotes
impact location)
The left front impact location was located 4.5 cm anterior of the headform centerline at
the level of the reference plane and 6.0 cm superior to the reference plane. The right side
impact location was located 5.0 cm posterior of the headform centerline at the level of
the reference plane and 4.5 cm superior to reference plane at that point. Figure 4 and
Figure 5 illustrate the left front and right side impact locations as seen on the helmet.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5. Right Side Impact Location (circle denotes
impact location)
A flat anvil and a hemispherical anvil were selected as the two impact test surfaces to be
evaluated as part of this study. The flat anvil consisted of a 25.4 mm thick steel plate with
a 12.7 cm diameter while the hemispherical anvil consisted of a 25 mm radius steel dome
which was moimted onto a 25.4 mm thick plate with a 12.7 cm diameter. These surfaces
were selected based on the fact that they represented standard test anvils used as part of
the impact attenuation evaluation of bicycle helmets currently tested to one or more of
the contemporary N orth American bicycle helmet standards. These impact surfaces also
represented the two extremes of impacting surfaces that are present in the real world of
bicycle accidents. In particular, the hemispherical anvil, with its sharp radius, presented
an aggressive impact surface over a very small contact area. It was hypothesized that this
anvil would likely result in higher local peak forces when compared to the local peak
forces observed during flat anvil impact tests.
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to evaluate the load distribution characteristics of the different helmet liner
densities, four drop heights were chosen as part of the test matrix. The drop heights
ranged from 50 cm to 2 m in 50 cm increments. It was felt that the impact velocities and
impact energies resulting from this range of drop heights encompassed both the impact
energy levels of contemporary bicycle helmet standards (ASTM F1446, Snell B90, CPSC)
as well as the impact energy levels observed during rephcation studies (Smith et al., 1994).
Transducer Preparation and Calibration
Prior to any impact testing, each UniForce transducer was cahbrated prior to application
to the test headform. This cahbration procedure involved pre-conditioning of the
transducer by successively applying a known load to the transducer a series of ten times.
According to the manufacturer, this pre-conditioning procedure conditioned each
individual sensor in such a manner as to ensure a reliable and repeatable force
measurement. Upon completion of these conditioning trials, each sensor was sequentially
loaded ten times throughout the expected loading range in order to generate calibration
constants for each sensor.
In order to ensure that the transducer remained in the appropriate position on the
headform throughout the entire test sequence, it was necessary to secure the sensor to the
headform with two sided tape. The two sided tape selected for this project was model
9482PC fabricated by 3M of Minneapolis, Mirmesota. This particular tape was selected
based on the fact that it was suitable for the adhesion of polyethylene to magnesium, had
a very high resistance to shear and was also very thin relative to other products currently
available (0.5 mm thick).
Pilot research on individual sensors strongly suggested that there would be a significant
difference in the output from an individual sensor depending upon whether or not tape
was applied to the underside of the sensor. Further investigation also found a significant
difference in sensor cahbration values depending upon whether the tape backing was
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
removed from sensor underside during the calibration testing. Given the fact that the
transducer was affixed to the headform during testing, this preliminary finding precluded
calibration of the sensor with the two sided tape and the adhesive backing.
It was decided that all calibrations would be conducted using the same configuration that
would be used for the test series (i.e. sensor with two sided tape, no backing). Given the
number of cahbration tests which were to be conducted (20 per sensor, 1260 total per
transducer) it was felt that the two sided tape would not be suitable for actual testing
following this cafibration procedure given the fact that each sensor cafibration involved
placement and removal of the two sided tape beneath the load actuator assembly. This
repetitive manipulation of the tape would likely serve to seriously degrade the adhesive
powers of the tape. Furthermore, the force required to remove the tape from the platen
of the arbor press and the extreme angles required for sensor removal would likely
caused microcracks within the surface of the sensor. These microcracks would then most
certainly affect the output voltage of the individual sensor channels; thereby invalidating
the generated cafibration constants since the sensor damage would not occur until after
completion of the cafibration procedures.
In order to avoid this potential problem, the test system configuration was simulated
during cafibration by placing a single layer of two sided tape directly to the platen of the
arbor press and coating the two sided tape with a layer of baby powder. The baby
powder reduced the adhesive strength of the two sided tape without affecting the
compliance of the two sided tape. In this way, the sensor could be placed on the platen of
the arbor press directly beneath the load applicator, conditioned and calibrated and then
removed without damage to the sensor. The platen was rotated after each sensor
cafibration series in order to ensure that there was a new sample of tape located directly
beneath each sensor prior to each cafibration procedure.
As mentioned earlier, a total of 20 tests were performed as part of the cafibration
procedure for each sensor. The first ten tests of the sensor involved application of a
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
known load to the sensor in order to condition the sensor. This known load was
nominally 225 N and was applied by hanging a mass on the end of the actuator arm of
the arbor press. Once the mass was placed on the armature, the data collection system
was triggered using the calibration software described previously (see Figure 1). Once the
data were collected, the armature was then raised and the procedure was repeated an
additional nine times. For each trial (including pre-conditioning trials), the mean force
from the load cell and mean sensor voltage from the UniForce sensor were computed
along with the standard deviation. These values were then stored in a unique summary
file for each particular sensor.
Upon completion of the initial 10 conditioning tests, each sensor was cahbrated over the
operational range of the sensor (0 to 10 volts) by applying a series of loads to the sensor
via the arbor press armature assembly. The loads apphed to the sensor were
incrementally varied by manually applying an increasing amount of force to the end of
the arbor press armature. Once the desired amount of force had been reached, the data
acquisition system was triggered and the output from the load cell was captured along
with the output from the UniForce sensor.
Once again the mean force output value from the load cell and the mean voltage output
from the UniForce sensor were determined using the cahbration software described
above. The mean output values and the standard deviations about the mean for both
sensors were stored into the same summary file that was used during the conditioning
test series. Once ah 20 tests were completed for a given sensor, the entire transducer was
raised off of the platen surface and the next sensor location was selected for cahbration.
Preparation of the Test Headform
The cahbration procedures were sequentiaUy repeated until all sensors had been
cahbrated for a given transducer. Upon completion of ah cahbration tests, the sensor was
cleaned of excess baby powder and a complete layer of two sided tape was apphed to the
underside of the transducer. Excess tape was trimmed using scissors and an exact-o knife.
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The centroid of the transducer was then located directly over the desired impact site
(either left front or right side) and carefully oriented such that sections 4 and 5 were
located at the top of the headform and sections 1 and 8 were located at the bottom of the
test headform (see Figure 3). This particular orientation allowed for the best routing of
the lead wires from each sensor. Figure 6 illustrates the sensor affixed in the left front
impact location.
Figure 6. Sensor Mounted At Left Front Impact Location
Special care was taken to carefully affix the sensor to the test headform without creasing
the transducer or any sensor within the transducer. This was accomplished by
sequentially affixing each section of the transducer over the curved headform surface and
carefully smoothing out each particular section to remove air bubbles trapped beneath
the transducer and to remove any creases that may result from the complex curve fit.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Once the transducer was affixed to the test headform, the headform was placed on the
twin wire drop frame using a ball and clamp assembly. A sample helmet with the impact
location marked on the external surface was then placed upon the headform. The entire
helmet/headform system was then rotated and oriented such that the desired impact
location represented the center of the impact against the desired test anvil. The headform
was then secured in place using the ball and clamp assembly.
A uniaxial piezoelectric accelerometer (Endevco Model 7701-A-50) was mounted at the
center of gravity of the entire headform system. The signal from the accelerometer was
carried away from the drop frame to a signal conditioner which amplified and calibrated
the signal which was then input into the AT-MIO-64E A /D board.
A series of 8, 12 conductor cables were required to transmit the transducer signals from
the sensors to the signal conditioning circuity. Each junction between the transducer
section and the 12 conductor cable was secured using duct tape and the entire cable
assembly was secured to the drop frame using tie wraps. The routing path of each
particular section of the transducer was selected such that it would not in any way affect
the energy absorbing characteristics of the liner nor would it prohibit proper placement
of the helmet on the test headform. All cables were secured with duct tape to the surface
of the headform. This limited any potential noise effects due to sudden cable movement
or connector failure. The opposite end of the cables were then secured to inputs of the
signal conditioning box for the transducer using standard DB15 connectors.
A total of six different transducers were used to conduct the impact testing. Four low
range (0 to 220 N) sensors were used (two per impact location) and two high range
sensors (0 to 2200 N) were used (one at each impact location) throughout the entire test
series. The use of several different transducers was based on preliminary work which
indicated that over time the signal output from the transducers would degrade and the
transducers themselves would become damaged.
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Impact Test Procedures
Once the instrumentation had been applied to the test headform, an appropriate test
helmet was selected at random from all available helmets of a given density. The helmet
was then secured to the test headform by ahgning the edge of the helmet with a known
reference mark on the test headform. This ensured that each helmet was placed in an
identical manner on the test headform. Typically the ahgnment was performed using the
test headform reference plane as a guide. Each helmet was secured to the test headform
using a single piece of pipe wrap tape. This ensured that the helmet remained in place
throughout the entire impact sequence.
In some instances, a dry transfer ink was apphed to the upper surface of each sensor using
a dry transfer marking pen prior to mounting the test headform. This allowed for a
transfer of ink onto the internal surface of the helmet as a result of a given impact. This
transfer information was then used as a guide to assist in the interpretation of the
transducer output and the distribution of the impact forces relative to the internal surface
of the helmet.
Once the helmet was secured to the test headform, the entire drop frame assembly was
raised to the appropriate drop height and released using a manually driven pneumatic
release mechanism. The drop frame then proceeded to fall in a guided free fall and
contacted the appropriate test anvil. Immediately prior to contacting the impact surface,
a GHI VS300 timing gate was used to monitor impact speed immediately prior to
impact. The timing gate consisted of an infrared single beam system which was activated
by a 25 mm flag located on the drop frame assembly. When the flag passed through the
sensing head of the GHI VS300, a high speed clock recorded the duration over which the
infrared beam was broken. Given the known width of the flag, it was then possible to
determine the velocity immediately prior to the impact against the test anvil.
As the helmet contacted the test anvil, a LabView program monitored the accelerometer
data which was being passed through the memory buffer. When the threshold value of 40
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
g was exceeded, the program began to capture 300 data points across all 64 data channels
at a rate of 7500 Hz per channel. The program was also designed to capture 100 data
points prior to the threshold value across all 64 data channels. The program then
displayed all 64 channels on the monitor and proceeded to write the raw data file to the
computer hard disk using a unique name for each trial.
Following the impact, the test helmet was removed and the UniForce transducer was
inspected for any signs of direct damage. Information from the 64 channel display was
reviewed to identify any potential failures in the electronic circuitry (e.g. loose
connection or signal shorting). Any necessary modifications or adjustments to the cable
system were made at this time. Following this system check, the next test helmet was
selected for testing and the entire procedure was repeated.
Post Test Procedures
Upon completion of a given portion of the test matrix (one impact location and one
impact anvil), the UniForce transducer was removed from the test headform using a
procedure which was designed to minimize additional strain on the sensors prior to the
post test cahbration procedures. This procedure involved heating the headform and the
sensor unit at a temperature of 50 degrees Celsius for a period of one hour. The
application of heat to the headform reduced the adhesive characteristics of the two sided
tape and allowed for easy removal of the transducer from the headform. Baby powder
was then apphed to the underside of the transducer in order to prevent any potential
contact between transducer sections or between the transducer and the headform.
Once the transducer was removed from the test headform, each individual sensor was
then subjected to the same cahbration procedures described previously. A visual
comparison was then made between the pre and post test cahbration values and any
possible sources of discrepancy were identified.
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Data Reduction
Upon completion of all testing, each channel of the raw data file was filtered using a
custom LabView program which utilized a 4“ * * order low pass Butterworth filter with a
cutoff frequency of 1000 Hz. This particular cutoff frequency was selected because it
conformed to the requirements of SAE J211 Channel Class 1000. This filter specification
is commonly used in bicycle helmet test standards throughout North America.
In order to properly reduce the large amount of data that had been collected, a custom
Visual Basic program was developed to read in the filtered data file, compute the
accelerometer based measures, cahbrate each individual force sensor channel and to
provide peak sensor force and peak sensor pressure values. Details of these computations
are presented below.
Accelerometer Data Reduction
As indicated above, the output from the accelerometer was filtered using a digital low
pass 4 '* ' order Butterworth filter with a cutoff frequency of 1000. Following the filtering
process, any bias in the signal was removed and the peak headform acceleration was
determined. The duration of each individual accelerometer trace was determined by
locating the peak headform acceleration value and locating those points before and after
the peak where the acceleration value dropped below a threshold value of 5 g. The
difference between these two times was considered to be the impact duration.
In addition to the peak headform acceleration value, a Head Injury Criteria (HIC) value
was determined for each impact test. This value was determined using the equation:
Head Injury Criteria (HIC) =
(^2 - f | )
~ ^ ] a d t
where a represents the acceleration and t^and /, are the time limits where HIC is a
maximum value.
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The Gadd Severity Index (GSI) was also calculated from the acceleration data using
equation:
f
Gadd Severity Index (GSI) = ja^^dt,
0
where a represents the acceleration pulse and t represents the pulse duration. Although
the GSI data was computed, it was not analyzed as part of this study.
Upon completion of the computations, all data for each trial was stored in a summary
file which included the trial filename, peak acceleration, impact duration, peak force,
HIC and GSI. An additional file (*.acc) was stored for each trial which included the
acceleration values over time as well as the product of the acceleration times the mass of
the test headform (total force).
Sensor Force Data Reduction
Prior to any data reduction, the raw data for each trial was analyzed and each data
channel was coded according to one of four categories. The channel was identified as
being either inactive (i.e. having a response output below 100 mv), active (signal response
greater than 100 mv), active but noisy, and active with excessive noise or signal shorting.
This coding information was necessary to identify force distribution patterns as well as to
identify true peak sensor force values, as opposed to peak sensor force values from faulty
data channels.
Prior to sensor calibration, the bias from each filtered channel was removed by visually
inspecting the data trace and identifying a portion of the trace prior to impact that best
represented the bias state of the transducer prior to impact. Given the fact that there
were forces being applied to the transducer as a result of the helmet being affixed to the
test headform (i.e., pre-load), the effect of any pre-load was also removed as part of the
bias removal process. Consequently, the resulting data represented the net change in local
contact force as seen by each individual sensor channel
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Calibration coefficients for each individual sensor within a transducer were determined
by reading in the summary calibration file for each sensor and then using a custom
Lab View program to perform a 2“* order polynomial curve fit to the cahbration data for
each sensor channel using the equation:
f t = ,
7=0
where /, represented the best polynomial fit, x/ represented the input sequence of
UniForce voltage values, Q j represented the polynomial fit coefficients and m
represented the polynomial order, which in this case was set at two. A mean square error
term was also calculated for each sensor curve fit using the equation:
1 % ! , ^
mean square error = " X ) >
^ 7=0
where y represented the force values obtained from the load cell and n represented the
number of data points. This mean square error value represented the relative measure of
the residuals between the expected curve values and the actual observed values.
The output from this program generated the coefficients and a mean square error term
for each individual sensor within a transducer. These coefficients were then stored in a
summary coefficient file for each transducer. The coefficient file was then read into the
Visual Basic program and used to cahbrate the individual sensor channels for each
particular transducer.
Once all channels were properly cafibrated the peak change in force for each individual
sensor was determined and stored in a summary file which included the trial filename
and the peak force for each individual sensor channel. The cafibrated force data across all
sensor channels for each trial was also stored in a separate file with a unique suffix
identifier (*.pfd). A transposed data file was also created to store the 100 frames of data
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
prior to the peak headform acceleration data frame and the 100 frames following the
peak headform acceleration data frame (*.tpfd). This file was created to assist in the visual
display of the data over time.
Given the fact that the sensing area of the transducer was known to be 7.92 mm^, a
simple calculation was made to convert the force values into pressure values in Kpa. The
pressure values for each sensor over time were then stored as pressure data files (*.ppd).
Transposed pressure data files (*.tppd) were also stored.
Statistical Analysis
Following data reduction, 20% trimmed mean calculations were made for the peak
acceleration values for each helmet liner density across each drop height, impact location
and impact surface. This was accomplished using Splus software which included the Splus
function mean(^,y) where y represented the amount of trimming, which for all analyses
was set at 20%. Trimmed mean values were selected in an effort to reduce the effects of
outliers within the data set (Wilcox, 1995). Similar trimmed mean calculations were made
using the impact duration data, HIC data, and the G ADD data. The peak sensor force
data was also subjected to trimmed mean computations; however, only after each trial
was reviewed and properly coded according to the criteria described above.
Upon completion of the trimmed mean calculations, linear contrasts between each group
at each drop height for a given impact anvil and location consideration were performed
using the SPLUS function lincon. Only linear contrasts were performed in order to
control the experimentwise Type I error and to maximize statistical power
(Bernhardson, 1975).
In order to analyze the relative importance of bicycle helmet liner density across all
impact drop heights, all data was collapsed across drop heights for subsequent analysis.
This decision was based on the fact that any given bicycle helmet must be capable of
performing well over a wide range of potential impact energy levels. Therefore, a
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
comparison of helmet liner density performance across drop heights (i.e. energy levels)
was performed using the same linear contrast procedures described above.
Upon completion of the linear contrast analyses described above, a percentage bend
correlation procedure was used to determine whether or not the peak headform
acceleration value for a given test was independent of the peak sensor contact force for
the same test. The SPLUS function pbcor was used for these tests (Wilcox, 1996). In
addition to the percentage bend correlation analysis, scatterplots of the peak headform
acceleration and peak sensor force for each trial were generated. It was expected that
these scatterplots might provide some insight into the relationship between peak
headform acceleration and peak sensor force.
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapt e r 4
RESULTS
The primary purpose of the present study was to design and develop a reliable transducer
which was capable of monitoring local forces on the surface of an ISO magnesium test
headform during bicycle helmet impact testing. A secondary purpose was to evaluate the
effect that bicycle helmet liner density had upon these local forces during standard
bicycle helmet impact tests.
Transducer Response Evaluation
Prior to any actual testing, the performance of the digital circuit was confirmed by
collecting the driving voltage output from the relay circuit and the output from the
UniForce sensor while a load was applied to the sensing area of the sensor. Figure 7
illustrates the voltage output from the UniForce Sensor as well as the -5 volt driving
voltage which was supplied to the transducer for a period of 80 ms. The data clearly
shows that the digital circuit was performing correctly in that sufficient voltage was
supplied to the sensor prior to impact in order to produce a voltage output ^ v e n an
applied loacQ and that the driving voltage was applied for a duration of 80 ms; thereby
ensuring that the transducer was not damaged due to prolonged voltage input. Evaluation
of the inherent noise of the system during these cahbration and system check tests
revealed that the maximum level of noise within any given test was 50 mv. This was felt
to be acceptable given the fact that the typical sensor output was expected to be in the
range of 2 to 7 volts.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I
î -,
CO
-3
-4
-5
-6
— — L oad Cell O utput
— — Driving V oltage
..................U niForce O utput
! to
---------- 1 ■
2 0 • 30
1
40 50 60 70 80 90 ICO 1 110 120 130 140
!
1
i
i
i
1
T im e (m s)
Figure?. Evaluation of Digital Circuit Performance
Phase response characteristics
Given the very short duration of a bicycle helmet impact, it was imperative that the
UniForce transducer system have a very rapid response time and minimum phase lag.
Since bicycle helmet impacts are currently evaluated using accelerometer based
instrumentation, it was felt that a correlation analysis between the accelerometer
response and the response of the UniForce transducer would provide insight into the
temporal response characteristics of this transducer. Figure 8 illustrates the response of
the UniForce transducer relative to a uniaxial accelerometer mounted at the center of
gravity of the test headform. As illustrated, the response of the UniForce transducer
actually leads the response of the accelerometer. This was not surprising given the fact
that the UniForce transducer was located on the surface of the test headform while the
accelerometer was located at the center of gravity of the test headform. A cross
correlation analysis of the thirty trials conducted yielded a trimmed mean Pearson
product moment correlation value of .967 (20% trimming). The range of correlation
values was from .961 to .977.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
&
a
«
u
3
%
C
(0
5
4
Accelerometer Voltage
UniForce Voltage
3
2
1
0
-1
Time (ms)
Figure 8. Phase Response Comparison Between Accelerometer and UniForce Sensor
Transducer Linearity Evaluation
Evaluation of the linearity of the transducer was performed by comparing the output
from a calibrated Sensotec Model 31 compression/tension load cell with the output from
the UniForce transducer. Loads over the response range of the sensor were applied to the
transducer over the entire sensor area using an arbor press apparatus. Once all data was
collected, the output was subjected to a linear regression analysis which reported a best fit
linear curve as well as the residual values for each fitted point. Figure 9 illustrates the
linearized response of a sample sensor. The r^ value for this particular test was .958.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
300
250
UniForce S e n so r 63
Linear C urve Fit
200
z
150
S
£
100
0 1 2 3 5 6 9 4 7 8
Transducer Output (volts)
Figure 9. Linearized Response From UniForce Transducer
After reviewing the residual values from the linear regression analysis, it became obvious
that the transducer did not respond in a linear fashion over the desired response range.
The distribution of the residual values strongly suggested that a 2nd order polynomial
curve fit would better approximate the response of the sensor. Figure 10 illustrates the
same data shown in Figure 9; however, a 2nd order polynomial curve fit was used to
model the response of the transducer instead of a linear curve fit.
4 0 0
350
UniForce S e n so r 63
2nd O rder Potynornal
300
250
z
200
§
£
150
100
0 1 2 3 6 5 7 8 9 4
Transducer Output (volts)
Figure 10. Second Order Polynomial Curve Fit of Calibration Data
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The mean square error value for this particular curve fit was found to be 2.10 N. Given
the fact that this approach provided a better fit to the output data from the UniForce
transducer, it was decided that all cahbration coefficients for each sensor would be
subjected to a 2nd order polynomial curve fit procedure.
Transducer Hysteresis Evaluation
Transducer hysteresis was not considered to be an important response characteristic with
respect to this study; however, it was felt that defining the hysteresis characteristics
would be useful in the event that the rate of loading versus unloading became an
important measurement variable. Hysteresis data came from two sources, the first one
being the manufacturer of the transducer and the second one being physical tests done on
the actual transducers used in this study. The manufacturer performed tests on single
sensor samples which were produced at the same time as the 65 channel transducer
system. The maximum hysteresis values for each sensor tested are illustrated in Table 3.
Sensor ED Sensor Range (N) Maximum Hysteresis (%)
476313R 220 14.0
476313L 220 15.0
476325R 220 13.5
476325L 220 12.5
476713R 2200 10.0
476709R 2200 10.0
476717R 2200 10.0
476325L 2200 12.5
476719R 2200 11.5
476711R 2200 9.5
Table 3. Hysteresis Values As Reported By Transducer
Manufacturer
As indicated above, the range of maximum hysteresis values as obtained by the
manufacturer was from 9.5% to 14%.
The maximum hysteresis values obtained using the instrumentation designed exclusively
for this study are located in Table 4. The range of maximum hysteresis values was from
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.56% to 17.01% with a trimmed mean value of 13.28%. The different hysteresis values
observed in Table 4 could be due to the fact that the sensors were evaluated over a larger
operational range (0 to 10 volts) when compared to the data obtained by the
manufacturer (0 to 5 volts). It should also be mentioned that the instrumentation used for
this study was not exactly the same as the instrumentation used by the manufacturers,
even though the circuit design was identical.
Trial Number Maximum Hysteresis (%)
1 12.91
2 17.01
3 15.57
4 8.56
5 11.35
Table 4: Hysteresis Values Obtained Using Experimental Instrumentation
Transducer Repeatability Analysis
An evaluation of the transducer's abihty to provide repeatable data was evaluated by
repetitive apphcation of a known load to a sample sensor over a period of three days.
Twenty tests were performed during each test period and the input load and transducer
output were collected for each test. Given the fact that there was a sUght deviation in the
input load between trials, each trial was normalized to a ratio of input force/output
voltage. Table 5 illustrates the trimmed mean ratios for each test session.
