Case1291318455 PDF

BIOMECHANICAL ENGINEERING ANALYSES OF HEAD AND SPINE IMPACT
INJURY RISK VIA EXPERIMENTATION AND COMPUTATIONAL SIMULATION
by
ADAM JESSE BARTSCH
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy
Department of Mechanical and Aerospace Engineering
CASE WESTERN RESERVE UNIVERSITY
January 2011
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of
ADAM JESSE BARTSCH

____________________________________________________
DOCTOR OF PHILOSOPHY
candidate for the ______________________________degree *.
VIKAS PRAKASH, PHD