Day 1 Response Ratio
(N/v)
Day 2 Response Ratio
(N/v)
Day 3 Response Ratio
(N/v)
48.34 (0.65) 46.42 (1.22) 53.36 (1.24)
Table 5; Trimmed V ean Response Ratios (± SE) For Repeatability Analysis
The distribution of the repeatability data also appears as a boxplot illustrated in Figure
11. As this figure indicates, there were several instances of outhers noted for the first day
repeatabihty testing while there were no outhers for the second or third day of data
collection.
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65 —I
o
1
55 —
g
I
50
3
45
4 0
2
Test Session By Day
3 1
Figure 11. Boxplot of Repeatability Data (n-20 trials per day)
A comparison of all pairwise differences using a percentile t bootstrap procedure
(pairdepb) reported no significant differences between pairs at an a level of .05. This lack
of significant difference between days indicates that the sensor may be considered to be
repeatable across different test days. With regards to the test procedures used during these
tests, the issue of transducer repeatability is not relevant since each transducer unit is
applied to the headform only once and then discarded once all tests are completed at the
specified impact location.
Sensor Output During Partial Coverage
In order to assess the effect that partial loading would have on the transducer output, a
series of constant loads were appUed to one individual sensor with varying amounts of
coverage over the sensing area. Figure 12 illustrates the transducer output given a
constant load application with 25%, 50%, 75% and 100% sensing area coverage. As this
figure indicates, there is a tremendous difference in the transducer output depending
upon the actual area over which the load is applied.
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
7
!:
I:
*■ 1 ■
♦ ♦ ♦ ♦ ♦ ♦
^ O u tp u t W ith 25 Percent
C overage
g O utput W ith 50 Percent
C overage
^ O u tp u t W ith 75 Percent
C overage
X O utput W ith 100 Percent
C overage
350 355 360 365
Applied Force (N )
370 375 380
Figure 12: UniForce Transducer Output Relative To Percentage Coverage of Sensing Area
If only 25 percent of the sensor is in contact with the test helmet, then the sensor output
is only approximately 1 volt compared to approximately 9 volts when the force is
applied over the entire sensor area. Even if the force is applied to 75 percent of the sensor
area the output is still below the output for the 100 percent coverage trials. These results
clearly indicate that in order to maximize transducer output, the impact forces should be
applied over the entire area of the sensor.
Sensor Calibration For Impact Testing
Upon completion and review of all preliminary tests, it was felt that the UniForce
transducer was a suitable test device for monitoring local forces during bicycle helmet
impacts. This decision was based largely on the abihty of the transducer to respond
rapidly under impact conditions (i.e. phase response) and the abihty of the transducer to
be cahbrated against a known reference standard.
A total of six individual transducers were utilized for this study. Four transducers were
used to monitor low level impacts (220 N load limit per sensor) and two were used to
monitor high level impacts (2200 N load limit per sensor). Each sensor which was used
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for testing as part of this study was calibrated using the methods described earher for a
total of 7,560 conditioning and calibration tests. A mean squared error (MSE) term was
calculated for each sensor using a 2°^ order polynomial curve fit of the calibration data
obtained for each individual sensor channel. The trimmed mean of the MSE over all 63
sensors for each transducer is presented in Table 6 with the standard error, range and
median MSE values for each transducer.
Sensor ED Trimmed
Mean MSE
(N)
Standard
Error (N)
Max MSE
(N)
Min MSE
(N)
Median
MSE(N)
9 50.21 6.41 336.12 5.21 45.55
11 47.62 7.27 664.69 4.96 40.80
16 21.79 1.81 138.74 2.10 21.16
17 14.55 1.71 88.81 2.09 12.94
18 38.67 2.93 114.99 2.43 38.85
24 19.28 2.13 111.28 2.85 16.36
Table 6: Trimmed Mean Values of MSE For Each Transducer Used During Testing
The largest MSE values were observed for the high range sensors as opposed to the low
range sensors. Given the fact that the maximum output of the high range sensors was in
excess of 2200 N, it was not surprising that the maximum MSE values would be larger
than the lower range sensors, which could provide maximum force output values in the
220 N range.
The high MSE values may have been due to incomplete coverage over the sensor area
during cahbration. This would have caused a reduction in the force output and an
increase in the MSE value observed. The high MSE values could also have been due to
the presence of a foreign body (such as a tape build-up) below the sensor during
calibration. If a foreign body created an irregularity in the surface of the sensor, the
voltage output would be quite erratic relative to the sensor output without the
irregularity. Typically the voltage output could be expected to increase due to this high
point loading in one area of the sensor. This high voltage would then create a large MSE
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
relative to the 2 °^ order polynomial curve fit. Every effort was made to ensure that the
sensor interface was free of foreign materials.
As indicated in the methodology section, each transducer system was calibrated before
and after testing. A comparison between pre and post test cahbration values indicated
either good agreement with the pre-test cahbration values or a consistent reduction in the
voltage output from the sensor foUowing testing. Good agreement was defined as a
similar cahbration slope through the range of measurement.
There were very few instances where the sensor output foUowing testing was greater
than the transducer output prior to testing. The cahbration values reported foUowing
testing were consistently lower than those reported prior to testing. Given the
propensity of the transducer to become damaged during testing or during the removal
process from the headform and the potential loss of tape to the backing surface, it was
felt that only pre test cahbration values would be used to determine sensor coefficient
values. A typical pre and post test comparison appears in Figure 13.
8
7
> 5
I
4
O
3
2
1
0
300 200 250 SO 100 0 150
Force (N)
Figure 13: Comparison of Pre-test and Post-test Calibration Values For Sensor 30
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Helmet Test Samples
A total of 200 helmets were fabricated exclusively for this project by a well known
bicycle helmet manufacturer. Helmets were produced of expanded polystyrene (EPS) and
were drawn from one of 12 cavities for this particular helmet model. The mean cavity
volume was found to be 2494.83 ml with a range between 2489.4 ml and 2498.8 ml. The
mean cavity volume was used to calculate the actual density of each individual hehnet
sample given the sample mass which was obtained via direct measurement using a
calibrated electronic scale. The trimmed mean helmet mass values appear in Table 7
along with a comparison between the actual helmet liner density and the desired helmet
liner density. As the table indicates, the helmet liner densities supphed by the
manufacturer were not exactly to the specifications requested. The upper limit of the
helmet liner density was found to be 98 kg/m^ as opposed to 128 kg/m^ as requested. Due
to the costs, time delays and the potential change in performance due to different EPS
bead batches, it was not possible to request an additional helmet finer density at 128
kg/m^. It was felt that the lack of an extremely high density helmet finer would not
adversely affect the conclusions of this study given the fact that the majority of
contemporary bicycle helmet liners are produced at a density at or below 97 kg/m^.
Furthermore, although the lack of a 128 kg/m^ helmet liner density does limit the
interpretation of the results, it was felt that this group of bicycle helmet liners was still
diverse enough to allow for an effective interpretation regarding the effect of helmet finer
density upon the selected test variables.
Group Desired
Density
(kg/m^
Target Mass
(gm)
Actual Mass
(gm)
Actual Density
(kg/m")
1 32 79.9 87.04 (.50) 34.89
2 64 159.8 140.16 (.53) 56.18
3 96 239.8 201.48 (.93) 80.72
4 128 319.7 243.31 (1.12) 97.53
Table 7: Summary of Helmet Mass Analysis (± SE)
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The data found in Table 7 also indicates that there was no significant variability among
helmets for a given helmet liner density. The increase in the standard error for the group
3 and 4 helmet liners suggests that as the helmet liner density increases, difficulties in
production tend to increase the variability between helmets.
Impact Test Results
A total of 327 helmet impact tests were conducted for this study across two impact
locations, four drop heights and two impact surfaces. The headform acceleration data for
the left front hemispherical anvil impact locations for each group across the four different
drop heights appears in Figures 14 through 17. These plots illustrate typical headform
acceleration data which was collected across all impact locations, drop heights and impact
surfaces. A summary of the headform acceleration plots across drop heights for each
helmet group for the remaining three impact locations appears in Appendix A.
500
40 0
0.5 m Drop Height
1 0 m Drop Height
1.5 m Drop Height
S
0 300
1
I 200
S
40
-100
Time (ms)
Figure 14: Group 1 Headform Acceleration Across Drop Heights - Left Front Impact Location,
Hemispherical Anvil
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
0 .5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
200
a
150
100
40
-50
Time (ms)
Figure 15: Group 2 Headform Acceleration Across Drop Heights - Left Front Impact Location,
Hemispherical Anvil
160
0 .5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2 .0 m Drop Height
140
§
1
I
I
100
40
-20
Time (ms)
Figure 16: Group 3 Headform Acceleration Across Drop Heights - Left Front Impact Location,
Hemispherical Anvil
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
160
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
_ 140
S
§ 120
=
e 100
4 0
20
40 20
-20
Time (ms)
Figure 17; Group 4 Headform Acceleration Across Drop Heights - Left Front Impact Location,
Hemispherical Anvil
As illustrated in the figures above, peak headform acceleration values increase with
increases in drop height. This was to be expected given the fact that as drop height is
increased, the kinetic energy of the impact is increased and the energy management
demands placed upon the helmet are increased. Under two test conditions, the 1.5 m and
2.0 m drop heights at the left front impact location, hemispherical anvil, the energy
demands of the impact were in excess of the energy absorbing requirements of the helmet
and the helmet liners effectively “bottomed out”, meaning that they had absorbed a
maximum amount of impact energy and once this energy limit was reached, any excess
energy was transferred directly to the headform. The net result of this excess energy
apphed to the headform was an excessively high force being apphed to the headform (as
illustrated by the acceleration curve). N o tests were performed at the 2.0 m drop height
for the group 2 helmet liners based on the results obtained from the 1.5 m drop height
which created a high risk of damage to the test instrumentation.
The headform acceleration data presented across groups at a drop height of 1.5 m are
presented in Figures 18 through 21. These plots are considered typical of all the data
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
observed. A complete set of data for each impact location across all drop heights may be
found in Appendix B. A summary of the trimmed mean peak headform acceleration
across all groups and drop heights is located in Tables 8 through 11.
500
Group 1
Group 2
Group 3
Group 4
^ 400
3
I 300
2
I 200
<
I 100
1
X
40
-100
Time (ms)
Figure 18: Headform Acceleration Data For Front Left Impact Location, Hemispherical Anvil, 1.5
m Drop Height
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
200
Group 1
Group 2
Group 3
Group 4
3
§ 150
I
<
i
X
100
-50
Time (ms)
Figure 19: Headform Acceleration Data For Front Left Impact Location, Flat Anvil, 1.5 m Drop
Height
500
G roup 1
G roup 2
G roup 3
G roup 4
400
300
200
100
0
40
-100
Time (ms)
o >
E
I
<
1
Figure 20: Headform Acceleration Data For Right Side Impact Location, Hemispherical Anvil, 1.5m
Drop Height
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
160
G roup 1
G roup 2
G roup 3
G roup 4
140
S 120
5 100
I .
I 60
J 40
«
I 20
5 0
-20
-40
Time (me)
Figure 21: Headform Acceleration Data For Right Side Impact Location, Flat Anvil, 1.5 m Drop
Height
50 cm Ah
(g)
100 cm Ah
(g)
150 cm Ah
(g)
200 cm Ah
(g)
Row Means
(g)
Group 1 59.4 (0.89) 86.95 (1.27) 109.33 (.38) 151.67 (1.57) 98.06 (11.27)
Group 2 82.63 (4.55) 121.6 (3.83) 153.4 (2.26) 177.93 (2.20) 139.27(11.21)
Group 3 101.28 (2.69) 148.00 (2.54) 179.75 (5.93) 216.18 (1.36) 162.67(13.95)
Group 4 113.63 (1.17) 158.63 (3.28) 198.85 (4.48) 237.13 (3.55) 190.16(17.58)
Table 8: 20% Trimmed Mean Peak Headform Acceleration Across Groups and Drop Heights - Left
Front Impact Location, Flat Anvil
50 cm Ah
(g)
100 cm Ah
(g)
150 cm Ah
(g)
200 cm Ah
(g)
Row Means
(g)
Group 1 42.40 (0.58) 84.55 (8.95) 472.33(21.52) not tested 64.58 (10.98)
Group 2 52.30 (2.29) 77.97 (0.11) 111.20 (3.39) 180.58(22.33) 88.00 (9.74)
Group 3 63.48 (2.08) 83.90 (1.98) 114.67 (2.18) 14473 (2.72) 97.39 (9.50)
Group 4 72.33 (1.77) 96.00 (1.60) 119.33 (0.71) 149.40 (3.69) 104.64 (8.64)
Table 9: 20% Trimmed Mean Peak Headform Acceleration Across Groups and Drop Heights - Left
Front Impact Location, Hemispherical Anvil
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah
(g)
100 cm Ah
(g)
150 cm Ah
(g)
200 cm Ah
(g)
Row Means
(g)
Group 1 53.53 (0.53) 79.47 (0.40) 107.67 (0.45) 152.03 (0.68) 95.05 (11.84)
Group 2 75.17 (2.12) 101.67 (0.50) 128.23 (2.62) 155.40 (0.56) 113.82 (9.06)
Group 3 90.97 (1.06) 121.47 (3.21) 147.27 (3.03) 178.33 (5.13) 133.80 (9.99)
Group 4 101.03 (1.44) 135.83 (0.23) 168.27 (3.34) 191.37 (2.01) 150.50(10.59)
Table 10: 20 % Trimmed Mean Peak Headform Acceleration Across Groups and Drop Heights
Right Side Impact Location, Flat Anvil
50 cm Ah
(g)
100 cm Ah
(g)
150 cm Ah
(g)
200 cm Ah
(g)
Row Means
(g)
Group 1 39.86 (0.42) 100.50(11.34) 437.57(18.64) not tested 170.40(61.48)
Group 2 53.00 (0.92) 75.57 (0.68) 102.48 (1.62) 186.53 (4.59) 96.32 (19.32)
Group 3 65.33 (0.85) 94.20 (2.39) 117.43 (1.94) 145.57 (0.91) 104.72 (9.43)
Group 4 70.50 (1.05) 99.63 (1.57) 121.00 (1.07) 150.57 (3.95) 110.28 (9.83)
Table 11: 20% Trimmed Mean Peak Headform Acceleration Across Groups and Drop Heights - Right
Side Impact Location, Hemispherical Anvil
Those tests conducted with group 1 helmet liners at the 1.5 m drop height against the
hemispherical anvil were found to produce the highest peak headform acceleration values
throughout the entire test series (a trimmed mean of 472 g for the left front impact
location and a trimmed mean of 438 g for the right side impact location). It should be
noted that these peak headform acceleration values are well in excess of the pass/fail
criteria for current bicycle helmet standards.
In most cases, peak headform acceleration increased as a function of drop height and as a
function of helmet liner density. The lowest peak headform acceleration values were
obtained during the 50 cm right side hemispherical and left front hemispherical impact
tests (a trimmed mean of 40 g and a trimmed mean of 42 g respectively). The highest
peak headform acceleration values (aside from the “bottomed out” group 1 tests)
occurred during the 2.0 m flat anvil tests at both the left front impact location and the
right side impact location (a trimmed mean of 237 g and a trimmed mean of 191 g
respectively).
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The only exceptions to this natural trend of increased peak headform acceleration across
drop heights and across helmet liner densities occurred during the hemispherical anvil
tests for the group 1 and group 2 helmet liners. At the 1.0 m drop height, the trimmed
mean peak headform acceleration for the group 1 helmet liners was found to be 85 g
while the trimmed mean peak headform accelerations for the group 2 and group 3
helmet liners at the 1.0 m drop height were found to be 78 g and 84 g respectively. The
occurrence of the higher trimmed mean peak headform acceleration at the 1.0 m drop
height was likely related to the fact that the helmet liner was very rapidly approaching its
maximum energy absorbing limit and the magnitude of the forces being apphed to the
headform were rapidly increasing. A similar phenomenon was likely occurring at the 1.0
m drop height, right side impact location against the hemispherical anvil where the
group 1 trimmed mean peak headform acceleration was found to be 101 g, which was
greater than all other groups tested at the 1.0 m drop height.
The group 2 helmet liners exhibited trimmed mean peak headform accelerations which
were contrary to this trend of increasing acceleration with increasing helmet liner
density. At the 2.0 m drop height against the hemispherical anvil, tor both the left front
and the right side impact locations, the trimmed mean peak headform acceleration for
the group 2 helmet liners was greater than the trimmed mean peak headform acceleration
values for the other two groups tested. Analysis of the data presented in figure 15 above
clearly shows that at the 2.0 m drop height against a hemispherical anvil, the group 2
helmet liner is rapidly approaching the limit of energy absorbing capacity relative to the
higher density helmet liners. The slope of the acceleration curve for the group 2 helmet
rapidly increases and begins to resemble the “bottomed out” acceleration curve noted for
the group 1 helmet Uner which appears in Figures 14 and 20. Any further increase in
impact energy would likely result in a peak headform acceleration value in excess of 300
g for the group 2 helmet liners if tested against the hemispherical anvil.
The results of the linear contrast analysis for the peak headform acceleration values
appears in Tables 12 through 15 The only consistent difference between groups appears
between the group 1 and group 4 helmet liners at both impact locations against a flat
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
anvil. Observation of the data presented in the tables above indicate that the trimmed
mean peak headform acceleration values for the group 1 helmet liners were consistently
lower than the mean peak headform acceleration values for the group 4 helmet liners.
This relationship did not continue across to the hemispherical anvil tests where
significant differences were observed at only the 50 cm and 1.50 m drop heights.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 4.57 9.05* 20.19* 9.72* 2.70
Group 1 V . Group 3 15.31* 22.64* 12.49* 29.81* 3.83*
Group 1 V . Group 4 36.61* 21.51* 20.98* 22.61* 4.73*
Group 2 V . Group 3 3.33 6.05* 4.38 15.54* 1.42
Group 2 V . Group 4 6.07 7.74* 9.55* 14.95* 2.66
Group 3 V . Group 4 4.43* 2.70 2.71 5.80* 1.34
Table 12: Results of Linear Contrast Analysis For Peak Headform Acceleration - Left Front Impact
Location, Flat Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 4.46 1.23 26.88* not tested not tested
Group 1 V . Group 3 10.41* 0.11 27.24* not tested not tested
Group 1 V . Group 4 17.12* 2.05 27.29* not tested not tested
Group 2 V . Group 3 3.82* 3.14 0.79 2.61 0.70
Group 2 V . Group 4 7.30* 11.84* 2.14 2.23 1.29
Group 3 V . Group 4 3.42 5.01 1.86 0.93 0.59
Table 13: Results of Linear Contrast Analysis For Peak Headform Acceleration - Left Front Impact
Location, Hemispherical Anvil { * indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 9.05* 31.84* 7.07 3.47 1.24
Group 1 V . Group 3 28.95* 11.84* 11.79* 4.64 2.46
Group 1 V . Group 4 28.37* 112.88* 16.43* 16.88* 3.43*
Group 2 V . Group 3 6.09* 5.56 4.34 4.06 1.46
Group 2 V . Group 4 9.23* 57.00* 8.61* 15.71* 2.59
Group 3 V . Group 4 5.16* 4.07 4.25 2.16 1.13
Table 14: Results of Linear Contrast Analysis For Peak Headform Acce
Location, Flat Anvil (* indicates significance for a
eration - Right Side Impact
- .05)
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 12.31* 2.31 16.39* not tested not tested
Group 1 V . Group 3 25.43* 0.57 15.59* not tested not tested
Group 1 V . Group 4 25.44* 0.08 15.48* not tested not tested
Group 2 V . Group 3 9.00* 6.86* 6.39* 8.00* 0.40
Group 2 V . Group 4 11.47* 12.89* 12.19* 5.43* 0.70
Group 3 V . Group 4 3.49 1.74 1.47 1.12 0.44
Table 15: Results of Linear Contrast Analysis For Peak Headform Acce
Location, Hemispherical Anvil (* indicates significance
eration - Right Side Impact
at a - .05)
Significant differences were observed between the peak headform acceleration values of
the group 2 helmet liners and the group 4 helmet liners across thirteen of the sixteen test
conditions analyzed. However, when the data was collapsed across all drop heights, no
significant differences between the group 2 and group 4 peak headform acceleration
values were observed.
Peak Headform Force
The trimmed mean peak headform force values obtained during this test series appears in
tables 16 through 19.
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N )
Row Means
(N)
Group 1 2063.28
(28.14)
4114.40
(435.72)
22984.84
(1047.28)
not tested 3142.51
(534.43)
Group 2 2545.04
(111.26)
3794.04
(5.50)
5411.25
(164.96)
8787.20
(1086.83)
4282.29
(474.83)
Group 3 3088.84
(100.96)
4082.77
(96.44)
5579.95
(106.13)
7043.06
(132.21)
4739.02
(462.38)
Group 4 3519.50
(86.29)
4671.58
(77.87)
5807.04
(34.56)
7270.15
(179.39)
5091.82
(420.41)
Table 16: 20% Trimmed Mean Peak Headform
Impact Location,
Force Across Groups and Drop Heig
Hemispherical Anvil
Its - Left Front
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N)
Row Means
(N)
Group 1 2890.54
(43.45)
4231.19
(61.72)
5320.42
(18.42)
7380.45
(76.44)
4771.90
(548.43)
Group 2 4021.13
(221.49)
5917.34
(186.40)
7464.80
(109.90)
8658.25
(106.83)
6777.04
(545.39)
Group 3 49.28.28
(130.94)
7202.03
(123.78)
8747.06
(288.44)
10519.58
(67.29)
7914.62
(678.68)
Group 4 55.29.26
(57.06)
7719.06
(159.35)
9676.51
(217.87)
11539.06
(172.60)
8613.23
(718.61)
Table 17; 20% Trimmed Mean Peak Headform
Impact Lee
*orce Across Groups and Drop Heig
ition. Flat Anvil
Its - Left Front
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N)
Row Means
(N)
Group 1 1939.68
(20.31)
4890.57
(551.71)
21293.02
(907.19)
not tested 6255.14
(2991.57)
Group 2 2579.10
(44.60)
3677.25
(32.81)
4986.67
(78.87)
9077.15
(223.19)
4687.31
(940.17)
Group 3 3179.27
(41.48)
4583.99
(116.17)
5714.58
(94.53)
7083.61
(44.47)
5067.37
(458.65)
Group 4 3430.70
(51.01)
4848.39
(76.19)
5888.14
(51.84)
7326.93
(192.00)
5366.24
(478.50)
Table 18: 20% Trimmed Mean Peak Headform Force Across Groups and Drop Heig
Impact Location, Hemispherical Anvil
Its - Right Side
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N)
Row Means
(N)
Group 1 2605.06
(25.60)
3867.03
(19.26)
5239.31
(22.00)
7398.30
(33.28)
4625.36
(576.11)
Group 2 3657.79
(103.11)
4947.34
(24.25)
6240.13
(127.35)
7562.13
(27.31)
5538.59
(440.71)
Group 3 4426.65
(51.42)
5910.85
(156.38)
7166.34
(147.61)
8678.12
(249.56)
6511.02
(486.21)
Group 4 4916.52
(69.84)
6606.72
(14.87)
8188.25
(162.32)
9312.35
(97.99)
7322.87
(515.20)
Table 19: 20% Trimmed Mean Peak Headform 'orce Across Groups and Drop Heig Its - Right Side
Impact Location, Flat Anvil
Peak headform force ranged between approximately 2000 N and 9000 N for the flat anvil
tests while the peak force headform range was found to be between 2000 N and
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approximately 23,000 N for the hemispherical anvil. The latter limit of peak headform
force illustrates the magmtude of forces apphed to the head when a helmet liner
effectively “bottoms out”. Given the fact that peak headform force is the product of the
mass of the test headform and the peak headform acceleration, the trends observed for
the peak headform force are identical to those observed for the peak headform
acceleration data. No statistical analyses were performed on the peak headform force
data.
Impact D uration
The trimmed mean impact duration values are presented in Tables 20 through 23. Impact
duration was found to decrease with increasing drop height. Therefore, the duration of
the energy absorption process was decreasing with increases in drop height (implying
that the helmets were capable of absorbing a greater amount of energy over a shorter
duration). In general, the impact duration observed for the hemispherical anvil tests were
greater than those observed for the flat anvil tests. This could likely be explained by the
increased intrusion of the test anvil into the helmet liner. The only exception to this
finding occurs during the 1.5 m impact tests wherein the group 1 helmet liner
experienced a peak headform acceleration value in excess of 300 g. In this case, the helmet
liner is rapidly deformed and due to the bottoming out of the hner, maximum
acceleration was reached quickly and the headform rapidly began to rebound away from
the test anvil, resulting in a shorter impact duration.
In all cases, impact duration decreased as helmet liner density increased. This would
suggest that the duration of force application to the head decreases when helmet liner
density increases; however, when the peak headform forces are taken into consideration
(i.e. increasing headform force with increasing helmet liner density), this decrease in
impact duration indicates that the impulse apphed to the headform is similar in all cases.
This would make sense since the change in velocity is relatively the same for all tests.
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah
(ms)
100 cm Ah
(ms)
150 cm Ah
(ms)
200 cm Ah
(ms)
Row Means
(ms)
Group 1 20.51 (1.80) 14.40 (2.29) 8.62 (0.20) not tested 16.71 (1.90)
Group 2 17.47 (1.05) 15.96 (0.50) 15.24 (0.70) 11.07 (0.75) 15.30 (0.68)
Group 3 14.37 (1.42) 17.50 (1.37) 11.51 (0.11) 11.38 (0.20) 13.14 (0.91)
Group 4 11.93 (0.50) 12.03 (0.12) 10.89 (0.05) 10.67 (0.49) 11.43 (0.23)
Table 20: 20% Trimmed Mean Impact Duration Across Groups and Drop Heights - Left Front Impact
Location, Hemispherical Anvil
50 cm Ah
(ms)
100 cm Ah
(ms)
150 cm Ah
(ms)
200 cm Ah
(ms)
Row Means
(ms)
Group 1 15.91 (0.35) 16.19 (0.87) 14.01 (0.53) 14.26 (0.08) 14.33 (0.69)
Group 2 10.86 (0.41) 9.90 (0.27) 9.47 (0.11) 9.10 (0.26) 9.70 (0.17)
Group 3 9.54 (0.24) 8.63 (0.15) 8.46 (0.11) 7.99 (0.25) 8.64 (0.15)
Group 4 8.89 (0.26) 8.28 (0.40) 7.93 (0.49) 6.85 (0.33) 8.01 (0.28)
Table 21: 20% Trimmed Mean Impact Duration Across Groups and Drop Heights - Left Front Impact
Location, Flat Anvil
50 cm Ah
(ms)
100 cm Ah
(ms)
150 cm Ah
(ms)
200 cm Ah
(ms)
Row Means
(ms)
Group 1 12.75 (2.58) 14.70 (0.27) 9.60 (0.30) not tested 12.50 (1.17)
Group 2 12.71 (0.30) 13.42 (0.53) 12.87 (0.27) 11.33 (0.10) 12.52 (0.30)
Group 3 10.58 (0.06) 11.51 (0.70) 11.20(0.33) 10.89 (0.22) 10.99 (0.20)
Group 4 10.62 (0.60) 10.36 (0.15) 10.71 (0.15) 10.22 (0.20) 10.42 (0.13)
Table 22: 20% Trimmed Mean Impact Duration Across Groups and Drop Heights - Right Side Impact
Location, Hemispherical Anvil
50 cm Ah
(ms)
100 cm Ah
(ms)
150 cm Ah
(ms)
200 cm Ah
(ms)
Row Means
(ms)
Group 1 13.78 (1.24) 12.89 (0.56) 12.62 (0.72) 11.73 (0.87) 12.67 (0.29)
Group 2 10.71 (0.30) 9.91 (0.35) 10.04 (0.30) 9.56 (0.22) 9.98 (0.23)
Group 3 8.84 (0.25) 8.71 (0.38) 8.80 (0.09) 9.24 (0.27) 8.89 (0.16)
Group 4 7.87 (0.58) 8.27 (0.42) 7.64 (0.31) 8.22 (0.47) 8.00 (0.25)
Table 23: 20% Trimmed Mean Impact Duration Across Groups and Drop Heights - Right Side Impact
Location, Flat Anvil
The results of the linear contrast analysis of impact duration across all groups, drop
heights and anvils appears in Tables 24 through 27. The largest presence of significant
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
differences across groups appeared at the front left impact location against the flat anvil
where twenty-four of the thirty contrasts were found to be significantly different. When
collapsed across all drop heights, all contrasts between groups at the right side impact
location against the flat anvil were found to be significant (even though many contrasts at
individual different drop heights were not found to be significant).
There were very few significantly different comparisons for the hemispherical anvil
impact tests relative to the flat anvil tests. The predominant contrasts appeared with the
group 1 impact duration data at the 1.5 m drop height. As noted above, this particular
test cell includes very high level, short duration acceleration values; therefore, it was not
surprising to note significant differences between the other three helmet liner groups at
this drop height. It was noted that a significant difference was observed between the
group 2 impact duration and the group 4 impact duration at both impact locations when
collapsed across drop heights. Observation of Tables 20 and 22 above indicate that the
trimmed mean impact duration for the group 2 helmet liners was consistently greater
than the trimmed mean impact duration for the group 4 helmet liners when tested
against the hemispherical impact anvil.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 1.71 1.05 8.57* not tested not tested
Group 1 V. Group 3 3.08 1.64 17.24* not tested not tested
Group 1 V. Group 4 5.54* 1.72 17.19* not tested not tested
Group 2 V. Group 3 1.85 1.09 4.83 0.62 1.95
Group 2 V. Group 4 5.01* 7.01 5.69 0.57 5.34*
Group 3 V. Group 4 1.70 4.17 4.66 1.22 1.91
Table 24; Linear Contrast Analysis of Impact Duration Between Groups and Across Drop Heights,
Left Front Impact Location, Hemispherical Anvil indicates significant at a - .05)
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 9.87* 7.96* 8.85* 8.29* 6.38*
Group 1 V . Group 3 16.19* 9.87* 10.85* 13.63* 7.91*
Group 1 V . Group 4 17.38* 9.51* 9.24* 14.28* 8.40*
Group 2 V . Group 3 3.00 4.73* 7.47* 3.58* 5.02*
Group 2 V . Group 4 4.37* 3.84* 3.54 6.17* 5.53*
Group 3 V . Group 4 2.10 0.92 1.21 3.19 2.14
Table 25: Linear Contrast Analysis of Impact Duration Between Groups and Across Drop Heights,
Left Front Impact Location, Flat Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 0.02 2.02 9.00* not tested not tested
Group 1 V . Group 3 0.97 3.96 3.30 not tested not tested
Group 1 V . Group 4 0.91 14.35* 3.04 not tested not tested
Group 2 V . Group 3 6.36 2.00 4.26 1.69 4.24*
Group 2 V . Group 4 2.85 5.10 9.41* 4.47 6.67*
Group 3 V . Group 4 0.07 1.48 1.24 2.04 2.36
Table 26: Linear Contrast Analysis of Impact Duration Between Groups and Across Drop Heights,
Right Side Impact Location, Hemispherical Anvil { * indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 2.19 4.13 3.02 2.22 7.18*
Group 1 V . Group 3 3.55 5.66* 4.81 2.50 11.33*
Group 1 V . Group 4 3.93 6.05* 5.80* 3.24 12.15*
Group 2 V . Group 3 4.36 2.13 3.58 0.81 3.77*
Group 2 V . Group 4 3.96 2.76 5.05* 2.33 5.72*
Group 3 V . Group 4 1.41 0.72 3.22 1.70 2.96*
Table 27: Linear Contrast Analysis of Impact Duration Between Groups and Across Drop Heights,
Right Side Impact Location, Flat Anvil (* indicates significance at a - .05)
Head Injury C riteria (HIC) Results
The trimmed mean values for Head Injury Criteria (HIC) appears in Tables 28 through
31.
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
Group 1 92.97
(1.82)
183.92
(11.45)
1477.75
(219.91)
not tested 125.15
(27.77)
Group 2 91.06 219.92 404.51 710.84 279.34
(6.87) (2.14) (6-94) . . (75.10) (54.52)
Group 3 126.29 266.63 483.49 719.40 358.08
(5.09) (14.54) (7.82) (4.29) (76.12)
Group 4 143.62 301.16 513.74 776.16 388.96
(10.31) (4.29) (19.35) (27.40) (75.03)
Table 28: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across Groups and Drop Heights ■
Left Front Impact Location, Hemispherical Anvil
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
Group 1 109.84
(5-37)
286.37
(5.56)
469.92
(3.78)
777.94
(2.94)
379.86
(86.63)
Group 2 170.21
(8.46)
438.34
(12.58)
751.72
(12.12)
1103.83
(29.95)
622.97
(112.68)
Group 3 236.73
(7.02)
594.72
(15.63)
912.35
(29.72)
1299.60
(105.44)
739.82
(121.10)
Group 4 293.03
(13.08)
662.41
(13.60)
1058.36
(33.56)
1573.71
(65.61)
866.72
(148.42)
Table 29: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across Groups and Drop Heights
Left Front Impact Location, Flat Anvil
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
Group 1 51.08
(6.79)
216.33
(11.95)
653.37
(98.28)
not tested 212.15
(75.92)
Group 2 94.63 221.39 375.98 751.76 328.97
(2.42) (1.21)
(17.71) (11.14)
(96.25)
Group 3 125.89 292.14 484.14 758.18 386.34
(1.88) (12.05) (11.68) (4.02) (74.02)
Group 4 145.33 321.26 519.58 804.45 425.53
(2.54) (3.86) (15.56) (28.18) (84.09)
Table 30: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across Groups and Drop Heights
Right Side Impact Location, Hemispherical Anvil
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
Group 1 98.24 221.38 392.81 674.91 320.81
(4.13) (0.28) (9.47) (8.30) (75.66)
Group 2 165.89 349.73 562.45 833.32 456.64
(11.18) (15.48) (4.36) (21.08) (78.91)
Group 3 224.59 459.12 740.16 1038.78 601.14
(3.58) (28.09) (22.06) (45.11) (104.37)
Group 4 260.93 544.61 816.77 1139.40 681.92
(10.74) (9.35) (13.70) (17.75) (101.87)
Table 31: 20% Trimmed Mean Head Injury Criteria (HIC) Values Across Groups and Drop Heights •
Right Side Impact Location, Flat Anvil
The HIC values for the left front impact location were found to range between 93 and
776 for the hemispherical anvil tests and 110 and 1573 for the flat anvil tests. The
observed HIC values increased with drop height and with increases in helmet liner
density. The highest HIC values were observed for the left front impact location, 2.0 m
flat anvil tests with group 2, 3 and group 4 helmets (trimmed mean values of 1103, 1299,
and 1573 respectively). The trimmed mean values observed during these tests were all in
excess of the recommended threshold level of 1000 as estabhshed by the automotive
industry. Similarly, the 1.5 m drop onto the flat anvil was also is in excess of 1000 with
an observed trimmed mean value of 1058.
Three instances of excess HIC values over 1000 were noted for the right side impact tests.
These values occurred during the group 3 and group 4 impact tests from a 2.0 m drop
height onto the flat anvil and during the 1.5 m hemispherical impact test with a group 1
helmet. This latter case once again emphasizes the increased risk of head injury whenever
a helmet liner fails to effectively absorb the energy of a given impact. As noted with the
left front impact data, HIC values increased with drop height and with increased helmet
liner density.
The results of the linear contrast analysis across impact locations, test anvils and drop
heights appears in Tables 32 through 35.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 4.20 4.96* 8.12* not tested not tested
Group 1 V . Group 3 12.52* 5.37* 7.52 not tested not tested
Group 1 V . Group 4 8.15* 14.68* 7.21 not tested not tested
Group 2 V . Group 3 4.34 3.34 6.90* 0.19 0.86
Group 2 V . Group 4 4.47 17.29* 4.85 1.21 1.21
Group 3 V . Group 4 1.59 2.40 1.32 1.87 0.30
Table 32: Linear Contrast Analysis of Head Injury Criteria Across Groups and Drop Heights - Left
Front Impact Location, Hemispherical Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 5.50* 11.65* 23.06* 11.40* 1.80
Group 1 V . Group 3 14.28* 19.60* 15.53* 5.21 2.58
Group 1 V . Group 4 13.34* 11.00* 18.33* 12.77* 3.05*
Group 2 V . Group 3 5.83* 8.22* 5.28* 1.88 0.77
Group 2 V . Group 4 7.93* 6.26* 9.06* 6.87* 1.43
Group 3 V . Group 4 4.00 1.84 3.43 2.33 0.73
Table 33: Linear Contrast Analysis of Head Injury Criteria Across Groups and Drop Heights - Left
Front Impact Location, Flat Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 6.73* 0.44 2.56 not tested not tested
Group 1 V . Group 3 11.97* 4.36* 1.56 not tested not tested
Group 1 V . Group 4 14.44* 8.67* 1.23 not tested not tested
Group 2 V . Group 3 9.31* 5.33 6.50* 0.49 0.50
Group 2 V . Group 4 13.19* 22.57* 7.49* 1.59 0.80
Group 3 V . Group 4 5.62* 2.10 1.73 1.48 0.34
Table 34: Linear Contrast Analysis of Head Injury Criteria Across Groups and Drop Heights - Right
Side Impact Location, Hemispherical Anvil (* indicates significance at a — .05)
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 5.18 7.57 14.86* 6.38* 1.22
Group 1 V . Group 3 21.11* 7.72* 13.22* 7.24* 2.14
Group 1 V . Group 4 12.91* 31.55* 23.24* 21.63* 2.80
Group 2 V . Group 3 4.56 3.11 7.22* 3.77 1.08
Group 2 V . Group 4 5.60* 9.84* 16.15* 10.14* 1.72
Group 3 V . Group 4 2.93 2.64 2.68 1.89 0.54
Table 35: Linear Contrast Analysis of Head Injury Criteria Across Groups and Drop Heights - Right
Side Impact Location, Flat Anvil (* indicates significance at a - .05)
As indicated above, the majority of contrasts between the group 1 helmet liners and the
other three groups were found to be significantly different (33 out of 48 comparisons).
Similarly a large number of group 2 and group 4 comparisons were found to be
significantly different (12 out of a possible 20 comparisons). All HIC value comparisons
between the group 2 and group 4 data at the right side impact location, flat anvil were
found to be significant although when collapsed across drop heights, no significant
difference was observed.
Peak Sensor Force Change Results
The trimmed mean peak sensor force change across impact locations, drop heights and
test anvils appears in Tables 36 through 39.
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N)
Row Means
(N)
Group 1 23.43 98.61 2086.85 not tested 295.37
(2.30) (66.42) (629.13) (367.26)
Group 2 62.61 77.97 219.74 335.66 152.77
(16.01) (12.04) (17.67) (139.36) (40.43)
Group 3 146.26 150.91 152.37 203.29 161.27
(22.63) (38.20) (21.13) (34.21) (10.52)
Group 4 252.37 272.91 208.96 288.05 252.40
(75.62) (31.95) (77.28) (82.45) (21.68)
Table 36: 20% Trimmed Mean Peak Sensor Force Across Groups and Drop Heights - Left Front
Impact Location, Hemispherical Anvil
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
(N)
(N)
(N)
(N) (N)
Group 1 19.19 27.78 26.95 54.84 29.46
(1.22) (4.42) (1.08) (1.13) (4.00)
Group 2 48.18 54.37 50.28 62.71 54.65
(9.08) (3.32) (2.83) (5.94) (2.45)
Group 3 109.05 101.10 126.33 112.07 113.15
(20.15) (10.35) (14.57) (17.26) (8.48)
Group 4 159.39 162.22 192.14 179.39 175.73
(26.36) (15.41) (10.18) (20.62) (9.43)
Table 37: 20% Trimmed Mean Peak Sensor Force Across Groups and Drop Heights - Left Front
Impact Location, Flat Anvil
50 cm Ah
(N)
100 cm Ah
(N)
150 cm Ah
(N)
200 cm Ah
(N)
Row Means
(N)
Group 1 26.02
(1.59)
489.51
(264.85)
4153.70
(203.23)
not tested 793.87
(623.18)
Group 2 58.93 75.71 180.05 598.22 123.51
(6.15) (4.21) (14.72) (259.59) (23.02)
Group 3 81.71 118.87 165.85 208.93 139.58
(7.29) (11.96) (9.55) (12.43) (15.08)
Group 4 132.90 160.99 169.31 201.33 164.79
(24.44) (9.49)
(19.11)
(33.75) (8.35)
Table 38: 20% Trimmed Mean Peak Sensor Force Across Groups and Drop Heights - Right Side
Impact Location, Hemispherical Anvil
50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Row Means
(N) (N) (N) (N)
(N)
Group 1 23.85 21.11 34.79 90.96 35.31
(1.63) (4.65) (6.14) (/.37) (8.89)
Group 2 38.80 38.03 54.60 59.65 49.39
(1.42) (0.88) (3.90) (1.89) (3.49)
Group 3 71.85 94.95 92.64 110.40 93.02
(3.01) (12.38) (2.97) (3.45) (6.06)
Group 4 121.17 134.40 134.02 159.19 137.07
(8.78) (8.05) (8.92) (4.81) (6.96)
Table 39: 20% Trimmed Mean Peak Sensor Force Across Groups and Drop Heights - Right Side
Impact Location, Flat Anvil
The general trend across all impact locations and test anvils was that trimmed mean peak
sensor force increased with increases in drop height and with increases in helmet liner
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
density. The lowest trimmed mean peak sensor force changes were observed during 50
cm drop height testing of the group 1 helmet liners across all impact anvils and impact
locations. The range of trimmed mean peak sensor force changes for these tests was
between 19 N and 26 N. The highest trimmed mean peak sensor force changes were
observed during the 1.5 m drop height group 1 tests against the hemispherical test anvil.
In these tests, the trimmed mean peak local contact sensor force was found to be 2086 N
for the left front impact location and 4153 N for the right side impact location. Recall
that these peak sensor force changes represent a force which is applied over a 31.67 mm^
area. The next largest peak local contact sensor force changes were observed during the
2.0 m drop height hemispherical impact tests for the group 2 helmet liners at both impact
locations. For this test configuration, the peak sensor force change was found to be 598
N at the right side impact location and 335 N at the left front impact location. This
finding once again suggested that the helmet was rapidly approaching its maximum
energy absorbing limit due to the fact that very large forces were already being
transmitted to the headform through the helmet liner.
As mentioned above, the general trend across drop heights and across helmet liner
density was for an increase in peak sensor force. However, there were several instances
where the peak sensor force was either equal to or less than the peak sensor force at the
lower drop height for a given helmet liner density. For example, the trimmed mean peak
sensor force change for the group 1 helmets at the 1.0 m drop height was found to be 28
N while the trimmed mean peak sensor force change for the 1.5 m drop height was
found to be 27 N. This relatively small force change can possibly be explained by the fact
that almost all of the impact energy was being absorbed by the helmet hner and the only
remaining force being transmitted to the headform was due to relatively small contact
forces which have no potential for head injury. As the impact energy levels increase,
more force is transmitted to the headform and this is reflected by an increase in the peak
sensor force change.
A second explanation for differences noted across drop heights and across helmet liner
density could be that the loads apphed to the helmet were shared across one or more
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transducers in the vicinity of the point of load application. Even though every effort was
made to place each test helmet on the headform in a similar manner, small changes in
helmet contour due to production differences and small changes in the helmet position
on the headform could change the load distribution pattern of a particular impact. In this
instance, smaller force changes would be measured on individual sensors even though the
total force change would be the same over the entire surface of the headform.
A last explanation could be that a given sensor malfunctioned or was damaged as a result
of continuous testing and consequently failed to monitor a significant local contact sensor
force change. It must be noted that every precaution was taken to ensure that this did not
occur; however, there were instances of signal shorting and instances of transducer
dropout due to damage noted during the testing.
Results of the linear contrast analysis peak sensor force appear in Tables 40 through 43.
As expected there were significant differences noted between the group 1 data and the
other three groups for the right side and left front hemispherical impact tests at the 1.5 m
drop height. This was primarily due to the rather large changes in peak sensor forces
noted during the group 1 tests.
The largest number of significant differences between groups was found to occur during
the flat anvil testing. A total of forty linear contrasts out of a possible 60 contrasts were
found to be significantly different. All comparisons between the group 1 peak sensor
force change and the group 4 peak sensor force change were found to be significantly
different. Similarly, all comparisons between the group 2 peak sensor force change and
the group 4 peak sensor force change were found to be significant except for the left front
impact location at a 50 cm drop height.
When collapsed across drop heights, all comparisons between groups for the flat anvil
tests were found to be significantly different except for the group 1 and group 2
comparison at the right side impact location (see Table 43). In contrast, only 1
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
comparison across drop heights was found to be significant for the hemispherical anvil
tests ^ o u p 3 and group 4 - left front impact location).
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 4.04 0.51 4.94 not tested not tested
Group 1 V . Group 3 9.00* 1.14 5.12 not tested not tested
Group 1 V . Group 4 5.04 3.94 4.93 not tested not tested
Group 2 V . Group 3 5.03* 3.04 4.08 1.54 0.23
Group 2 V . Group 4 4.09 9.52* 0.23 0.49 2.44
Group 3 V . Group 4 2.24 4.08 1.18 1.58 4.11*
Table 40: Linear Contrast Analysis of Peak Sensor Force Across Groups and Drop Heights - Left
Front Impact Location, Hemispherical Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 2.89 5.07* 7.95* 1.37 5.54*
Group 1 V . Group 3 4.69 6.87* 7.16* 3.48 9.66*
Group 1 V . Group 4 5.60* 8.84* 16.98* 6.35* 15.49*
Group 2 V . Group 3 2.82 4.53 5.40* 2.85 7.27*
Group 2 V . Group 4 4.13 7.21* 14.15* 5.73* 13.64*
Group 3 V . Group 4 1.60 3.47 3.90* 2.64 5.42*
Table 41: Linear Contrast Analysis of Peak Sensor Force Across Groups and Drop Heights - Left
Front Impact Location, Flat Anvil (* indicates significance at a - .05)
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 4.79 1.65 17.83* not tested not tested
Group 1 V . Group 3 6.87 1.47 17.89* not tested not tested
Group 1 V . Group 4 3.99 1.30 17.81* not tested not tested
Group 2 V . Group 3 2.18 3.11 1.03 1.37 0.63
Group 2 V . Group 4 2.68 7.50* 0.47 1.38 1.86
Group 3 V . Group 4 1.83 2.52 0.14 0.19 1.44
Table 42: Linear Contrast Analysis of ^eak Sensor Force Across Groups and Drop Heig Its - Right
Side Impact Location, Hemispherical Anvil (* indicates significance at a - .05)
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparison 50 cm Ah 100 cm Ah 150 cm Ah 200 cm Ah Across All
Ah
Group 1 V . Group 2 7.00* 3.26 2.49 3.76 1.45
Group 1 V . Group 3 18.54* 5.10 7.74* 2.18 5.27*
Group 1 V . Group 4 24.32* 11.12* 8.37* 7.08* 8.85*
Group 2 V. Group 3 16.09* 4.19 7.08* 11.77* 6.13*
Group 2 V. Group 4 22.61* 10.86* 7.45* 17.57* 11.06*
Group 3 V. Group 4 10.74* 2.44 4.02 7.52* 4.69*
Table 43: Linear Contrast Analysis of i*eak Sensor Force Across Groups and Drop Heig Its - Right
Side Impact Location, Flat Anvil (* indicates significance at a - .05)
Correlation Analysis: Peak Headform Acceleration and Peak Sensor Force Change
In an effort to better understand the relationship between peak headform acceleration
and peak sensor force change, a percentage bend correlation procedure was performed to
test the hypothesis of independence between peak headform acceleration and peak sensor
force change. The results of this procedure are presented in Table 44. The data suggests
that for all groups, the peak headform acceleration and the peak sensor force as observed
at left front hemispherical impact location and the right side hemispherical impact
location are not independent. The only observation where the null hypothesis of
independence was not rejected occurred for the group 4 percentage bend correlation
which was found to be 0.07. When collapsed across all groups, peak headform
acceleration and peak sensor force change cannot be considered to be independent.
Left Front
Hemi
Left Front Flat Right Side
Hemi
Right Side Flat
Group 1 0.98* 0.80* 0.98* 0.88*
Group 2 0.90* 0.36 0.82* 0.69*
Group 3 0.59* 0.10 0.91* 0.60*
Group 4 0.07 0.19 0.56* 0.37
Across All Groups 0.78* 0.58* 0.72* 0.70*
Table 44: r Values Obtained From Percentage Bend Correlation Procedure (* indicates that null
hypothesis of independence was rejected with a < .05)
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It is interesting to note that the r values decreased as the helmet liner density increased.
The group 4 percentage bend correlation values were the lowest across all impact
locations and impact anvils while the group 1 percentage bend correlation values were
the highest across all impact locations and impact anvils.
In an effort to better understand the relationship between the peak headform acceleration
and the peak sensor force, a series of scatterplots were generated to show the relationship
between these two variables. These plots appear in Figures 22 through 25. The data
presented in Figures 22 and 24 (hemispherical impact tests) suggest that some relationship
may exist between peak headform acceleration and peak sensor force. However, the
results presented for flat anvil impact tests suggest that the relationship between peak
headform acceleration and peak sensor force is very poor.
As noted above, the tests for independence proved to be weaker as the helmet liner
density increased. The group 4 helmet liner data illustrated only one instance where the
null hypothesis of independence could be rejected (i.e., the right side hemispherical anvil
tests). Figure 26 illustrates the plot of the group 4 data for the right side hemispherical
anvil test data. Once again, the plot suggests that the relationship between peak headform
acceleration and peak sensor force is very poor.
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I I
100 200 3 0 0 4 0 0 5 0 0
P e a k H eadform A cceleratio n (g)
Figure 22: Scatterplot of Left Front Impact Location, Hemispherical Anvil Impact
Test Data
5 0 100 150 2 0 0 250
P e a k H eadform A cceleratio n (g)
Figure 23: Scatterplot of Left Front Impact Location, Flat Anvil Impact Test Data
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I
s
I
m o
c o
C O C M
I
1 0 0 200 3 0 0 400 5 0 0
P e a k H eadform A cceleratio n (g)
Figure 24: Scatterplot of Right Side Impact Location, Hemispherical Anvil Impact Test Data
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I
s
50 100 150
P e a k H eadform A cceleration (g)
200
Figure 25: Scatterplot of Right Side Impact Location, Flat Anvil Impact Test Data
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
g I
£
8
8 0 100 120 140 160
P e a k H eadform A cceleration (g)
Figure 26: Scatterplot of Group 4 Right Side Impact Location, Hemispherical Anvil Impact Test Data
UniForce Transducer Distribution Profiles
The largest benefit gained from the development of this unique transducer was the ability
to evaluate the load distribution patterns for a given bicycle helmet impact over time. A
three dimensional display of the load transmission pattern to the ISO test headform was
generated for each group for the trial with the greatest peak sensor force change across all
impact locations, drop heights and test anvils. Time series distribution patterns for the
group 4 helmet liners are illustrated in Figures 27 through 33 for the hemispherical anvil
and Figures 34 through 39 for the flat anvil impacts. In order to reduce the number of
graphs which were generated, only the data frame that contained the greatest peak sensor
force change was evaluated. It was felt that this trial would represent the worst case
scenario for a given test condition.
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
All graphs were generated using the surface fitting algorithm available with Harvard
Chart XL software. Each surface fit procedure used the x,y coordinates of each sensor
along with the sensor force change to form a three dimensional surface with color
gradations relative to the sensor force change. The color range was fixed at a maximum
of too N in order to discriminate between trials with peak sensor force changes above
100 N.
-p250
J.200
Force (N)
■ 100+
£.150
I 91 to 100
I 82 to 91
[lOO I 73 to 82
.50
■ 64 to 73
1 55 to 64
0
I 45 to 55
1 36 to 45
1 27 to 36
1 to 27
g 9 to 18
1 0 to 9
Figure 27: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at Contact (Headform Acceleration - 2.2 g)
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
-8 -8
Figure 28: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 2.2 ms after Contact (Headform Acceleration - 54.5 g)
-p250
4.200
Force (N)
■ 100+
|l5 0
1 91 to 100
1 82 to 91
[lOO 1 73 to 82
.50
I 64 to 73
■ 55 to 64
0
1 45 to 55
I 36 to 45
I 27 to 36
1 18 to 27
3 9 to 18
g Oto9
Figure 29: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 5.2 ms after contact (Headform acceleration — 100.8 g)
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
-8 -8
Figure 30: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 7.8 ms after contact (Headform acceleration - 133.8 g)
250
.200
150
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
Figure 31: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 10.4 ms after contact (Headform acceleration - 157.1 g)
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-r-250
-IlOO
Force (N)
■ 100+
1150
■ 91 to 100
■ 82 to 91
[io o ■ 73 to 82
.50
■ 64 to 73
■ 55 to 64
0
I 45 to 55
I 36 to 45
■ 27 to 36
I 18 to 27
8 9 to 18
0 Oto9
Figure 32: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 13 ms after contact ^eadform acceleration - 95.2 g)
250
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 33: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height Load
Distribution at 15.6 ms after contact (Headform acceleration - 18.6 g)
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
•8 -8
Figure 34: Group 4 Right Side Impact Location, Plat Anvil, 2.0 m Drop Height Load Distribution at
contact (Headform acceleration - 2.9 g )
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
I
9 to 18
0to9
-8 -8
Figure 35: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height Load Distribution at
2.6ms after contact (Headform acceleration - 124.9 g )
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure 36: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height Load Distribution at
5.2 ms after contact (Headform acceleration - 177.2 g )
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure 37: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height Load Distribution at
7.8 ms after contact (Headform acceleration - 185 g )
10 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 38: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height Load Distribution at
10.4 ms after contact (Headform acceleration - 47. Ig )
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 39: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height Load Distribution at
13 ms after contact (Headform acceleration - 14.2 g )
1 0 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The load distribution patterns for group 3 helmet liners when subjected to a
hemispherical impact test at the left front impact location are illustrated in figures 40
through 43. These data are presented to illustrate the change in the load distribution
profile as the drop height was increased for a given impact location and test anvil. A
complete set of graphs illustrating the load distribution profile at the time of peak sensor
force change across all groups, impact locations, drop heights and test anvils are located
in Appendices C and D.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
Oto9
Figure 40: Group 3 Load Distribution Profile for a 50 cm Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position, z axis represents Force)
1 0 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 41: Group 3 Load Distribution Profile for a 1.0 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position, z axis represents Force)
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
g 9 to 18
1 0to9
•8 -8
Figure 42: Group 3 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
11 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 43: Group 3 Load Distribution Profile for a 2.0 m Drop Height, Left Front Impact Location,
Hemispherical Anvü (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
The load distribution profiles seen above clearly indicate that as the drop height
increased, the change in peak sensor force increased. The area over which the forces was
apphed also tended to increase with increases in drop height. The sensor distribution
pattern indicated that at the front left impact location the forces were distributed in two
main regions which were separated by a channel (see Figure 40). Inspection of the
internal surface of the bicycle helmet revealed a concave vent channel on the internal
surface of the helmet which traveled near the center of the impact site. Any sensor which
was located on the headform beneath this concave vent would not experience any
loading due to the space created by the vent channel. This was clearly seen in Figures 40
and 41 where the load was clearly distributed into two regions on the headform.
The load distribution patterns for the 1.5 m drop height onto the hemispherical anvil are
presented by group in Figures 44 through 47. These graphs are presented to illustrate the
effect that helmet liner density has upon the load distribution pattern for a given drop
height and impact location.
I l l
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 44: Group 1 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 45: Group 2 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 46: Group 3 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N)
-i.200 ■ 100+
■ 91 to 100
i l 5 0
I 82 to 91
[lOO
I 73 to 82
■ 64 to 73
Iso I 55 to 64
0
I 45 to 55
I 36 to 45
I 27 to 36
I 18 to 27
■ 9 to 18
g Oto9
Figure 47: Group 4 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Hemispherical Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The data presented in Figure 44 illustrates the change in sensor force when a helmet liner
“bottoms out” due to an inability to effectively absorb the energy of the impact. It was
apparent that the high forces were distributed over a very small region near the center of
the impact, probably within 2 cm of the center of the impact. The eight radial portions
of the sensor can be clearly seen in this figure and as indicated above, the change in sensor
force outside the “bottom out” region was minimal (in the range of 20 to 30 N).
As the helmet liner density was increased, the magnitude of the forces increased and the
region of high forces (over 100 N) increased as well. The presence of high localized forces
for the group 2 load profile distribution suggested that this particular helmet was rapidly
approaching its energy absorbing limit. The load distribution profile for the group 3
helmet indicated that the magnitude of the local forces for this particular test were less
than those seen for either the group 1 or the group 2 helmets. Consequently, the risk of
local skull bending and subsequent brain injury was likely reduced due to the fact that
the helmet liner had not bottomed out.
A comparison of the load distribution profile for flat anvil impacts for the same group 3
helmets when tested at different drop heights is illustrated in Figures 48 through 51.
1 1 4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 48: Group 3 Load Distribution Profile for a 50 cm Drop Height, Left Front Impact Location,
Fiat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure 49: Group 3 Load Distribution Profile for a 1.0 m Drop Height, Left Front Impact Location,
Flat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 50: Group 3 Load Distribution for a 1.5 m Drop Height, Left Front Impact Location, Fiat
Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
- 8 -8
Figure 51: Group 3 Load Distribution Profile for a 2.0 m Drop Height, Left Front Impact Location,
Flat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The local contact sensor force changes for the flat anvil impacts appeared to be
distributed over a wider area than the hemispherical impact tests which were performed
at the same location. Once again, the data indicated that the sensor was sensitive enough
to illustrate the effect of the vent channel which was located immediately beneath the
impact site by clearly indicating two regions of load distribution.
As noted for the hemispherical impacts, the amplitude of the local sensor force change
increases as a function of drop height. However, for this particular group of helmets
^roup 3), the 2.0 m drop height shows a reduction in the ampHtude of the local sensor
force change and wider distribution of the local sensor force change as compared to the
1.5 m drop height test. A possible explanation for this could be the fact that this
particular helmet experienced a fracture of the EPS liner as a result of the flat anvil
impact from this height. The resulting fracture would effectively dissipate a great deal of
impact energy and the subsequent headform contact forces would be effectively reduced.
A similar phenomenon was observed for the 1.5 m and 2.0 m impacts for the group 4
helmets at this same location.
The effects of helmet hner density upon the load distribution profile for the flat anvil
impact tests are illustrated in Figures 52 through 55.
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure 52: Group 1 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Fiat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 53: Group 2 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Fiat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure 54: Group 3 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Flat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
•8 -8
Figure 55: Group 4 Load Distribution Profile for a 1.5 m Drop Height, Left Front Impact Location,
Flat Anvil (x,y axes represent sensor position in cm, z axis represents Force in Newtons)
1 1 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Once again, as the helmet liner density increased between group 1 and group 3 the
magnitude of the local sensor force changes increased. The change between groups was
most marked between the group 1 test and the group 2 test wherein there was Uttle or no
force change seen at the helmet headform interface for the group 1 data. However, for
the group 2 data, there was a well distributed impact which did apply some force to the
headform. Comparison with the group 3 data indicated that these forces then increased as
the helmet liner density was increased. The area of load distribution appeared to be
similar between both the group 2 and the group 3 flat anvil tests at this location.
The load distribution profile for the group 4 helmets during the flat anvil tests appeared
to decrease relative to the group 2 and group 3 load distribution profiles. As mentioned
earher, this may have been due to a better contour match between the helmet and the
headform which acted to increase the area of load distribution and to increase the
amount of liner deformation. However, this reduction in local contact forces was more
likely due to helmet liner fracture during the impact against the flat anvil.
Radial Analysis
In an effort to better understand the load distribution patterns of the different impacts,
an approach was developed to analyze the manner in which the load was distributed
throughout the quadrants of the UniForce transducer. Since the transducer had been
developed using an equally distributed radial system, the distribution of responses from
each sensor within a transducer were equally divided into four quadrants according to a
simple cartesian coordinate system. The overall sensor pattern radiated from a central
origin; therefore, the average sensor force change for a given radius could be computed
for each quadrant of the sensor. This resulted in six different radial values for each
quadrant. In this manner it became possible to evaluate the manner in which the load
was distributed over the surface of the headform. In particular, it was possible to evaluate
whether or not there was a specific headform loading pattern for a given impact location
or anvil configuration. If the load distribution pattern was symmetrical, then the radial
lines would all be of equal value and gradually decay as the radius increased. However, if
12 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the load distribution pattern was not symmetrical or if it was concentrated in one area
then there would be differences between each radial.
Since the orientation of the UniForce transducer was known relative to the headform
and relative to the helmet, this analysis also provided some insight into those portions of
the headform which would experience the largest change in sensor force due to a given
impact configuration. It was felt that this information could be compared to the human
head in order to identify areas which may be more susceptible to injury due to a lower
tolerance to skull fracture or bending.
Figure 56: Radial Analysis Quadrant Distribution For Left Front Impact Location
1 2 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 57: Radial Analysis Quadrant Distribution For Right Side Impact Location
An analysis of the tests conducted at the left front impact location against the
hemispherical anvil indicated that the predominant load path was through the first
quadrant which was the medial superior portion of the headform. Anatomically this
would correspond to the frontal bone. Figure 58 illustrates the radial loading pattern for
the group 2 helmets from a drop height of 50 cm against the hemispherical anvil.
Although not consistent across all groups and drop heights, the load path does appear to
generally travel superior and anterior relative to the original impact location on the
headform for those tests conducted at the left front impact location against a
hemispherical anvil. A complete set of hemispherical impact radial distribution patterns
for each group at the left front impact location are presented in Appendices E and F.
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Q u ad ran t
2nd Q u ad ran t
3rd Q u ad ran t
4th Q uadrant
9 30
R adius 0 R adius 1 Radius 2 R adius 3
R a d iu s L o c a tio n
R adius 4 R adius 5 R ad iu s 6
Figure 58: Radial Analysis For Group 2 Left Front Impact Location - Hemispherical Anvil, 50 cm
Drop Height
The radial analysis for a typical left front hemispherical impact suggested that the largest
change in sensor force did not occur at the center of the transducer (Radius 0) but rather
9.5 mm away from the center of transducer within the first and second quadrants. The
primary load pathway appeared to be through the first quadrant as indicated by the fact
that the change in sensor force for quadrant one remained high relative to the other three
quadrants. The majority of the force was applied within 38 mm of the center of the
transducer as indicated by the dramatic reduction in sensor force change at the fourth
radius.
Figure 59 illustrates the radial distribution pattern during a test wherein the helmet liner
experienced “bottoming out”. This particular trial indicates that the high forces are
concentrated at the center of the impact and are distributed over a very small area of
approximately 20 cm^ as indicated by the very low forces monitored at the second radius
(12.7 mm from center of impact) and beyond.
12 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3000
♦ Overall
— • — 1st Q u a d ran t
A 2nd Q u a d ran t
K 3rd Q u a d ran t
4tti Q u a d ran t
2500
2000
8 1500
1000
500
R ad iu s 6 R ad iu s 5 R adius 2 R adius 4 R adius 0 R ad iu s 1 R adius 3
Radius Location
Figure 59: Radial Analysis of Group 1 Impact From 1.5 m Drop Height, Left Front Impact
Location, Hemispherical Anvil
The radial analysis data for the flat tests conducted at the left front impact location
indicated that the impact forces were distributed over a wider area relative to the
hemispherical tests and they also tended to be distributed anteriorly relative to the
impact point. Figure 60 illustrates the radial distribution of a flat anvil impact at the left
front impact location from a drop height of 1.0 m. The data indicates that the load was
primarily distributed in the first and second quadrants; however, the load does also
appear to be somewhat evenly distributed through all quadrants as the radius increases
from the original point of impact.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1 st Q u a d ran t
2nd Q u a d ran t
3rd Q u a d ran t
4th Q u a d ran t
9 30
R adius 0 R ad iu s 1 Radius 2 R adius 3
R a d iu s L o ca tio n
R adius 4 R adius 5 R ad iu s 6
Figure 60: Radial Distribution of Group 1 Sensor Forces For a 1.0 m Impact at the Left Front
Impact Location, Flat Anvil
Once again, the peak sensor force changes were not seen at the center of the impact, but
rather within 12.7 mm of the center of the impact in the first quadrant and 9.5 mm from
the center of the impact in the second quadrant of measurement. This non-symmetrical
distribution could have been due to contour differences between the helmet and the
headform or due to the presence of the local venting chaimels on the internal surface of
the helmet. Visual inspection of the helmet did reveal the presence of local ridges at the
edge of the ventilation channel due to the production process and these ridges may have
caused areas of local stress which correspond to high local forces.
The radial analysis distribution for the right side impact location (flat and hemispherical
anvil tests) appear in figures 61 and 62. Given the subtle design differences of this helmet
model, in particular the differences in helmet radius at the two locations, it was expected
that a more symmetrical distribution pattern would appear for the right side impact test
results. The absence of adjacent ventilation holes and the presence of additional
polystyrene was also expected to facilitate symmetrical load distribution profiles. An
analysis of the radial data for the hemispherical impact tests indicate that the load was
12 5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
symmetrically distributed away from the center of the impact; however, there was a
predominant load distribution within the second quadrant. Relative to the headform,
this quadrant was located superior and posteriorly relative to the center of the impact.
Anatomically, this region would correspond to the upper parietal bone on the right side
of the skull. As illustrated in Figure 61 and as seen previously, the peak sensor force
change is not located at the center of the impact, but rather at a site immediately adjacent
to the center of the impact. Once again, this was most likely due to the contour matching
profile at this particular impact location. After this local maximum peak, the sensor
forces appear to be well distributed throughout all quadrants of the sensor, gradually
decreasing to a minimum value at the perimeter of the sensor.
Overall
1 st Q u adrant
2nd Q u adrant
3rd Q u adrant
4th Q u adrant
R 150
R adius 0 R adius 1 R adius 2 R ad iu s 3
R a d iu s L o ca tio n
Radius 4 R ad iu s 5 R adius 6
Figure 61: Radial Analysis For Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m
Drop Height
1 2 6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1 st Q u ad ran t
2nd Q u ad ran t
3rd Q u ad ran t
4th Q u ad ran t
o 3 0
R adius 0 R adius 1 R adius 2 R ad iu s 3
R a d iu s L o catio n
R adius 4 R adius 5 R adius 6
Figure 62: Group 2 Radial Analysis, Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
Analysis of the radial distribution patterns for the right side flat anvil impacts suggests
that the sensor forces are predominantly distributed in the first and second quadrants as
opposed to the third and fourth quadrants. This corresponds to a superior distribution of
the forces on the headform relative to the center of the impact. The sensor loads at the
right side impact location also appeared to be more symmetrical in their distribution as
opposed to the flat anvil tests conducted at the left front impact location. This was likely
due to the fact that there were no significant ventilation channels or ventilation holes in
the vicinity of the impact site. The only ventilation holes appeared in the distal portions
of quadrants one and two; however, the presence of these vent holes did not appear to
affect the overall distribution pattern in any great manner. The high sensor forces
observed in quadrant one may have been affected by the ventilation hole if there was a
sUght irregularity in the EPS in the vicinity of the ventilation hole. Any irregularity
which made one portion of the EPS more prominent than the other would result in an
area of local stress concentration. This may explain why the forces in quadrant one were
found to be higher than those observed in the other quadrants. There also may be a
1 2 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
better contour match between the bicycle helmet liner and the headform in this region;
subsequently, more of the impact energy was be absorbed in this region.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
chapter 5
DISCUSSION
Head injury has significant economic and emotional costs to society. The use of proper
head protection has been shown to be effective in reducing the risks of head injury due to
both direct contact and subsequent inertial loading of the brain. Current bicycle helmet
standards measure impact acceleration using an accelerometer mounted at the center of
gravity of a magnesium test headform. Although this technique is useful at evaluating the
general physical motion of the headform, the relative risk of head injury due to inertial
effects, and the magnitude of the total force applied to the headform, little is known
about the manner in which this load is applied to the headform. More importantly, very
little is known regarding the manner in which these forces can induce local skull bending
and subsequently induce brain injury due to local deformation and local contact forces.
In order to address the lack of information in this area, the present study was undertaken
to develop an accurate and reliable methodology for the evaluation of the load
distribution characteristics of bicycle helmet impacts. As a first step, a custom transducer
was developed exclusively for this project to measure local contact forces and although
the results indicate that there are some limitations with regards to the applicability of the
transducer to general local contact force measures, the transducer was found to be very
cost effective and able to monitor local contact forces observed during bicycle helmet
impacts without altering the physical interface between a bicycle helmet and test
headform.
As a method by which to evaluate this new transducer, a study was designed to evaluate
the effect of helmet hner density upon the load distribution characteristics of a bicycle
helmet impact using an identical bicycle helmet model fabricated in four different helmet
liner densities. Local contact forces were monitored in addition to headform acceleration
1 2 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
data collected from an accelerometer located at the center of gravity of the test headform.
The impact test results obtained from the new transducer as well as the accelerometer
were quite comparable to existing literature and found to be quite repeatable. Overall,
the peak headform acceleration and peak local contact sensor force values were found to
be significantly lower for the low density helmet liners when compared to the highest
density of helmet liners during low to moderate energy impacts. During the high energy
impacts against the hemispherical anvil, the lower density helmets bottomed out,
resulting in high local contact forces and high peak headform accelerations relative to the
higher density helmets. These results suggest that a tradeoff does exist in terms of the
protection offered by low density helmets at low to moderate energy impacts compared
to the performance of higher density helmets during the higher energy impacts.
Development of the UniForce Transducer
Space limitations and the need for a non-invasive transducer system required the
development of a thin layer technology which was capable of monitoring many
individual channels over a large area without significantly altering the structure being
monitored (i.e. the headform). Current designs for in-shoe pressure measurement were
found to be both costly and too slow for the purpose of capturing high speed bicycle
helmet impacts. For this reason, a thin layer resistive ink technology was developed and a
custom sensor was designed to capture up to 63 channels of force information at a sample
rate of 7500 Hz per channel.
Sensor Response Characteristics
Analysis of the transducer linearity indicated that the UniForce Transducer does not
remain linear throughout the entire measurement range; however, it is predictable. In
order to accurately calibrate the transducer, each sensor had to undergo a complete
calibration throughout the entire operating range of the transducer. A 2"‘ ^ order
polynomial curve fit was found to model the response of the transducer with a minimum
mean square error (MSE). The MSE of the high range sensors was found to be larger than
the MSE for the low range sensors. The reason for this may be due to differences in the
13 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sensor range but it was most likely due to incomplete coverage over the sensing area
while the force was being applied to the sensor. This mismatch between the area of force
application and the sensing area could possibly create large discrepancies in the sensor
output and the subsequent sensor calibration factor. Preliminary data obtained during
this study noted that depending upon the amount sensing area covered by the applied
force, the voltage output from the transducer does drops quite dramatically.
Limitations of the Measurement System
The methodology developed during this study overcame several of the limitations
presented by previous technologies. Existing technology (i.e. Fuji film) is unable to
monitor local contact forces in the time domain and in addition is unable to temporally
relate the magnitude of the local contact forces to the overall acceleration measurement
obtained at the center of gravity of the test headform. Furthermore, given the presence of
contours within the helmet, ventilation holes that distribute local forces in a unique
manner and helmet fractures during impact, it was felt that this transducer system would
provide much more additional information regarding local EPS response characteristics
when compared to the existing Fuji film technology.
Preliminary evaluation of the sensor used in this study indicated that the maximum
difference in sensor output response during loading to full scale and unloading to zero
could be as high as 17%. Fortunately, during short duration bicycle helmet impacts,
hysteresis does not become an issue due to the fact that the interpretation of the data is
largely confined to the loading phase of the impact rather than the unloading phase of the
impact. Future research in head injury might suggest that the rate of helmet unloading
(and consequently skull unloading may be an important factor; however, at the present
time no research is available to support this hypothesis.
Although there were some limitations to the transducer, its ability to monitor local
forces without significant modification to the test headform far outweighed any
disadvantages presented by non-linearity and poor hysteresis. The only difficulties with
the transducer appeared to be in the application of the transducer to the surface of the
13 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
test headform. Extreme care had to be taken in order to ensure that each sensor was well
affixed to the headform surface. If creases or bubbles appeared on the surface of the
transducer, there was a high probability of damage to an individual sensor during normal
operation. Sunilarly, if there was any excess two sided tape on the headform, the tape
would tend to adhere to the helmet being tested and then when that helmet was
removed, the mylar backing of the transducer would begin to peel away from the
headform. In at least one instance during the preliminary testing the mylar tape pulled
away from the headform and actually severed one portion of the transducer thereby
losing the signal transmission from six sensors.
In addition to the physical damage which could occur to the transducer during normal
operation, there appeared to be a finite lifespan for each individual transducer. It was
noted that after approximately 40 trials, the signals from the dominant sensors in the
center of the transducer tended to experience short circuits and in some cases failed to
respond to load input altogether. Given the fact that these sensors experience the highest
changes in force, the loss of information from these sensors could reduce the overall peak
local contact force reported during a particular test.
The number of cables required to supply the transducer with the appropriate driving
voltage and to monitor the sensor outputs must be considered a limitation of this
measurement system. A total of 63 channels were monitored during this testing;
therefore a total of eight, twelve conductor cables were required for the instrumentation
setup. This equated to a large collection of cables that were attached directly to the drop
frame. Although the drop frame was free to travel on two guide wires, the additional
mass presented by these cables may have created a shift in the headform center of gravity
of the system. This may result in the development of moments about the center of
gravity of the drop frame system. These moments would then cause the drop frame
system to rotate during the impact sequence. This rotation would result in a transfer of
kinetic energy into rotational kinetic energy. Therefore, the helmet would not be
required to absorb as much energy for a given impact, since some of the kinetic energy
has been converted into rotational kinetic energy. For this particular test series, it would
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have to be assumed that the amount of rotational kinetic energy would be consistent
across all tests at a given drop height and impact location; therefore, the conclusions
regarding helmet liner density load distribution would remain the same. Future
developments should consider the use of smaller gage cables or some form of fiber optic
signal transmission system. These two developments would likely significantly reduce
the mass of the cables and have less of an effect upon the behavior of the headform and
drop frame system.
One final limitation of this new measurement system was the amount of time which was
required to calibrate the entire system. According to the manufacturers
recommendations, each sensor was pre-conditioned at least ten times prior to calibration.
Following this, each sensor was then calibrated at ten discrete loads in order to generate
the cahbration coefficients for each sensor. For a 63 channel transducer, this resulted in
1260 tests prior to any actual tests. The typical duration for this type of calibration
processing was five to six hours. From a standards testing perspective, this represents far
too much time for a technician to spend preparing for a given test series. The
development of an automated servo control system which could sequentially move to a
specific sensor and calibrate that sensor could easily accomplish the same calibration
procedures automatically and likely with more precision.
Linear Headform Accelerations
The results of this study are in good agreement with the literature in that Hurt and
Thom (1985) found linear headform accelerations to range between 105 g and 190 g for a
group of bicycle helmets with EPS liners that were tested using the same drop frame
system that was used for this particular study. Tests conducted by Bishop and Briard
(1984) on a group of seven bicycle helmets with expanded polystyrene liners found mean
peak headform accelerations of 135 g to 137 g for flat anvil impacts from a drop height of
1.0 m. These results appear to be very consistent with the data obtained for the group 3
and group 4 1.0 m impacts onto a flat anvil, confirming an earlier hypothesis that the
group 3 and group 4 helmet liner densities were very close to current production bicycle
1 3 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
helmet liner densities. It should be mentioned that although similar, the data collected
from these tests include a large percentage of hard shell helmets, i.e., helmets with a very
stiff outer shell. Current production bicycle helmets have a thin shell material and the
response of these thinshell helmets are slightly different than their hard shell
counterparts.
The present study was unique in that identical helmets were fabricated from the same
molding process, the only difference being the density of the EPS helmet liner. This
effectively isolated helmet liner density as a variable for observation. As mentioned
above, the headform acceleration values increased with an increase in helmet liner
density. The statistical analysis consistently indicated significant differences in trimmed
mean peak headform acceleration between groups, particularly between the group 1
helmets (low density) and the group 4 helmets (high density). There was also a large
number of instances where there was a significant difference between the trimmed mean
peak headform accelerations for the group 2 helmets and the trimmed mean peak
headform accelerations for the group 4 helmets. There were only three instances where
there was a significant difference between the group 3 helmets and the group 4 helmets.
These were at the 50 cm and 2.0 m drop heights against the flat anvil at the left front
impact location and at the 50 cm drop height against the flat anvil at the right side impact
location. This may largely be due to the fact that there was not a very large difference in
the actual helmet liner density of these two groups; consequently, a large difference in
peak headform accelerations would not be expected.
It is interesting to note that when collapsed across drop heights, only three significant
differences in the trimmed mean peak headform acceleration were observed between
groups and these occurred only for the flat anvil impact tests. Group 1 was found to be
significantly different from both group 3 and group 4 when tested at the front left impact
location against a flat anvil. Group 1 was also found to be significantly different from
group 4 when tested at the right side impact location against the flat anvil. This would
suggest that in general the group 1 linear headform acceleration values were less than the
group 4 linear headform acceleration values. Relating this to the human system, it may
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be possible to suggest that the use of lower density bicycle helmets could reduce the peak
headform accelerations and consequently reduce the risk of brain injury. However, it
must be mentioned that the group 1 helmet hners were also found to “bottom out” at the
1.5 m drop height when impacted onto a hemispherical anvil. The trimmed mean peak
headform accelerations for these tests were found to be 472 g for the left front impact
location and 437 g for the right side impact location. Both of these values are well in
excess of the limits of contemporary bicycle helmet standards and would very likely
result in serious head injury.
The data obtained during this study does indicate that the group 1 helmets have lower
peak headform accelerations than the higher density helmets and consequently can be
considered to be very effective at absorbing impact energy up until the point in which
they bottom out. Given the consistent higher acceleration values as helmet liner density
increases, it becomes clear that for a similar change in kinetic energy, the higher density
helmets are not experiencing as much deformation when compared to the group 1
helmets. Returning to our work-energy theorem, this lower deformation equates to a
higher force on the helmet (and consequently a higher acceleration) for a given change of
kinetic energy.
Although these group 1 helmets do perform well at low to moderate energy impacts,
when compared relative to bicycle helmet standard test procedures they will not pass the
current test requirements. If on the other hand, these group 1 helmets are discussed in
light of the performance of bicycle helmets in real world accidents then they appear to be
a potential alternative to the current group of bicycle helmets. Research by Smith et al.
(1994) on a group of 72 bicycle helmets found that the majority of the helmet liner
deformation observed in the sample population could be replicated by conducting impact
tests onto flat surfaces from drop heights of 1.0 m or less. Although this research was
conducted on only a small group of helmets, it does suggest that some people would
benefit from lower density bicycle helmets. Accident involved damage replication
conducted by McIntosh and Dowdell (1991) for a group of 42 accident involved helmets,
found that a 1.5 m drop height was an accurate representation of real world accidents.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Therefore, although the test failure at the 1.5 m drop height would preclude the
introduction of bicycle helmets with very low densities (e.g. group 1 helmets), bicycle
helmet hners with a nominal density of 64 kg/m^ might present a suitable alternative to
the current density of bicycle helmets. Unfortunately, only two impact locations were
evaluated during this study; consequently, the behavior of the group 2 helmet liners
when impacted at other locations is not known. O f particular interest would be any
hemispherical anvil tests which would be directed at a vent opening. Given the minimum
EPS coverage in these locations, an impact to a vent opening would hkely represent a
worst case test for these particular helmets.
Current bicycle helmet standards are the driving force behind the development of
relatively stiff bicycle helmet liners. This phenomenon has been criticized by Mills and
Gilchrist (1991) wherein they found that the use of a hemispherical anvil forces the
manufacturers to develop bicycle helmets which have very high densities. Furthermore,
they concluded that this requirement does not maximize the potential for energy
absorption within the helmet, i.e. complete deformation of the helmet liner in order to
perform as much work upon the helmet as possible. They proposed that if the
hemispherical anvil test were eliminated, then bicycle helmets with much lower densities
could be introduced into the marketplace. Unfortunately, this assumes that all impacts
are going to occur onto relatively flat surfaces and although this represents approximately
70% of all instances of impact (Smith et al., 1994, McIntosh and Dowdell, 1991), it does
not represent the entire continuum of all impacts. Therefore, hemispherical anvils should
likely remain as part of the battery of tests to evaluate the performance of a bicycle
helmet.
The maximum acceleration limit of contemporary bicycle helmet standards has also been
criticized by several authors (McIntosh et al. 1995, Corner et al., 1985, Lane, 1989) as
being too high to be representative of the Umits of human injury. It has been suggested
that the maximum peak headform acceleration levels could be reduced to as little as 200 g
to promote effective reduction of head injury. Clearly the results of the tests obtained
during these studies indicate that bicycle helmets can be manufactured in such a manner
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that the peak headform acceleration values can be reduced to levels below a 200 g
threshold for injury. However, as mentioned above, the current demands of bicycle
helmet standards, in particular the aggressive hemispherical anvil test, do not permit the
production of lower density bicycle helmets.
Published hterature suggests that current production bicycle helmets are doing an
adequate job at protecting the head from significant injury (Rivera et al., 1996). There is
no question that helmets are an effective agent against head injury; however, much of the
field data collected indicates that head injuries still occur in approximately 25 % of
helmeted riders (Smith et al., 1990, McIntosh and Dowdell, 1992). The presence of an
instance of a head injury while wearing a helmet leads to the question of whether or not
the injury was due to excessive impact energy or due to fundamental helmet design. It
would be interesting to investigate the potential outcome of injury if lower density
bicycle helmets were used in place of the current density of bicycle helmets; however,
this becomes more of an issue of ethics given the potential risk of high energy failure
with the lower density bicycle helmets. Given the assumption that the general
distribution of head injuries while wearing a bicycle helmet is skewed to the left, i.e. a
larger percentage of low severity head injuries, a reduction in peak headform acceleration
as a result of a lower density bicycle helmet could potentially reduce the frequencies of
these head injuries. Conversely, the lower density bicycle helmet would also have a
reduced upper limit of performance; therefore, there would also likely be an increase in
the number of severe injuries while wearing the lower density bicycle helmet liner.
Given the relative socioeconomic costs of these latter injuries, this may not be an
acceptable compromise to the current situation.
Impact Duration
Impact duration was found to decrease as drop height increased and as the helmet liner
density increased. The trimmed mean duration was the longest for the hemispherical
impact tests and shortest for the flat anvil tests. The hemispherical anvil trimmed mean
durations ranged between 20.5 ms and 11.4 ms for the left front impact location and 12.7
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ms and 10.4 ms for the right side impact location. There were significant differences
observed between the group 1 helmet impact durations and all other helmet groups at the
1.5 m drop height; however, this was largely due to the fact that the group 1 helmet liner
had “bottomed out” resulting in a very high amplitude, short duration acceleration
profile. When collapsed across drop heights, the only significant differences between
groups were found between the group 2 and group 4 helmet hners at both locations and
between the group 2 and group 3 helmet liners at the right side impact location.
To date, no research has been made available to evaluate the impact duration
characteristics of bicycle helmet impacts. The only available information from the
literature indicates that an effective protective system is capable of increasing the duration
of the acceleration pulse to a maximum in order to distribute the impact forces over as
large an area as possible (Patrick, 1972). Furthermore, increasing the impact duration is
an effective means of reducing the maximum force applied to the headform when one
considers the impulse momentum relationship that occurs as the headform center of mass
velocity is brought to zero against the impact anvil. Since the impact velocity is
controlled by the drop height, the change in velocity times the mass of the headform (i.e.
the change in momentum) will be equivalent to the impulse applied to the test headform.
If the duration of the impact can be maximized, then the maximum force applied to the
headform will be reduced to a minimum. If on the other hand, the impact duration
cannot be increased, then the forces applied to the head must increase for an equivalent
change in momentum. This increased force equates directly to an increase in headform
acceleration.
The relatively large difference in impact duration between the two impact locations for
the hemispherical tests could be explained by the differences in helmet contours and
helmet thicknesses at each of these particular impact locations. The left front impact
location has a smaller radius as compared to the right side impact location and this may
increase the amount of time available for the helmet liner velocity to come to zero.
Furthermore, the helmet thickness at the front left impact location was found to be 32
mm nominally and 27 mm at the right side location. This increased thickness at the left
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
front location would most certainly increase the amount of material available for energy
absorption and potentially increase the impact duration.
The results of the statistical analysis would suggest that helmet liner density does play a
significant role in the impact duration observed during flat anvil impact tests. The softer
helmet liners appear to be more effective at increasing the impact duration and effectively
bringing the headform center of mass velocity to zero. Using the impulse-momentum
relationship once again, this increase in impact duration corresponds to a reduction in the
peak forces seen by the headform during impact tests with the lower density helmet
liners. Conversely, helmet liner density does not appear to play a significant role in
affecting impact duration for the hemispherical anvil tests.
Head Injury Criteria
This study found that the proposed HIC tolerance limit of 1000 was exceeded for both
the group 3 and group 4 helmet liners when tested against a flat anvil at both the right
side and left front impact locations from a drop height of 2.0 m. The only trimmed mean
HIC value greater than 1000 for the group 2 data occurred during the 2.0 m drop height
tests against the flat anvil at the left front impact location.
As expected, the trimmed mean HIC value for group 1 at a drop height of 1.5 m at the
left impact location against the hemispherical anvil was found to be in excess of the HIC
limit of 1000. Given the high level of acceleration observed as the helmet liner bottomed
out, it was understandable that this particular series of tests presented a very high risk of
head injury. What wfs not expected however was the relatively low HIC value obtained
from the group 1 tests conducted at a drop height of 1.5 m onto the hemispherical anvil
at the right side impact location. The trimmed mean HIC value for these particular tests
was 653, a value well below the proposed head injury limit of 1000. This finding is not
consistent with a trimmed mean peak headform acceleration of 438 g. Since the HIC
value is the computed by integration of the acceleration signal, the corresponding
duration of these particular tests must have been of a sufficiently short duration to keep
the HIC value to a nominal value. This finding could perhaps cast some doubt on the
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ability of the HIC algorithm to determine the risk of head injury in relatively short
duration impacts. The current recommendations proposed by the Society of Automotive
Engineers suggest that the HIC algorithm should only be applied for impacts which are
under 36 ms in duration (Prasad et al., 1985). These particular tests would clearly fall
under this category.
As with the results of the impact duration analysis, there is no published literature
available on the HIC values obtained during bicycle helmet impacts. The data obtained
during this study indicate that trimmed mean HIC values will increase with helmet liner
density and with drop height. The flat anvil HIC values tend to be greater than the
hemispherical anvil values, a trend which was also noticed during the analysis of the
acceleration data. There did not appear to be a large difference between the HIC values
obtained for the right side impact location and the left side impact location (regardless of
impact test anvil). This would suggest that of the two impact locations analyzed, one
impact location does not present a greater risk of head injury when compared to the
other impact location (although this is likely not true with respect to the human skull
and brain).
Although HIC is commonly used as a measure of risk of head injury in the automotive
injury, it has received criticism as a tool for evaluating the performance of helmets
(Newman, 1982). It has been suggested that since the acceleration response of the helmet
is merely a reflection of the mechanical response of the helmet and dropframe system,
HIC cannot be correlated to actual head injury. What is known however, is that there
remains some fundamental relationship between time duration and head acceleration.
The Wayne State Tolerance Curve clearly illustrates this finding by illustrating that high
levels of acceleration can be tolerated for short durations while lower levels of
acceleration can be tolerated for longer durations before the onset of head injury. Until
additional research becomes available to clearly define this relationship, HIC will remain
an indicator of the relative risk of head injury between different impact situations.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The development a HIC failure criterion of 1000 was based on data obtained from
human cadavers when impacted at a forehead impact location and extrapolated to the
Hybrid Ed test dummy. Developed in 1976 by General Motors, the Hybrid IE is
considered to be state of the art in impact research. The head of the Hybrid HI is made
up of a soft polyurethane outer skin which surrounds a hollow magnesium headform.
During typical automotive crash tests, triaxial accelerometers are mounted at the center
of gravity of the headform in order to monitor resultant head acceleration values and to
compute a HIC value. In our study, a solid magnesium headform was used as the impact
test headform. Given the fact that there is no outer skin on the magnesium headform, the
actual HIC values may be sUghtly higher than those that would have been obtained had a
Hybrid HI been used for this test series. Furthermore, there is no evidence available to
suggest that the HIC value is reliable for any impact location other than the forehead.
However, given these limitations, it must be assumed that any relative differences in HIC
values between helmet liner densities would have remained regardless of the test
headform or the impact location selected; therefore, the overall findings of lower HIC
values with lower helmet liner densities would have also been observed.
UniForce Transducer Measurements
The present study is unique in that it is the only research to date that has captured in the
time domain the local contact forces experienced by a solid test headform. Since the
relative risk of contact injury is related primarily to the magnitude of these contact
forces, only peak sensor contact force results are discussed although the duration over
which these forces are applied to the headform is critical in assessing the risk of head
injury due to contact injury.
The change in peak sensor force was found to increase as a function of drop height and as
a function of helmet liner density (Tables 36 through 39). There were only a few
exceptions to this general trend and they primarily occurred during the 1.5 hemispherical
impacts at the left front and right side impact locations using group 1 helmets. The peak
sensor force change for the left front hemispherical impact test from a drop height of 1.5
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m was found to be 2087 N while the peak force change for the same test at the right side
impact location was 4153 N. The trimmed mean peak force change for the left front
impact location is quite close to the suggested tolerance value of 2446 N for the
temporoparietal bone as suggested by Nahum et al. (1968). The value observed for the
right side impact location can also be considered to be very close, if not above, the
tolerance levels for skull fracture in that region. Based on these findings, it is suggested
that had these similar impacts occurred to helmeted humans wearing the same helmets,
the likely outcome would have been some form of skull fracture with potential
commensurate brain injury.
Tremendous variability in behavior of a particular helmet liner was observed when a
helmet experiences “bottoming out”. If the failure tends to be over a very small area,
then the local forces would tend to be very high; however, if the failure is distributed
over a slightly larger area, then the average load over the “bottomed out” area will be
much less. This latter situation likely occurred for several of the hemispherical tests given
the fact that the range of the peak sensor force change for the left front hemispherical test
at 1.5 m was from 1804 N to 2834 N. The range of peak sensor force change for the right
side hemispherical anvil test at 1.5 m was from 2527 N to 4390 N. A more detailed
analysis within this test series indicates for those tests with lower peak sensor force values
(i.e. 1800 to 2500 N) the next highest peak sensor force change after the peak value was
very similar to the highest sensor force change, suggesting a more distributed loading
pattern over a small area.
In general, the trimmed mean peak sensor force changes were greater for the
hemispherical anvil than for the flat anvil. This could potentially be explained by a
dramatic difference in the area of load application, with the hemispherical impact tests
being distributed over a smaller area compared to the flat anvil impact tests from the
same drop height. There did not appear to be a large difference in the peak sensor forces
when compared between left front and right side impact locations, once again supporting
the theory that relative to the helmet, one impact location does not present a potential
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for transferring higher sensor forces onto the headform relative to the other impact
location.
Statistical analysis revealed once again a large number of significant differences between
groups for the flat anvil impacts. When collapsed across drop heights, all groups were
found to be significantly different except for the group 1 and group 2 comparison at the
right side impact location. This lack of significance could be related to the fact that the
peak sensor forces for both group 1 and group 2 during flat anvil testing were found to
be below SON and almost inconsequential in terms of the risk of head injury due to local
contact since nearly all of the impact energy is absorbed through the finer deformation
process.
Comparison of the peak sensor force changes with the localized loading test results
obtained by Long et al. (1989) show very good agreement. The results of the group 4 tests
indicates that the for the hemispherical anvil tests at the left front impact location all
peak sensor forces were in excess of 200 N. The data observed for the other three test
configurations ranged between 121 N and 201 N. These values are only slightly lower
than the 196 N local peak forces measured by Long et al. (1989). The lower force values
observed during these tests could likely be due to differences in the test equipment or due
to differences in the contour match between helmet and headform. A better contour
match between helmet and headform would result in a better loading distribution
pattern and consequently a lower peak sensor force change.
Given the fact that each of the sensors used in this particular study covered an area of
0.32 cm*, this failure criterion would be approximately equivalent to a peak sensor force
value of 79.5 N. This assumes that an equivalent load of 245 N was distributed over a 1
cm* area surrounding the sensor. A review of the data found in Tables 36 through 39
suggests that none of the group 3 and group 4 helmets would be capable of meeting the
requirements of this test except for the group 3 results from the right side flat impact
location. It must be emphasized once again that this observation is based on extrapolation
of tests using a different anvil and a different procedure; furthermore, the assumptions
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regarding the distribution of the load over the sensing area may not be true. Both of these
factors would play a major role in determining whether or not a given helmet would fail
this proposed localized loading test.
The results of this study found that higher density helmets elicited higher local sensor
contact forces relative to the lower density helmets. However, the lower density helmets
experience catastrophic failure (as defined by sensor forces beyond the known limits of
skull fracture tolerance) when subjected to hemispherical anvil impacts at drop heights of
1.5 m. Consequently, any advantages that are offered by the lower density helmet in
terms of reduced head contact forces at lower impact energy levels are lost when the
impact energy gets moderately high. This would support the concept proposed by
Newman (1995) in that the material behavior of the higher density helmets may
represent the most ideal approach to helmet design technology. Higher contact forces
may be experienced at the lower energy impact levels as a tradeoff for the additional
protection offered at the higher impact energy levels. As long as these contact forces
remain below some threshold for brain injury (either due to local deformation or due to
skull fracture) then the outcome of a given head impact will be favorable. Although it
was not possible to obtain helmet liners which had a density greater than commercially
available products, it is highly likely that increases in helmet liner density will
correspond to increases in local contact forces during impact. A situation may develop in
the future wherein the use of high density bicycle helmets will result in high local contact
forces, which even at the lower energy impacts may be beyond the tolerance for brain
injury.
If the potential spectrum of impact energy was known or could be predicted with a great
deal of accuracy, one could easily determine the appropriate bicycle helmet liner density
to provide the maximum amount of protection against head injury. Unfortunately, this
is not the case and we can only speculate on the actual true spectrum of impact
exposures. However, it may be possible to introduce lower density helmets for children
due to the fact that they should not theoretically be exposed to extremely high impact
energy levels (although this is never completely true!). In the event of an impact, the
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
helmet liner will deform adequately and the magnitude of the contact forces to the head
will be minimized when compared to the existing bicycle helmet designs. This particular
approach has been suggested by several researchers (Comer et al., 1985 Lane, 1986) given
the fact that small children have very soft and pliable skulls. In the event of an impact to
a small child, it is highly likely that instead of work being done on the bicycle helmet
liner, the work will be done on the child’ s skull, thereby limiting the role of the bicycle
helmet.
Correlation of Peak Sensor Force Change With Peak Headform Acceleration
The results of the correlation analysis between peak sensor force change and peak
headform acceleration strongly suggest that there exists a need to incorporate peak sensor
force into contemporary bicycle helmet standards. As an example, the trimmed mean
peak headform acceleration data obtained during the right side hemispherical anvil
impacts from 1.5 m using the group 2 helmet liner was found to 187 g and the trimmed
mean HIC value was found to be 752 while the trimmed mean peak sensor force was
found to be 596 N. Consequently, even though the helmet may be considered to pass a
given contemporary helmet test standard (since the peak headform acceleration was
below 300 g and the HIC was below 1000), the local sensor forces were found to be high
and therefore present a potential risk of brain injury.
The example given above illustrates the need to consider local contact force measurement
as part of the evaluation process for bicycle helmet standards. In order to effectively
measure local headform contact forces during standard bicycle helmet tests, a specific test
procedure must be developed to evaluate the local contact forces. Perhaps a smaller
version of the transducer used during this study could be developed in order to monitor
local contact forces during typical standard bicycle helmet impacts. The results of the
radial analysis suggest that the peak sensor forces do not necessarily occur at the center of
the impact; therefore, the use of a single point load cell to monitor helmet load
distribution does not appear to be a useful solution.
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Distribution Analysis
The data presented in this study represents the only available research demonstrating the
temporal load distribution patterns of bicycle helmet impacts during impact. Previous
work by Long et al. (1989) was limited to the capture of peak forces using Fuji Film. The
current transducer which was developed exclusively for this project provides time based
information regarding the development and distribution of a typical bicycle helmet
impact onto a flat anvil and onto a hemispherical anvil. Furthermore, the transducer
developed for this project was designed to illustrate the development of local contact
forces during the entire impact sequence, behavior that cannot be observed using Fuji
film.
The results of this study clearly show that tests conducted using both test anvils result in
two distinct impact profiles. As originally hypothesized, the hemispherical anvil impacts
tend to be localized while the flat anvil impacts tend to be more distributed. The
localized sensor forces tend to be higher for the hemispherical impacts, suggesting that
these type of impacts (i.e. aggressive impact surfaces) pose the greatest threat to the
helmet user in terms of head injury risk due to high local contact forces. As the forces
increase, the amount of local skull bending will increase and the potential for localized
trauma at the skull/brain interface will also increase. From a protective point of view,
the lower density helmet liners appear to be far more effective at reducing the local
sensor forces applied to the headform. However, this reduction in local contact forces
does come with a tradeoff in that there is an upper limit to the performance of these
helmets which must be considered to be in the range of foreseeable impacts while
wearing a bicycle helmet.
Comparison of the data with the recommended tolerance threshold for brain injury of
245 N (25 kgF/cm^ for a 29 J impact against a hemispherical anvil (Long et al., 1989)
indicates that peak local contact forces for both the group 3 and group 4 bicycle helmet
liners would be in excess of this threshold limit. This result assumes that the load seen
over a 0.32 cm^ sensor area can be equated to the recommended tolerance criteria. Closer
1 4 6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
examination of the distribution response for the hemispherical anvil tests suggests that
although the local forces at one particular sensor may exceed the recommended pass/fail
criterion, the amplitude of those sensors in close proximity to the maximum sensor were
of a smaller amphtude. This would seem to suggest that for hemispherical impact tests,
the local forces are focused over a very finite area and then tends to decrease rapidly in
magnitude. If the decay of the force (in a three dimensional sense) is rapid, then the local
forces may not exceed 245 N (25 KgF/cm^ over a 1 cm' area. This is most likely what
occurs during real life accident situations given the fact that very few instances of skull
fracture are observed when the subject is wearing a properly fitted bicycle helmet during
an accident (Hurt and Thom, 1994, Smith et al., 1994, McIntosh and Dowdell, 1992).
The data obtained by this study strongly shows a trend towards higher local contact
forces as the density of the bicycle helmet liner is increased. Consequently, further
increases in helmet liner density will most assuredly increase the possibility that skull
fracture may occur during some impact events. In particular, given the presence of high
sensor forces during low level impacts, accidents involving higher density bicycle helmets
may cause excessive skull bending and possibly skull fracture even during minor impacts
onto flat surfaces. Clearly a better understanding of this phenomenon will require
additional research with helmet samples which possess liner densities that are higher than
those that were evaluated during this study.
An additional confound to the risk of skull bending and skull fracture is the presence of
venting channels and ventilation openings. A bicycle helmet by its very nature possesses
ventilation holes to facilitate air flow and reduce local temperature buildup within the
helmet. Unfortunately, these design features tend to limit the available area for load
distribution and energy absorption and consequently may result in very high loads being
placed on the headform over a very small area. Although there may be no forces applied
to the head in the area of a vent hole, the surrounding area must distribute those forces to
the headform. Furthermore, due to the polystyrene production process, the EPS beads
near ventilation openings may not be completely expanded; consequently, they may be
quite stiff relative to the overall helmet. This potentially could correspond to a reduced
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ability to absorb impact energy and could also potentially result in high local forces in
the vicinity of the ventilation holes. More research is necessary to fully understand the
behavior of load distribution near ventilation holes and ventilation channels.
Regardless of the presence or absence of ventilation holes, the results of the radial analysis
suggest that the sensor forces tend to distribute themselves towards the center of the
skull. This is a more desirable situation than one in which the forces distribute
themselves towards the more sensitive temporoparietal bones of the skull. The results
also indicate that helmet impacts are not symmetrical in their distribution about the
center of the impact. They are particularly affected by the presence of ventilation
channels and ventilation openings that are part of the design of the bicycle helmet. The
presence of ventilation chaimels and ventilation openings result in local “high spots” due
to the polystyrene production process. Under impact situations, these “high spots” result
in areas of high localized loading which in turn can cause large amounts of local skull
bending. This local skull bending corresponds directly to an increase in risk of brain
injury due to local contact. This becomes particularly important at the lower edges of the
helmet which may be in close proximity to the temporoparietal bone. Given the fact that
the middle meningeal artery is embedded within this bone, any excessive skull bending in
this region will increase the possibility of local arterial rupture and possible subdural
bleeding. Based on the findings of this study, future bicycle helmet designs should make
every effort at reducing the number of ventilation openings and ventilation channels in
order to minimize the potential for local high stress areas.
Another factor affecting the load distribution characteristics of a bicycle helmet impact is
the contour match profile between the headform and the test helmet. As indicated
earlier, the test samples were fabricated from twelve separate cavities; therefore, there
were twelve unique helmet contour profiles being tested. Although an assumption must
be made regarding the homogeneity of the sample population, it is possible that subtle
variations between cavities do exist. These variations could be a result of the EPS cooling
process or they could be due to variations between the cavities themselves. It is possible
that these variations could affect the contact regions between the headform and the test
1 4 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
helmet. Subsequently, the load distribution pattern and the radial profile could vary
simply as a function of the differences between the match of helmet contour and
headform contour. A small alteration in the placement of the test helmet could also affect
the radial distribution for a given test, although every effort was made to place the
helmet on the test headform in a consistent manner. These variations likely represent the
real world situation wherein a large number of people will be wearing the same helmet in
a variety of different ways; consequently, the load distribution profiles may vary from
person to person depending upon their individual anthropometric characteristics.
The test data indicated that as the impact energy increases, the sensor forces tend to
increase and the area of the load distribution also tends to increase. The only instance
where this did not hold true was during the left front impact location tests using the flat
anvil. The group 3 and group 4 tests from 1.5 m and from 2.0 m drop heights showed
dramatically reduced local sensor forces. As indicated, this was likely due to significant
helmet fracture during the impact sequence. As the helmet liner fractures, energy is
dissipated through the process of crack propagation throughout the helmet.
Consequently, the local sensor forces would be dramatically reduced and the relative risk
of brain injury from skull bending would also be reduced. The data observed during
these tests is the first representation of the benefits of helmet fracture during impact
situations.
Helmet fracture was not observed during those tests conducted at the right side impact
location. This may be due to the differences in helmet radius at these two particular
impact locations. It may also be due to the proximity of the left front impact location to
both the edge of the helmet and to vents and ventilation channels. The presence of
ventilation channels create an area of high local tensile stress during impact situations and
since EPS beads possess a bond which is very weak in tension, the helmet impac: near the
edge of the helmet may cause high tensile stresses and subsequent helmet fracture.
1 49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Current Limitations In Head Injury Research
A fundamental purpose of the present study was to develop and validate a reliable system
for monitoring local contact forces during helmeted head impacts. It was expected that
this instrumentation would provide some insight into the behavior of bicycle helmets
during impact situations and a potential means for evaluating helmet performance under
standard test conditions. The use of solid magnesium headforms as a means to monitor
head acceleration and local contact forces most certainly affects the ability to relate the
information gathered to the human head and brain. It is known that the human head is
not a solid rigid object, but rather a non-rigid system which can undergo significant
deformations during impact situations (Nusholtz, 1984). These skull deformations
effectively contribute to the energy absorbing process; consequently, the use of a solid
magnesium head must be considered to represent a worst case scenario in terms of head
impact response.
There are headforms currently available which exhibit more biofidelic characteristics
than a solid magnesium headform. The Hybrid E H headform is used regularly in the
automotive industry and the Wayne State University Headform or NOCSAE headform
is used routinely for football and baseball helmet testing in the United States. Although
these headforms were originally considered for this project, it was decided that the best
approach was to adopt the instrumentation commonly used for bicycle helmet standard
testing. In this way, the effect of bicycle helmet liner density could be discussed in light
of the effect that its manipulation would have on the ability of a helmet to meet one or
more of the impact test criterion for bicycle helmet standards. As well, the presence of
the soft outer layer on both of the biofidelic headforms would certainly affect the
performance of the UniForce transducer. In particular, the transducer would likely
respond to both normal force application as well as to bending stresses as a result of
compression of the outer skin of the test headform.
Although this study measured local contact forces on a headform, the problem still exists
to correlate these local contact forces to the relative risk of brain injury. Typical cadaver
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
research is capable of evaluating only the tolerance threshold for injuries that can be
observed during gross autopsies (i.e. lacerations, fractures and some contusions). It is not
possible to determine the presence of brain injury, particularly concussive injury, in a
cadaver simply due to the fact that in many situations, there is no obvious clinical
damage to the skull or to the brain. Proper diagnosis of concussive brain injury requires
clinical observation following the impact event. This poses an obvious ethical problem
for human subjects; therefore, in many cases animals have been used as experimental test
subjects. This presents a secondary problem as the subtle anatomical and inertial
differences between the animal model and the human model require the extrapolation of
the experimental findings to the human.
Given the fact that the present study employed a soUd rigid headform, there is no
possible means available to directly extrapolate these findings to the human. Available
data which compares cadavers and rigid headforms under identical impact conditions has
clearly indicated that the rigid headforms are far stiffer than there human counterparts
and consequently, data obtained with these headforms is considerably higher than data
obtained under the same conditions with cadavers (Hodgson et al. 1970, Webster and
Newman, 1976). Given this information, it was assumed that the differences noted
between helmet liner groups would also appear if the exact same procedures were
repeated with human cadaver subjects, only the magnitude of the forces would change.
In addition to linear acceleration measurements, angular acceleration, when coupled with
linear acceleration, has clearly been identified as a primary mechanism for concussion
(Gennarelli et al., 1972). Unfortunately, to date, a suitable test methodology has not been
developed to evaluate angular acceleration of helmeted impacts. Current helmet test
methodologies are restricted primarily to guide wire systems which control the travel of
the drop frame apparatus and permit accurate measurement of linear acceleration effects.
Future development of helmet test standards (and head injury biomechanics in general)
must include angular acceleration instrumentation in an effort to try to understand the
effects of helmet performance and to try to develop new helmet technologies which may
reduce angular acceleration effects (Ommaya, 1985). At the present time, only the British
1 5 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Standards Institute is developing test instrumentation to evaluate the angular acceleration
response of motorcycle helmets (Mills, 1997).
In addition to a lack of understanding regarding helmet response to angular acceleration,
there is a tremendous lack of detailed information regarding actual helmet performance
in real world accidents. Many studies have clearly indicated that the use of a bicycle
helmet can significantly reduce the likelihood of head injury (Thompson et al., 1989,
Baker et al., 1993, Rivara et al., 1996); however, no study has been able to provide a
detailed correlation between helmet damage and clinical injury. Unfortunately, until
only recently, the number of helmeted bicyclists was as low as 18% of all cyclists
(Rodgers, 1993). This corresponds to a relatively low number of helmeted cyclists who
would appear at a hospital for medical treatment. Furthermore, current accident
databases in the United States (Fatal Accident Reporting System, FARS and National
Electronic Injury Surveillance System, NEISS) only record injury information if a motor
vehicle is involved. Consequently, the number of solo helmeted bicycle riders with
clearly documented accident and injury information remains limited.
The field data indicates that bicycle helmet usage rates are increasing (Vulcan et al., 1992,
Bicycle Helmet Safety Institute, 1997) and as a result, the number of helmeted bicycle
riders who appear at hospitals for medical treatment is also increasing. As this number
increases, so too does the number of helmeted bicycle riders experiencing head injury.
Research has suggested that the percentage of head injured helmeted riders is
approximately 22% to 25% for adults (McIntosh and Dowdell, 1992, Smith et al., 1994)
and has been found to be as high as 76% for children (Grimard, 1995). This leads to the
question of whether or not the incidence of injured helmeted bicycle riders represent a
failure or success of modern bicycle helmets? Are these head injuries due to poor
performance of the helmet or are they due to the excessive impact energy?
There are two components to the answer to these questions. Firstly, the data obtained
from field accident research suggests that current bicycle helmets are not effectively
absorbing a maximum amount of impact energy during accident situations. This is
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reflected in the low values of liner compression and small areas of impact distribution as
measured on actual accident involved bicycle helmets, particularly those riders who
sustained some form of head injury (Smith, 1997). If the helmet does not exhibit a high
percentage of compression (relative to its original thickness) then there is material
available for energy absorption that has not been efficiently utilized. If a head injury were
involved, this would suggest that there were local contact forces present sufficient to
induce some form of brain injury, as opposed to an effective load distribution over a
large area. The data obtained during this study would support this finding in that higher
density bicycle helmet liners (such as those used in contemporary helmets) produce
higher local contact forces and higher headform acceleration levels.
Similarly, the data obtained during this study suggests that the introduction of lower
density bicycle helmet liners could reduce headform acceleration values, HIC injury
indices as well as local contact forces for low to moderate energy impacts. This could
potentially translate to a net overall reduction of head injuries if one is to assume that the
general distribution of all head injuries is proportionately skewed to the lower severity
head injuries.
Unfortunately, there are a limited number of databases which categorize helmeted
bicycle riders who have been involved in accidents. Consequently, it is not possible to
evaluate and understand the overall continuum of bicycle related accidents and the role of
bicycle helmets in those accidents. Very often the epidemiologist is only interested in the
injury classification or type rather than the cause of that injury. If accurate injury
information was gathered along with a detailed examination of the helmet damage, then
it would be quite simple to determine whether or not a head injury was due to some
hmitation of the physical characteristic of the bicycle helmet (i.e. helmet liner did not
absorb enough energy) or due to the fact that the amount of energy applied to "he helmet
during the impact was in excess of the material characteristics of the helmet. If the former
situation were true, then the use of a bicycle helmet does not represent a totally effective
countermeasure to head injury. Appropriate steps could be taken to develop test
standards which would evaluate the loading characteristics of bicycle helmets relative to
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
helmet stiffness in an effort to reduce local contact forces and inertial forces. It would be
expected that the instrumentation developed for this project would be well suited for this
task. If on the other hand, the latter situation were true then future developments in
helmet standards would increase the energy absorbing requirements of the helmet.
The problem of collecting and analyzing helmets involved in real world accidents cannot
be limited to the clinical environment. Many bicycle riders reap the benefits of proper
head protection and do not experience any injuries as a result of a bicycle accident.
Consequently, they do not need to appear at a medical facility for treatment and their
favorable outcome does not appear in any accident database. Each year bicycle helmet
manufacturers receive hundreds of such low severity cases as part of their helmet return
programs; therefore, the exact number of non-injury cases involving bicycle helmets
remains unknown. Research by Rivara et al. (1996) have estimated this number to be
approximately 1.2 million cases per year. It is imperative that this relatively large group
be considered before invoking any significant change to an existing bicycle helmet
standard. For example, any increases in impact energy requirements may result in
helmets that potentially could increase the probability of head injury for these lower
severity impacts. The increased risk would then theoretically be reflected in an increase
in hospital admissions of helmeted bicycle riders and this would be direct evidence that
the proposed standard change did not reduce the frequency and severity of head injuries
but rather caused them to increase.
In conclusion, it can be seen that there a many limitations to the present level of
understanding of head injury. The current study has addressed the issue of contact injury
and developed a custom transducer to monitor local contact forces during bicycle helmet
impacts. The results indicate that under certain circumstances, lower density bicycle
helmet liners result in a lower risk of head injury as measured by linear headform
acceleration, head injury criteria, impact duration and load concentration. However,
there are certain circumstances under which these lower density bicycle helmets do not
perform as well as their higher density counterparts. Consequently, situations do exist
where a lower density helmet is not as effective as current bicycle helmets available on
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the market. This may not be an acceptable situation given the fact that there is no way to
accurately predict the impact surfaces, and impact energy which are present during a
typical bicycle accident. More research is necessary to better understand the role that the
helmet plays in a real world accident situation.
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ch a p t e r 6
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
The purpose of the preceding study was to design and validate a custom transducer
system for the evaluation of load distribution patterns during typical helmeted bicycle
impacts. It was expected that the development of this transducer and the evaluation of
the effect of helmet liner density could potentially provide some indication regarding the
potential for reducing head injury during typical bicycle accidents. The effect of helmet
hner density upon the load distribution and impact attenuation characteristics of bicycle
helmet impacts was also measured.
The results of the instrumentation evaluation indicated that the UniForce transducer was
accurate and rehable when measuring local contact forces applied to an ISO test
headform. The phasic characteristics of the transducer matched a linear accelerometer
located at the center of gravity of the test headform. The transducer was found to be non
linear in terms of its output relative to a known input force and although this was not an
ideal situation, the application of a 2 " * ^ order polynomial curve fit was found to provide
accurate calibration results through a wide range of forces. The hysteresis characteristics
of this transducer were found to be very poor; consequently, the system was felt to be
inaccurate under test conditions where force loading and unloading parameters are of
interest. Overall, it was felt that there were some limitations to the transducer system;
however, the transducer does represents a low cost alternative to current load
distribution systems and furthermore provides unique information that has not been
previously obtained.
The impact test data collected during this study indicated that the lower density helmets
produced significantly lower peak headform accelerations relative to the highest density
of bicycle helmet liners under low to moderate energy impacts. Headform acceleration
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
values from hemispherical anvil tests were generally lower than those tests conducted
from the same drop height and at the same impact location onto a flat anvil. This would
suggest that both helmet liner density and impact surface curvature play an important
role in determining headform acceleration and consequently, the relative risk of head
injury.
Impact duration was found to decrease with increases in helmet liner density except for
those tests conducted with the lowest density of helmet liner at a 1.5 m drop height
against a hemispherical anvil. The hemispherical impact durations in general were greater
than the flat anvil impact durations and significant differences were found between all
groups for the flat anvil tests when collapsed across all drop heights. This would suggest
that lower density bicycle helmet liners could be effective at increasing impact duration
and potentially reducing headform acceleration for a given impact energy.
HIC values were found to increase with increases in helmet liner density and HIC values
at the flat anvil, 2.0 m drop height tests for the 80 kg/m^ and 98 kg/m^ were in excess of
the recommended HIC tolerance level of 1000. The HIC values were also in excess of
1000 for the hemispherical anvil, left front impact location, 1.5 drop height tests with the
lowest density helmet liners (35 kg/m^. HIC values for the flat anvil tests were found to
be higher than the HIC values for the hemispherical anvil tests. Significant differences
were noted between the low density helmets and the high density helmets for the flat
anvil impacts. There were significant differences in the HIC values for the hemispherical
impacts conducted at the low to moderate impact energy levels; however, this difference
disappeared as the impact energy levels increased.
Peak contact sensor forces were found to be below the published threshold levels for
skull fracture during low to moderate impact energy levels. This was not surprising in
that very few instances of skull fracture have been reported for bicycle riders who have
been involved in accidents while wearing a bicycle helmet which meets one or more of
the current bicycle helmet standards. In those instances where the helmet liner
experienced material failure, the peak contact sensor force did exceed the pubhshed
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
thresholds for skull fracture. Although the actual tolerance threshold for brain injury is
not known, the lower peak contact sensor forces observed for the lower density helmets
strongly suggest that the risk of brain injury due to local contact stresses is less when
compared to the higher density helmet liners. Correlation analysis of peak sensor force
and peak headform acceleration suggests that peak sensor contact force is independent of
peak headform acceleration. The correlations were found to decrease with increases in
helmet liner density. This finding supports the concept that future bicycle helmet
standards should attempt to incorporate load distribution criteria.
The load distribution patterns for each particular impact appeared to be affected by local
contour matches between the helmet and the test headform. The transducer was sensitive
to changes in the inner contour pattern of the helmet in that the transducer was able to
monitor local contact forces in and around venting channels and ventilation holes.
During a typical bicycle helmet impact in a laboratory, the forces move superiorly across
the top of the headform as opposed to inferiorly towards the edge of the bicycle helmet.
The contact forces do not distribute radially from the center of the impact as originally
hypothesized. The superior load path reduces the probability of skull fracture due to the
fact that the bones located in the upper portion of the skull are more tolerant to fracture
as opposed to the bones in the lower portion of the skull (i.e. temporoparietal bone). Due
to the design of the twin wire impact drop frame system, it is not known if these same
distribution patterns would exist during a bicycle accident with a human subject.
Conclusions
1. The custom transducer designed for this project represents an effective means of
monitoring local contact forces on an ISO test headform during typical helmet
impacts.
2. Impacts with the lower density bicycle helmets resulted in lower peak headform
acceleration values, lower HIC values and greater impact durations for low to
moderate energy impacts. This would suggest that at these levels of impact energy, a
bicycle helemt with a lower density liner has a lower risk of injury due to excessive
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
head acceleration when compared to a bicycle helmet with a higher density liner. At
high energy impact levels against curved surfaces, the bicycle helmets with higher
density liners result in lower peak headform acceleration values relative to those
helmets with lower density liners. Consequently, under these impact situations, the
helmets with the higher density liners offer a greater amount of protection against
head injury.
3. Impacts with the lower density bicycle helmets resulted in lower changes in peak
force over those conditions which did not exceed the material limitations of the
lower density helmet liner. When the material was compromised, the peak forces
were in excess of known tolerance limits. It is not known whether or not the forces
seen during these tests are in excess of the tolerance limits for brain injury due to
contact. It is known that a higher local contact force does present a greater potential
for brain injury.
4. During high level impacts, the higher density helmets performed better than the
lower density helmets in that they were effectively able to manage the impact energy
without bottoming. The presence of helmet liner bottoming during low density
helmet liner testing caused high impact forces to be transmitted directly to the test
headform. These high impact forces can be related directly to a high risk of head
injury.
5. The contact forces observed for the lower density bicycle helmet liners at low to
moderate impact energy levels indicates that bicycle helmets manufactured at these
densities would provide a lower risk of brain injury than contemporary bicycle
helmets which are manufactured at higher helmet liner densities.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Recommendations
1. The process to calibrate the transducer should be automated and future research
should continue to explore other transducer technologies that may provide better
mechanical and electrical response characteristics.
2. Future research should include a wider range of bicycle helmets, more impact
locations and additional test anvils in order to develop a database regarding the load
distribution characteristics of bicycle helmets currently available. Specific attention
should be paid to internal contour and ventilation designs and their affect upon load
distribution characteristics.
3. Future bicycle helmet standards should support the development of lower density
helmet liners. This would reduce the risk of brain injury at low to moderate impact
energy levels by reducing the local contact forces as well as acceleration based
measures. Due to the behavior of these lower density helmets at higher impact energy
levels, a separate standard could possibly be introduced for those bicycle riders who
anticipate being involved in higher energy impacts.
4. Future research on accident involved bicycle helmets must include a detailed
examination of the bicycle helmet liner in order to adequately characterize the
helmet’s performance and the relationship of performance to the injuries sustained by
the wearer.
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIOGRAPHY
Aldman, B. (1984). A method for the assessment of the load distributing capacity ofprotective
helmets proposed to replace the current resistance to penetration test. Paper presented at the
International Research Council on the Biokinetics of Impact.
Allsop, D. L. (1993). Skull and facial bone trauma: experimental aspects. In A. M.
Nahum & J. W. Melvin (Eds.), Accidental Injury: Biomechanics and Prevention, (pp. 247-
267). San Diego, CA: Springer-Verlag.
American National Standards Institute. (1984). American National Standard for
Protective Headgear - for Bicyclists: ANSI Z90.4-1984 . New York, NY: American
National Standards Institute.
American Society For Testing and Materials. (1997). Standard Specification for Protective
Headgear Used in Bicycling: ASTM F1447-94 . Philadelphia, PA: American Society For
Testing and Materials.
Anderson, J. E. (1978). Grants Atlas of Anatomy. Seventh Edition. Baltimore, Maryland:
Williams and Wilkins Company.
Association For The Advancement of Automotive Medicine and Volvo Car
Corporation. (1987). Head Injury Mechanisms. New Orleans, LA.
Bandak, F. (1995). On the mechanics of impact neurotrauma: a review and critical
synthesis. Journal of Neurotrauma /2(4), 635-649.
Bernhardson, C. (1975). Type I error rates when multiple comparison procedures follow
a significant F test of ANOVA. Biometrics, 31, 719-724.
Bicycle Helmet Safety Institute. (1997). States, Counties and Localities With Mandatory
Bicycle Helmet Laws . Washington, D.C.: Bicycle Helmet Safety Institute.
Bishop, P. J., & Briard, D. B. (1984). Impact performance of bicycle helmets. Canadian
Journal of Applied Sport Science, 9, 94-101.
Buntain, W. L. (1985). Bicycle injuries in children: an economic insight and literature
review. Paper presented at the 2 9'*’ Annual Meeting of the Association for the
Advancement of Automotive Medicine, Washington, D.C.
Canadian Standards Association. (1989). Standard for Protective Headgear for Cyclists:
CSA D 113.2-89 . Toronto, Ontario, Canada: Canadian Standards Association.
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cavanagh, P. R., & Ae, M. (1980). A technique for the display of pressure distributions
beneath the ioot. Journal of Biomechanics, 13, 69-75.
Consumer Product Safety Commission (1997). Proposed Rule: Safety Standard for Bicycle
Helmets. Washington, DC: Federal Register.
Comer, J. P., Whitney, C. W., O'Rourke, N., & Morgan, D. E. (1987). Motor cycle and
bicycle protective helmets: requirements resulting from a post crash study and experimental
research. (CR 55): School of Civil Engineering, Queensland Institute of Technology.
Denny-Brown, D., & Russell, W. R. (1941). Experimental Cerebral Concussion. Brain,
64, 93-164.
Dorsch, M. M., Woodward, A. J., & Somers, R. L. (1984). Ejfect of helmet use in reducing
head injury in bicycle accidents. Paper presented at the 28'*' Annual Conference,
Association for the Advancement of Automotive Medicine.
Evans, P. G., Lissner, H. R., & Lebow, M. (1958). The relation of energy, velocity, and
acceleration to skull deformation and fracture. Surgery, Gynecology and Obstetrics, 107,
593-601.
Fife, D., Davis, J., & Tate, L. (1983). Fatal injuries to bicyclists: the experience in Dade
County, Florida./o«r72^/ of Trauma, 23, 745-755.
Gennarelli, T. A. (1985). The state of the art of head injury biomechanics - a review. Paper
presented at the 29'*’ Annual Conference, Association for the Advancement of
Automotive Medicine.
Gennarelli, T. A. (1987). Head Injury Biomechanics: A Review. Paper presented at the
Head Injury Mechanisms, New Orleans, LA.
Gennarelli, T. A., & Thibault, L. E. (1982). Biomechanics of acute subdural hematoma.
Journal of Trauma, 22(8), 680-686.
Germarelli, T. A., Thibault, L. E. & Ommaya, A.K. (1972). Comparison of translational
and rotational head motions in experimental cerebral concussion. Paper presented at 1 6 '* ’
Stapp Car Crash Conference, Society of Automotive Engineers, 296-308.
Grimard, G., Nolan, T., & Carlin, J. B. (1995). Head injuries in helmeted child bicyclists.
Injury Prevention, 1, 21-25.
Gurdjian, E. S., & Lissner, H. R. (1945). A study with the "Stresscoat" technique.
Surgery, Gynecology and Obstetrics, 81, 679-687.
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gurdjian, E. S., & Lissner, H. R. (1946). Deformation of the skull in head injury studied
by the "Stresscoat" technique, quantitative determinations. Surgery, Gynecology and
Obstetrics, 83, 219-233.
Gurdjian, E. S., Lissner, H. R., & Webster, J. E. (1947). The mechanism of production of
linear skull fractures. Surgery, Gynecology and Obstetrics, 85, 195-210.
Gurdjian, E. S., & Webster, J. E. (1958). Head Injuries - Mechanisms, Diagnosis and
Management. Boston: Little Brown.
Hennig, E. M., & Milani, T. L. (1995). In shoe pressure distribution for running in
various types of footwear./owttm/ of Applied Biomechanics, //(3), 299-310.
Hopes, P. D., & Chinn, B. P. (1989). Helmets: a new look at desigt and possible protection.
Paper presented at the International Research Council on the Biokinetics of Impact
(IRCOBI), Stockholm, Sweden.
Hurt, H. H., Jr., & Thom, D. R. (1985). Laboratory tests and accident performance of
bicycle safety helmets. Paper presented at the 29'*’ Annual Conference of the Association
for the Advancement of Automotive Medicine.
Hurt, H. H., Jr., & Thom, D. R. (1992). Motorcycle head injury mechanisms - with and
without helmets. Paper presented at the 3 6 '* * Annual Proceedings of The Association For
The Advancement of Automotive Medicine, Portland, Oregon.
Insurance Institute for Highway Safety. (1990). Bicycles. Fatality Facts.
Johnson, R. J. (1991). Intracranial injuries: anatomy of the brain and its coverings. In J. S.
Torg (Ed.), Athletic injuries to the head, neck and face (pp. 213-224). St. Louis, Missouri:
Mosby-Year Book, Inc.
Lane, J. C. (1986). Helmets For Child Bicyclists: Some Biomedical Considerations (CR 47).
Canberra: Federal Office of Road Safety, Department of Transport.
Levin, H. S., Mendelsohn, D., Lilly, M. A., Yeakley, J., Song, J., Scheibel, R. S.,
Harward, H., Fletcher, J. M., Kufera, J. A., Davidson, K. C., & Bruce, D. (1997).
Magnetic resonance imaging in relation to functional outcome of pediatric closed head
injury: A test of the Ommaya-Cennarelli model. Neurosurgery, 40(3), 432-440.
Lissner, H. R., Lebow, M., & Evans, F. C. (1960). Experimental studies on the relation
between acceleration and intracranial pressure changes in man. Surgery, Gynecology and
Obstetrics, 111, 329-338.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lombard, C. F., & Advani, S. H. (1966). Impact protection of the head and neck. Paper
presented at the Air Force Industry Two Wheel Motor Vehicle Safety Seminar,
California.
Long, G. J., Dowdell, B., & Griffiths, M. (1989). Development of a localised loading test for
pedal cycle helmets. . Rosebery, New South Wales: Crashlab, Road Safety Bureau, Roads
and Traffic Authority, New South Wales.
McIntosh, A., & Dowdell, B. (1992). A field and laboratory study of the performance of
pedal cycle helmets in real accidents. Paper presented at the International Research Council
on the Biokinetics of Impact (IRCOBI).
McIntosh, A. S., KaUieris, D., Mattem, R., Svensson, N. L., & Dowdell, B. (1995). An
evaluation of pedal cycle performance requirements. Paper presented at the 39'*' Stapp Car
Crash Conference, San Diego, CA.
McPoil, T. G., Cornwall, M. W., & Yamada, W. (1995). A comparison of two in-shoe
plantar pressure measurement systems. The Lower Extremity, 2(2), 95-103.
Mills, N. J., & Gilchrist, A. (1991). The effectiveness of foams in bicycle and motorcycle
helmets. Accident, Analysis and Prevention, 23, 153-163.
Moulton, J. R., Warner, C. Y., & Mellander, H. (1984). Design, development and testing of
a load sensing crash dummy face. Paper presented at the Advances in Belt Restraint
Systems: Design, Performance and Usage Conference, Detroit, Michigan.
Nahum, A. M., Gatts, J. D., Gadd, C. W., & Danforth, J. (1968). Impact tolerance of the
skull and face. Paper presented at the 12'*' Stapp Car Crash Conference, Detroit,
Michigan.
National Highway Traffic Safety Administration. (1981). Head and Neck Injury Criteria -
A Consensus Workshop, Washington, DC.
Nigg, B.M. (1990). The validity and relevance of tests used for the assessment of sports
surfaces. Medicine and Science in Sports and Exercise, 22(1), 131-139.
Newman, J. A. (1982). The influence of time duration as a failure criterion in helmet
evaluation. SAE Paper Number 821088, Society of Automotive Engineers, Warrendale,
PA.
Newman, J. A. (1993). Biomechanics of head trauma: head protection. In A. M. Nahum
& J. W. Melvin (Eds.), Accidental Injury: Biomechanics and Prevention . San Diego, CA:
Springer-Verlag.
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nusholtz, G. S., Lux, P., Kaiker, P., & Janicki, M. A. (1984, ). Head impact response -
skull deformation and angular accelerations. Paper presented at the 2 8'*’ Stapp Car Crash
Conference.
Ommaya, A. K. (1973). Head Injury Mechanisms. (226828). Washington, D C.: U.S.
Department of Transportation.
Ommaya, A. K., & Gennarelli, T. A. (1974). Cerebral concussion and traumatic
unconsciousness: correlation of experimental and clinical observations on blunt head
injuries. Brain, 97, 633-654.
Ommaya, A. K., & Yarnell, P. (1969). Subdural hematoma after whiplash. Lancet, 2, 237-
239.
Perl, T. R., Nilsson, S., Planath, I., & Wille, M. C. (1989). Deformable load sensing
Hybrid III face. Paper presented at the 33"* Annual Stapp Car Crash Conference,
Washington, D.C.
Prasad, P., Mertz, H.J., and Tarriere, C. (1985). Report to NHTSA by ISO Working Group
6 in regards to use of the HIC in the automotive environment. Paper presented at the Wayne
State University Head Injury Conference, Detroit, Michigan.
Rivara, F. P., Thompson, D. C., & Thompson, R. S. (1996). Circumstances and severity of
bicycle injuries. . Harborview Injury Prevention and Research Center: Snell Memorial
Foundation.
Rodgers, C.. (1993). Bicycle use and hazard patterns in the US, and options for injury
reduaion. Consumer Product Safety Commission, Washington, DC.
Sacks, J. J., Holmgreen, P., Smith, S. M., & Sosin, D. M. (1991). Bicycle associated head
injuries and deaths in the United States from 1984 to 1988 - how many are preventable?
Journal of the American Medical Association, 266, 3016-3018.
Smith, T. A. (1997). Evaluation of One Hundred Accident Involved Bicycle Helmets - Final
Report. Los Angeles, CA: University of Southern California, Head Protection Research
Laboratory.
Smith, T. A., Tees, D., Thom, D. R., & Hurt, H. H., Jr. (1994). Evaluation and
replication of impact damage to bicycle helmets. Accident, Analysis and Prevention.
Snell Memorial Foundation. (1995). 1995 Standard For Protective Headgear . St. James,
NY: Snell Memorial Foundation.
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Standards Australia/ Standards New Zealand Committee CS/14 on Safety Helmets for
Sport and Recreation and Subcommittee CS/14/2 on Pedal Bicycle Helmets. (1996).
AS/NZS 2512.9 Methods of testing protective helmets. Method 9; Determination of load
distribution. Standards Austraha and Standards New Zealand.
Thomas, L. M., Roberts, V. L., & Gurdjian, E. S. (1967). Impact induced pressure
gradients along three orthogonal axes in the human skull. Journal of Neurosurgery, 26,
316-321.
Thompson, R. S., Rivara, F. P., & Thompson, D. C. (1989). A case-control study of the
effectiveness of bicycle safety helmets. New England Journal of Medicine, 320, 1360-1367.
Ueno, K., Melvin, J. W., Li, L., & Lighthall, J. W. (1995). Development of tissue level
brain injury criteria by finite element inilysls. Journal of Neurotrauma, /2(4), 695-706.
United States Department of Transportation. (1981). Head and neck injury criteria: a
consensus workshop. Washington, D.C
UniForce Technical Notes (1996). Force Imaging Technologies, Chicago, Illinois.
Unterhamscheidt, F. J., & Sellier, K. (1966). Mechanics and pathomorphology of closed
brain injuries. In A. E. W. W.F. Caveness (Ed.), Head Injury . Philadelphia: J.B.
Lippincott.
Viano, D. (1988). Biomechanics of head injury - toward a theory linking head dynamic
motion, brain tissue deformation and neural trauma. Paper presented at the 3 2 " * ^ Stapp
Car Crash Conference,
Vulcan, A. P., Cameron, M. H., & Watson, W. L. (1992). Mandatory bicycle helmet use:
experience in Victoria, Australia. World Journal of Surgery, 16, 389-397.
Warner, C. Y., Wille, M. G., Brown, S. R., Nilsson, S., & Mellander, H. (1986). A load
sensing face form for automotive collision crash dummy instrumentation. Paper presented at
the Passenger Comfort, Convenience and Safety: Test Tools and Procedures., Detroit,
MI.
Wilcox, R. R. (1996). Statistics for the social sciences. San Diego, California: Academic
Press.
Wilcox, R. R. (1997). Introduction to robust estimation and hypothesis testing. San Diego,
California: Academic Press.
Williams, M. (1991). The protective performance of bicyclists' helmets in accidents.
Accident, Analysis and Prevention, 23, 119-131.
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wisman, J. A. (1996). Finite element modeling of a motorcycle helmet impact. Paper
presented at the AGAARD Conference on Head Injury, Alamagordo, NM.
Wood, T., & Milne, P. (1988). Head injuries to pedal cyclists and the promotion of
helmet use in Victoria, Australia, ylcrident. Analysis and Prevention, 20, 177-185.
Zhou, C., Khalil, T. B., & King, A. I. (1995). A new model comparing impact responses of
homogeneous and inhomogeneous human brain. Paper presented at the 39'*' Stapp Car
Crash Conference, San Diego, CA.
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix A
Headform Acceleration Data Sorted By Group
1 6 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0.5 m D rop Height
1.0 m D rop Height
1.5 m D rop Height
400
300
I
S 200 -
<
I 1 0 0 .
I
-100
Time (ms)
Figure Al: Group 1 Left Front Impact Location, Hemispherical Anvil
0
u
<
1
500
0 .5 m Drop Height
1.0 m D -o p Height
1.5 m Drop Height
2.0 m Drop Height
400
300
20 0
1 0 0
0
-100
Time (ms)
Figure A2: Group 2 Left Front Impact Location, Hemispherical Anvil
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
I
I
:
500
400
% 300
I
200 .
1 0 0
-100
0 5 m Drop Height
• _______ 1.0 m Drop Height
................1.5 m Drop Height
________ 2 .0 m Drop Height
10 20 30 40 5 0
Time (me)
Figure A3: Group 3 Left Front Impact Location, Hemispherical Anvil
500
4 0 0 .
0.5 m Drop Height
1.0 m Drop H eight
1.5 m Drop H e ig h t,
2.0 m Drop H eight
r 300
200
1 00 .
-100
Time (me)
Figure A4: Group 4 Left Front Impact Location, Hemispherical Anvil
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0.5 m Drop Height :
1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
400 .
f 300 .
1
e
s 200 .
<
I
1 0 0 .
I
I
-100
Time (ms)
Figure A5; Group 1 Left Front Impact Location, Flat Anvil
500
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
400 -
f 300
o
S
s 200
<
I
1 0 0
I
-100
Time (ms)
Figure A6: Group 2 Left Front Impact Location, Flat Anvil
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
400
~ 300
I
o
<
I
200
1 0 0
1 0
-100
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height i
2.0 m Drop Height
f L
2 0 50
Time (ms)
Figure A7: Group 3 Left Front Impact Location, Flat Anvil
500
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
4 00 -
f 300 -
I
8 200 -
<
I 1 0 0 .
I
-100
Time (ms)
Figure A8: Group 4 Left Front Impact Location, Flat Anvil
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
400
a
300 .
e
0
1
I
200 .
o
1 0 0
I
-100
Time (ms)
Figure A9: Group 1 Right Side Impact Location, Hemispherical Anvil
500
_ 0 .5 m Drop Height
. 1 . 0 m Drop Height
. . 1.5 m Drop Height
_ 2 .0 m Drop Height
400
300
I
£
s 200
<
I
1 00 .
-100
Time (ms)
Figure AlO: Group 2 Right Side Impact Location, Hemispherical Anvil
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0 .5 m Drop Height
1 .0 m Drop Height 1
1.5 m Drop Height;
2 .0 m Drop Height
400
f 300 .
I
*
S 200 .
<
£
■ o
I
1 0 0
40
-100
Time (ms)
Figure A ll: Group 3 Right Side Impact Location, Hemispherical Anvil
500
0 .5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2 .0 m Drop Height
400
f 300
I
»
S 200
<
I
I 100
I
-100
Time (ms)
Figure A 12: Group 4 Right Side Impact Location, Hemispherical Anvil
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0.5 m Drop Height
_ . 1.0 m Drop Height
1.5 m Drop Height
2.0 m Drop Height
400
f 300
0
1
£
S 20 0
<
I
1 0 0
-100
Time (ms)
Figure A13: Group 1 Right Side Impact Location, Flat Anvil
500
0.5 m Drop Height
1.0 m Drop Height
— 1.5 m Drop Height
2.0 m Drop Height
400
O )
% 300
I
£
8 200
<
I
X
1 0 0
-100
Time (ms)
Figure A 14: Group 2 Right Side Impact Location, Flat Anvil
175
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
0.5 m Drop Height
1.0 m Drop Height
1.5 m Drop Height
2 .0 m Drop Height
400
■f 300
I
s 200
<
I
1 0 0
£
■ o
I
-100
Time (ms)
Figure A15: Group 3 Right Side Impact Location, Flat Anvil
500
0 .5 m Drop Height
_ . 1.0 m Drop Height
1.5 m Drop Height
2 .0 m Drop Height
400
300
I
I
I
200 .
1 0 0
I
40
-100
Time (ms)
Figure A16: Group 4 Right Side Impact Location, Flat Anvil
176
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix B
Headform Acceleration Data Sorted By Drop Height
1 7 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
400
G roup 1
G roup 2
G roup 3
G roup 4
f 300
0
1
I 200
I
1 0 0
-100
Time (me)
Figure Bl: Left Front Impact Location, Hemispherical Anvil, 50 cm Drop Height
500
G roup 1
G roup 2
G roup 3
G roup 4
400
S
c 300
I
«
g 200
<
I
£
-100
Time (ms)
Figure B2: Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop Height
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5 0 0
Group 1
Group 2
Group 3
Group 4
400
o >
c
0
1
300
I 200 .
g
100
i
3
X
40 20
-100
Time (ms)
Figure B3: Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop Height
500
G roup 2
G roup 3
G roup 4
400
S
c 300
I
g 200
i
I
a 100
S
40
-100
Time (ms)
Figure B4: Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop Height
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
Group
Group
2
Group
Group
400
300
§
1
§
<
20 0
1 0 0
40
-100
Time (ms)
Figure B5: Left Front Impact Location, Flat Anvil, 50 cm Drop Height
500
Group 1
Group 2
Group 3
Group 4
400
3
c 300
0
1
3
8 200
I
% 1 0 0
40
-100
Time (ms)
Figure B6: Left Front Impact Location, Flat Anvil, 1.0 m Drop Height
1 8 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
400
Group 1
Group 2
Group 3
Group 4
S
e 300
0
1
f
I 200
I
1 0 0
40 20
-100
Time (ms)
Figure B7: Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
500
Group 1
Group 2
Group 3
Group 4
400
f 300
I
8 200
<
i
1 0 0
I
40 20
-100
Time (ms)
Figure B8: Left Front Impact Location, Flat Anvil, 2.0 m Drop Height
181
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
Group 1
Group 2
Group 3
Group 4
400
S
e 300
I
§ 200
<
i
I
100
4 0
-100
Time (ms)
Figure B9: Right Side Impact Location, Hemispherical Anvil, 50 cm Drop Height
500
Group 1
Group 2
Group 3
Group 4
400
3
c 300
I
§ 200
i
I
1 0 0
40
-100
Time (ms)
Figure BIO: Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop Height
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
Group 1,
Group 2i
Group 3 1
Group 4 1
400
O )
ê 300
1
I
8 200
I
1 0 0
I
40
-100
Time (ms)
Figure B ll: Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop Height
500
Group 2
Group 3
Group 4
400
f 300
0
1
I 200
I
1 0 0
40
-100
Time (ms)
Figure B12: Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop Height
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
G ro u p 1
G roup 2
G roup 3 i
G roup 4
4 00
a
300
§
E
"5
8 200
I
1 0 0
I
-100
Time (ms)
Figure B13: Right Side Impact Location, Flat Anvil, 50 cm Drop Height
500
G roup 1
G roup 2
G roup 3
G roup 4
400
S
c 300
I
8 200
<
p
1 0 0
-100
Time (ms)
Figure B14: Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
Group 1
Group 2
Group 3
Group 4
400
f 300
I
0 200
1
1 100
2
20
-100
Time (ms)
Figure B15: Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
500
400
Group 1
Group 2
Group 3
Group 4
S
a 300
Î
8 200
1
I
X
1 0 0
-100
Time (ms)
Figure B16: Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix C
Load Distribution Profiles At Time of Peak Sensor Force
Left Front Impact Location
1 8 6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C l: Group 1 Left Front Impact Location, Hemispherical Impact, 50 cm Drop
Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C2: Group 1 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-p30Q0
-hsoo
-pooo
Force (N )
100+
^1500
9 1 to 100
L I 000
82 to 9 1
Isoo
73 to 82
64 to 73
0
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9to18
1 0to9
Figure C3: Group 1 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
-8 -8
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C4: Group 2 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
188
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C5: Group 2 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
Figure C6: Group 2 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
1 8 9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C7: Group 2 Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C8: Group 3 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
y2S0
Force (N )
p o o ■ 100+
[iso
■ 91 to 100
■ 82 to 91
L i 00
1 73 to 82
■ 64 to 73
50 1 55 to 64
I 45 to 55
J
1 36 to 45
I 27 to 36
1 18 to 27
3 9 to 18
1 0to9
-8 -8
Figure C9: Group 3 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
■pso
p o o
Force (N )
■ 100+
p 5 0
■ 91 to 100
I 82 to 9 1
[lO O I 73 to 82
L 5 0
■ 64 to 73
I 55 to 64
0
■ 45 to 55
■ 36 to 45
1 27 to 36
1 1Bto27
S 9 to 18
0 Oto9
Figure CIO: Group 3 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
•8 -8
Figure C il: Group 3 Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure C12: Group 4 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C13: Group 4 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
Force (N)
100+
9 1 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
•8 -8
Figure C14: Group 4 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C15: Group 4 Left Front Impact Location, Hemispherical Impact, 2.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
•8 -8
Figure C16: Group 1 Left Front Impact Location, Fiat Anvil, 50 cm Drop Height
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-8 -8
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C17: Group 1 Left Front Impact Location, Fiat Anvil, 1.0 m Drop Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
•8 -8
Figure C18: Group 1 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
Oto9
Figure C19: Group 1 Left Front Impact Location, Fiat Anvil, 2.0 m Drop Height
Force (N )
100+
91 te 100
82 te 91
73 te 82
64 te 73
54 to 64
45 to 54
36 to 45
27 to 36
18 to 27
9 to 18
Oto9
-8 -8
Figure C20: Group 2 Left Front Impact Location, Flat Anvil, 50 cm Drop Height
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-8 -8
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C21: Group 2 Left Front Impact Location, Flat Anvil, 1.0 m Drop Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C22: Group 2 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C23: Group 2 Left Front Impact Location, Fiat Anvil, 2.0 m Drop Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C24: Group 3 Left Front Impact Location, Fiat Anvil, 50 cm Drop Height
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9to18
1 0to9
•8 -8
Figure C25: Group 3 Left Front Impact Location, Fiat Anvil, 1.0 m Drop Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C26: Group 3 Left Front Impact Location, Fiat Anvil, 1.5 m Drop Height
199
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C27: Group 3 Left Front Impact Location, Fiat Anvil, 2.0 m Drop Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
Figure C28: Group 4 Left Front Impact Location, Fiat Anvil, 50 cm Drop Height
200
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C29: Group 4 Left Front Impact Location, Fiat Anvil, 1.0 m Drop Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure C30: Group 4 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
-8 -8
Figure C31: Group 4 Left Front Impact Location, Fiat Anvil, 2.0 m Drop Height
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix D
Load Distribution Profiles At Time of Peak Sensor Force
Right Side Impact Location
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D l: Group 1 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
Figure D2: Group 1 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
•8 -8
Figure D3: Group 1 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
Figure D4: Group 2 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure D5: Group 2 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure D6: Group 2 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
206
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
-8 -8
Figure D7: Group 2 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
-8 -8
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D8: Group 3 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
1 0to9
Figure D9: Group 3 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure DIO: Group 3 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure Dll: Group 3 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D12: Group 4 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D13: Group 4 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
Force (N )
100+
9 1 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D14: Group 4 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
-8 -8
Figure DIS: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
Legend
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D16: Group 1 Right Side Impact Location, Fiat Anvil, 50 cm Impact
211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Legend
100+
91 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure DU: Group 1 Right Side Impact Location, Fiat Anvil, 1.0 m Drop Height
Legend
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure DIS: Group 1 Right Side Impact Location, Fiat Anvil, 1.5 m Drop Height
212
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D19: Group 1 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
Legend
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
a 9 to 18
1 0to9
Figure D20: Group 2 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Legend
64 to 73
Figure
D21: Group 2 R ig > > ‘ Side Impact
Location,
Flat Anvil, 1.0 m Drop Height
^ -8
L e g e n d
82 to
55 to 64
Figure D22: Group 2
Right Side Impaa Location, Flat
Anvil, 1.5 m Drop Height
214
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D23: Group 2 Righr Side Im paa Locarion. Flat Anvü. 2.0 m Drop Herght
rporce (N T
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D24: Group 3 Right Side Im paa Location, Flat Anvil, 50 cm Drop Herght
215
Force (N )
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D25: Group 3 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0 to 9
-8 -8
Figure D26: Group 3 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
216
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D27: Group 3 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
I 9 to 18
1 OtoS
Figure D28: Group 4 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
217
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D29: Group 4 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
Force (N )
100+
9 1 to 100
82 to 9 1
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
1 9 to 18
II 0 to 9
Figure D30: Group 4 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Force (N)
100+
91 to 100
82 to 91
73 to 82
64 to 73
55 to 64
45 to 55
36 to 45
27 to 36
18 to 27
9 to 18
0to9
Figure D31: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix E
Radial Distribution Analysis At Time of Peak Sensor Force
Left Front Impact Location
220
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
* Overall
0 1st Quadrant
^ 2nd Quadrant
3rd Quadrant
4th Quadrant
20
15
10
5
0 -----
Radius 0 Radius 1 Radius 2 Radius 5 Radius 6 Radius 3 Radius 4
Radius Location
Figure El: Group 1 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
250
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
200
150 -
z
1 0 0
s
£
Rad us 0 Radius 1 Radius 2 Radius 5 Rad us 6 Radius 3 Radius 4
-50
Radius Location
Figure E2: Group 1 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop Height
221
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3000
Overall
1st Quadrant
2rxJ Quadrant
3rd Quadrant
4th Quadrant
2500
2000
z
1500
£
1000
500
R adius 6 Radius 0 Radius 1 Radius 3 Radius 4 Radius 5 Radius 2
R a d iu s L o catio n
Figure E3: Group 1 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop Height
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
o 30
Radius 0 Radius 1 Radius 2 Radius 3
R a d iu s L o catio n
Radius 4 Radius 5 R adius 6
Figure E4; Group 2 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ovefail
1 st Quadrant
2nd Q uadrant
3rd Quadrant
4tti Quadrant
e 40
Radius 0 Radius 1 Radius 2 R adius 3
R a d iu s L o catio n
Radius 4 R adius 5 Radius 6
Figure E5: Group 2 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop Height
250
* Overali
0 1st Q uadrant
^ 2nd Q uadrant
X 3rd Q uadrant
4th Q uadrant
200
150 .
z
100 -
Radius 6 R adius 3 Radius 4 R adius 5 R adius 1 Radius 2 Radius 0
R a d iu s L o catio n
Figure E6: Group 2 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop Height
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
600
Overall
1st Q uadrant
2nd Q uadrant
500 -
3rd Q uadrant
4th Q uadrant
400 -
z
300 .
i
£
200
100
Radius 6 Radius 4 Radius 0 Radius 1 Radius 5 Radius 2 Radius 3
Radius Location
Figure E7: Group 2 Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop Height
140
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
120 .
100
z
8
£
60
4 0 -
Radius 6 Radius 0 Radius 1 Radius 4 Radius 5 Radius 2 R adius 3
Radius Location
Figure E8: Group 3 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
224
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
Overall
1st Q uadrant
2nd Quadrant
160
140 -
3rd Quadrant
4th Quadrant
120 -
z 100
i
80 .
|2
40 -
2 0 -
Radius 0 Radius 1 Radius 6 R adius 5 Radius 2 Radius 3 Radius 4
Radius Location
Figure E9: Group 3 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop Height
Radius 0
Overall
g 1 st Quadrant
^ 2 nd Quadrant
X _ 3 r d Quadrant
4th Quadrant
Radius 1 Radius 2 Radius 3
Radius Location
Radius 4 R adius 5 Radius 6
Figure ElO: Group 3 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Overall
1 st Quadrant
2nd Q uadrant
200 -
—X— 3rd Quadrant
4th Quadrant
150 .
z
g
Radius 6 Radius 0 Radius 1 Radius 2 Radius 3 Radius 4 R adius 5
Radius Location
Figure El 1: Group 3 Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
300
Overall
1st Quadrant
2nd Quadrant
250
_ X _ 3rd Q uadrant
4th Quadrant
200 -
z
£
100
Radius 6 R adius 0 Radius 1 Radius 2 Radius 3 Radius 4 Radius 5
Radius Location
Figure E12: Group 4 Left Front Impact Location, Hemispherical Anvil, 50 cm Drop
Height
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
350
* Overall
g 1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
300
250 -
Z 200
150 .
100 .
Radius 5 Radius 6 R adius 4 R adius 2 Radius 3 Radius 0 Radius 1
Radius Location
Figure E l3: Group 4 Left Front Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
250
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
u. ICO
Radius 0 Radius 1 Radius 2 Radius 3 R adius 4
Radius Location
Radius 5 Radius 6
Figure E l4: Group 4 Left Front Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
400
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
350 -
300 .
250
z
200
150
100 .
50 .
Radius 0 Radius 4 Radius 5 R adius 6 R adius 1 Radius 2 Radius 3
Radius Location
Figure E l5: Group 4 Left Front Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
Overail
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
2 0 .
1 0 .
Radius 0 Radius 1 Radius 2 Radius 3 Radius 4
Radius Location
Radius 5 R adius 6
Figure E l6: Group 1 Left Front Impact Location, Flat Anvil, 50 cm Drop Height
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
£
3 0
* Overall
1st Q uadrant
4 Q uadrant |
X 3rd Q uadrant .
4th Q uadrant
25
2 0
15
1 0
5
0 ----
Radius 0 Radius 6 Radius 2 Radius 4 Radius 5 Radius 1 Radius 3
R a d iu s L o catio n
Figure E l7: Group 1 Left Front Impact Location, Flat Anvil, 1.0 m Drop Height
£
30
Overall
1st Q uadrant
2nd Q uadrant
25
3rd Q uadrant
4th Q uadrant
2 0
15
1 0
5
0 ----
Radius 0 Radius 6 Radius 4 Radius 5 Radius 2 Radius 1 Radius 3
R a d iu s L o catio n
Figure El 8: Group 1 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Q uadrant
2nd Q uadrant
50 '
X 3rd Q uadrant
4th Q uadrant
z
30 -
2 0 -
Radius 0 Radius 1 Radius 2 Radius 4 Radius 6 Radius 3 Radius 5
R a d iu s L o catio n
Figure E l9: Group 1 Left Front Impact Location, Flat Anvil, 2.0 m Drop Height
Overall
1st Q uadrant
2nd Q uadrant
40 -
X 3rd Q uadrant
4th Q uadrant
z
Radius 0 Radius 1 Radius 2 Radius 4 R adius 6 Radius 3 Radius 5
R a d iu s L o catio n
Figure E20: Group 2 Left Front Impact Location, Flat Anvil, 50 cm Drop Height
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
2 - 40 .
8
i 30
2 0 .
Radius 5 Radius 6 Radius 0 R adius 1 Radius 4 Radius 2 Radius 3
R a d iu s L o ca tio n
Figure E21 : Group 2 Left Front Impact Location, Flat Anvil, 1.0 m Drop Height
Overall
1st Q uadrant
2nd Quadrant
3rd Q uadrant
4th Quadrant
o 30
Radius 0 R adius 1 Radius 2 Radius 3
R a d iu s L o catio n
Radius 4 Radius 5 R adius 6
Figure E22: Group 2 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
_ * Overall
* 1st Quadrant
^ 2nd Quadrant
X .3rd Quadrant
4th Q uadrant
100 -
z
60 .
I
£
40 .
Radius 6 Radius 4 Radius 5 Radius 0 Radius 2 Radius 3 Radius 1
R ad iu s L o catio n
Figure E23: Group 2 Left Front Impact Location, Flat Anvil, 2.0 m Drop Height
160
Overall
1st Quadrant
2nd Q uadrant
140
120
__X— 3rd Q uadrant
4th Quadrant
100
z
80
8
£
R adius 4 Radius 5 R adius 6 Radius 0 Radius 1 Radius 2 Radius 3
R ad iu s L o catio n
Figure E24: Group 3 Left Front Impact Location, Flat Anvil, 50 cm Drop Height
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200
* .Overali
—a — 1st Quadrant
4 2nd Quadrant
X 3rd Quadrant
4th Quadrant
180
160
140
1 2 0
z
100 .
80
2 0
Radius 6 Radius 4 Radius 5 Radius 0 R adius 3 Radius 1 Radius 2
R a d iu s L o c a tio n
Figure E25: Group 3 Left Front Impact Location, Fiat Anvil, 1.0 m Drop Height
Overall
1st Q uadrant
2nd Q uadrant
3nJ Q uadrant
4th Q uadrant
Radius 0 Radius 1 Radius 2 R adius 3
R a d iu s L o c a tio n
Radius 4 Radius 5 R adius 6
Figure E26: Group 3 Left Front Impact Location, Fiat Anvil, 1.5 m Drop Height
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
Radius 1 Radius 2 Radius 3 Radius 4 Radius 5 Rad u s 6
R a d iu s L o catio n
Figure E27: Group 3 Left Front Impact Location, Flat Anvil, 2.0 m Drop Height
180
Overall
0 1st Q uadrant
^ 2nd Q uadrant
_ X _ 3 r d Q uadrant
4th Q uadrant
160
140
120 -
Z 1 00 -
Ô 80
u.
Radius 1 R adius 6 Radius 0 Radius 2 Radius 5 Radius 3 Radius 4
R a d iu s L o catio n
Figure E28: Group 4 Left Front Impact Location, Flat Anvil, 50 cm Drop Height
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
Overall
1 st Q uadrant
2nd Q uadrant
140
120
3rd Q uadrant
4th Q uadrant 100
z
Radius 6 Radius 4 R adius 5 Radius 0 Radius 1 Radius 2 Radius 3
R a d iu s L o ca tio n
Figure E29: Group 4 Left Front Impact Location, Flat Anvil, 1.0 m Drop Height
160
- Overall
g 1st Quadrant
^ 2nd Q uadrant
X 3rri Q uadrant
4th Q uadrant
140
120 -
100
z
i
£
2 0 .
Rad Rad
-20
Radius 3 Radius 4 R adius 5 us 6 u s 0 Radius 1 R adius 2
R a d iu s L o ca tio n
Figure E30: Group 4 Left Front Impact Location, Flat Anvil, 1.5 m Drop Height
2 3 5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
200
Ovefall
1st Quadrant
2nd Quadrant
180
160
3rd Q uadrant
4th Quadrant
140
120
5. 100
i
£
Radius 3 R adius 5 Radi Radius 4 Radius 1 R adius 2 I us 0
R a d iu s L ocation
Figure E31 : Group 4 Left Front Impact Location, Flat Anvil, 2.0 m Drop Height
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix F
Radial Distribution Analysis At Time of Peak Sensor Force
Right Side Impact Location
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Radius 0 Radius 1 Radius 2 Radius 3
R a d iu s L o catio n
R adius 4
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
Radius 5 Radius 6
Figure Fl: Group 1 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
1200
Overall
1st Quadrant
2nd Quadrant
1000
800 - 3rd Quadrant
4th Quadrant
£• 600
8
£
400 .
200 .
Radius 5 Rad us 6 R adius 4 R adius 1 Radius 3 Rad us 0 Radius 2
-200
R a d iu s L o cation
Figure F2: Group 1 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4500
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
4000
3500
3000
_ 2500
% 2000
<2 1500
1000
500
R adius 3 Radius 4 R ad us 6 us 0 Radius 1 Radius 2 R adius 5 R ad
-500
R a d iu s L o catio n
Figure F3: Group 1 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
60 -
50 -
z
40 -
8
£
30 -
Radius 3 Radius 4 R adius 5 Radius 6 Radius 1 Radius 2 R adius 0
R a d iu s L o catio n
Figure F4: Group 2 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Quadrant
2nd Quadrant
80 -
70 -
3rd Quadrant
4th Quadrant
60 .
z
I
o
u.
2 0 .
Radius 6 Radius 4 Radius 5 R adius 1 Radius 2 Radius 3 Radius 0
R a d iu s L o ca tio n
Figure F5: Group 2 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
200
Overall
1 st Quadrant
2nd Quadrant
180 -
160
140 3rd Quadrant
4 th Quadrant
120
z
100
I
£
Radius 6 R adius 5 Radius 4 Radius 3 R adius 1 Radius 2 Radius 0
R a d iu s L o catio n
Figure F6: Group 2 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1200
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
1000
800 -
£ • 600 .
i
o 400
200 -
Radius 5 Rad us 6 Radius 4 R adius 3 Radius 1 Radius 2 Rad u s 0
-200
R a d iu s L o catio n
Figure F7: Group 2 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
1 2 0
Overall
1st Q uadrant
2nd Q uadrant
100 -
3rd Q uadrant
4th Q uadrant
z
60 .
I
u.
40 -
Radius 6 Radius 5 Radius 4 Radius 1 Radius 2 R adius 3 Radius 0
R a d iu s L o catio n
Figure F8: Group 3 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 6 0
Overall
1st Q uadrant
2nd Q uadrant
140 .
1 2 0 .
3rd Q uadrant
4th Q uadrant ,
1 0 0 .
z
Radius 2 Radius 3 Radius 4 Radius 6 Radius 0 Radius 1 Radius 5
R ad iu s L o catio n
Figure F9: Group 3 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
250
* Overall
0 1st Q uadrant
2nd Q uadrant
X 3rd Q uadrant
4th Q uadrant
200
150
z
I
R adius 2 R adius 6 Radius 3 Radius 4 Radius 5 Radius 0 Radius 1
R ad iu s L o catio n
Figure FIG: Group 3 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Ovefall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
200
150
z
100
Radius 4 R adius 6 Radius 2 Radius 5 R adius 0 Radius 1 Radius 3
R a d iu s L o ca tio n
Figure F ll: Group 3 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
180
Overall
g 1st Quadrant
4 2nri Quadrant
X 3nr Quadrant
4th Quadrant
160
140
1 2 0
z 100
i
o
u.
60
40
R adius 6 Radius 2 Radius 4 R adius 0 Radius 1 Radius 3 Radius 5
R a d iu s L o c a tio n
Figure F12: Group 4 Right Side Impact Location, Hemispherical Anvil, 50 cm Drop
Height
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250
Overall
1st Quadrant
2nd Quadrant
200 -
3rd Quadrant
4th Quadrant
150
z
Radius 2 Radius 4 Radius 6 R adius 1 Radius 5 Radius 0 Radius 3
R a d iu s L o catio n
Figure F13: Group 4 Right Side Impact Location, Hemispherical Anvil, 1.0 m Drop
Height
250
* Overall
_ B _ 1 s t Quadrant
_ ^ _ 2 n d Quadrant
X 3rd Quadrant
4th Quadrant
200
150
z
e
Radius 2 Radius 4 Radius 6 Radius 0 R adius 1 Radius 5 Radius 3
R a d iu s L o catio n
Figure F14: Group 4 Right Side Impact Location, Hemispherical Anvil, 1.5 m Drop
Height
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
300
Overall
— e — 1st Quadrant
— ^ 2nd Quadrant
_ X _ . 3rd Quadrant
4th Quadrant
250
200
z
150
100
R adius 6 Radius 4 Radius 0 Radius 1 Radius 2 Radius 3 Radius 5
R a d iu s L o catio n
Figure F15: Group 4 Right Side Impact Location, Hemispherical Anvil, 2.0 m Drop
Height
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
25
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
2 0
15
1 0
5
0
R adius 0 R adius 1 R adius 2 Radius 3
R a d iu s L o catio n
Radius 4 Radius 5 Radius 6
Figure F16: Group 1 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
9 10
Radius 0 Radius 1 Radius 2 Radius 3
R a d iu s L o catio n
Radius 4 Radius 5 Radius 6
Figure F17: Group 1 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
^ Overall
* 1st Q uadrant
^ -2 n d Q uadrant
^ 3rd Q uadrant
4th Q uadrant
40
z
i
£
R adius 6 R adius 5 Radius 0 R adius 3 Radius 4 Radius 1 R adius 2
R a d iu s L ocation
Figure F18: Group 1 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
140
Overall
1st Q uadrant
2nd Q uadrant
120 -
100
3rd Q uadrant
4th Q uadrant
£ • 80 -
8
£
60
Radius 6 Radius 5 Radius 3 Radius 4 Radius 0 R adius 2 Radius 1
R a d iu s L ocation
Figure F19: Group 1 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
247
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
40 -
35
30 -
Z 25
8
Ô 20
Li.
Radius 5 Radius 6 Radius 4 Radius 1 Radius 2 Radius 3 Radius 0
R a d iu s L ocation
Figure F20: Group 2 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
Overall
1st Quadrant
2nd Quadrant
35
X 3rd Quadrant
4th Quadrant
z 25 -
Radius 6 Radius 5 Radius 2 Radius 3 Radius 4 R adius 0 Radius 1
R a d iu s L ocation
Figure F21: Group 2 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
248
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall
1s t Quadrant
2nd Quadrant
3rd Q uadrant
4th Q uadrant
9 30
Radius 0 Radius 1 R adius 2 Radius 3
R a d iu s L ocation
Radius 4 R adius 5 Radius 6
Figure F22; Group 2 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
S
Radius 0 Radius 1 R adius 2
.Overall
.1 s t Quadrant
.2 n d Quadrant
. 3rd Quadrant
.4 th Quadrant
Radius 3
R a d iu s L ocation
Radius 4 Radius 5 Radius 6
Figure F23: Group 2 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
Overall
a 1st Quadrant
^ 2nd Quadrant
3rd Quadrant
4 th Q uadrant
50 .
4 0 .
z
S
S.
1 0 .
R adius 5 Radius 6 Radius 0 Radius 1 Radius 2 Radius 3 R adius 4
R ad iu s L o catio n
Figure F24: Group 3 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
Overall
1st Quadrant
2nd Quadrant
80
X 3rd Quadrant
4th Quadrant
60
Z 50
30 -
2 0 .
1 0 .
R adius 5 Radius 6 Radius 0 Radius 2 Radius 3 R adius 4 Radius 1
R ad iu s L o catio n
Figure F25: Group 3 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
2 5 0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
120
Overall
1st Quadrant
2nd Quadrant
100
3rd Quadrant
4th Quadrant
80 .
z
S
£
40
Radius 5 R adius 6 Radius 1 Radius 2 Radius 3 R adius 4 Radius 0
R ad iu s L o catio n
Figure F26: Group 3 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
120
Overall
1st Quadrant
2nd Quadrant
3rd Quadrant
4th Quadrant
Radius 0 Radius 1 Radius 2 Radius 3
R ad iu s L ocation
R adius 4 Radius 5 R adius 6
Figure F27: Group 3 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
2 5 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
* Ovefall
a 1st Quadrant
_ ^ _ 2 n d Quadrant
_ ^ ( _ 3 r d Quadrant
4tti Quadrant
30
20
Radius 0 R adius 1 Radius 2 Radius 3
R a d iu s
R adius 4 Radius 5 R adius 6
Figure F28: Group 4 Right Side Impact Location, Flat Anvil, 50 cm Drop Height
R a d iu s L o catio n
120
* Overall
__0 _ 1s t Quadrant
_ ^ _ 2 n d Quadrant
X 3rd Quadrant
4tti Quadrant
100
z
i
£
20 -
R adius 4 Radius 5 Radius 6 Radius 0 R adius 1 Radius 2 Radius 3
Figure F29: Group 4 Right Side Impact Location, Flat Anvil, 1.0 m Drop Height
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
180
Overall
1st Q uadrant
2nd Q uadrant
3rd Q uadrant
4th Q uadrant
160 .
140 .
120
z 100
80 .
60 .
40 .
20 .
R adius 6 Radius 0 Radius 1 Radius 2 Radius 4 Radius 5 Radius 3
R a d iu s L o catio n
Figure F30: Group 4 Right Side Impact Location, Flat Anvil, 1.5 m Drop Height
120
Overall
1st Q uadrant
2nd Q uadrant
100 -
3rd Q uadrant
4th Q uadrant
z
a
2
£
40 -
R adius 6 Radius 0 Radius 1 Radius 2 Radius 4 Radius 5 Radius 3
R a d iu s L o catio n
Figure F31: Group 4 Right Side Impact Location, Flat Anvil, 2.0 m Drop Height
253
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TEST TARGET (Q A -3 )
✓
/
1 . 0
l.l
1.25
I g i a W à
12.2
JX2
13 6
: m m
L. ^
1 . 8
1.4
1 .6
15 0 m m
V
V
/ A P P L I E D A IIV 14G E . I n c
1653 East Main Street
Rochester, NY 14609 USA
Phone: 716/482-0300
- = ~ ^ ^ = Fax: 716/288-5989
O 1993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly firom the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
fi’ om any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing fi'om left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Beil & Howell Inform ation Company
300 North Zed) Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mil Number: 9835137
Copyright 1997 by
Smith, Terrance Alan
All rights reserved.
UMI Microform 9835137
Copyright 1998, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The effects of impact on bone mineral density over the course of a sports season

PDF

The role of the vasti in patellar kinematics and patellofemoral pain

PDF

The project life cycle and budgeting functions: planning, control, and motivation

PDF

The effects of varying levels of physical activity on psychological well-being in adults 60 and over

PDF

Cathodoluminescence studies of the influence of strain relaxation on the optical properties of InGaAs/GaAs quantum heterostructures

PDF

Alternate models of women's health care policy in the United States

PDF

Biomechanical factors that contribute to verticle jump performance

PDF

The isolation and characterization of rat plasma albumins

PDF

The role of the sensorimotor cortical system in skill acquisition and motor learning: a behavioral study

PDF

Elevated inflammation in late life: predictors and outcomes

PDF

Semantic heterogeneity resolution in federated databases by meta-data implantation and stepwise evolution

PDF

The impact of paperless systems and other technological changes upon auditing

PDF

The economics of tenure: understanding the effects of ethnicity, status and discipline on faculty attitude, workload and productivity

PDF

Novel phenylpolyene-bridged second-order nonlinear optical chromophores and new thermally stable polyurethanes for electro-optic applications

PDF

Protocol evaluation in the context of dynamic topologies

PDF

Internet security and quality-of-service provision via machine-learning theory

PDF

¹⁹F-NMR studies of trifluoroacetyl insulin derivatives

PDF

Age-related blunting of growth hormone secretion during exercise is not due solely to increased somatostatin tone

PDF

Experimental and computational analysis of microsatellite repeat instability in mismatch-repair-deficient mice

PDF

Colorectal cancer: genomic variations in insulin-like growth factor-1 and -2

Asset Metadata

Creator Smith, Terrance Alan (author)

Core Title The effect of helmet liner density upon acceleration and local contact forces during bicycle helmet impacts

School Graduate School

Degree Doctor of Philosophy

Degree Program Exercise Science

Degree Conferral Date 1997-12

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag engineering, biomedical,engineering, materials science,health sciences, occupational health and safety,health sciences, public health,health sciences, recreation,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-327769

Unique identifier UC11350793

Identifier 9835137.pdf (filename),usctheses-c17-327769 (legacy record id)

Legacy Identifier 9835137.pdf

Dmrecord 327769

Document Type Dissertation

Rights Smith, Terrance Alan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA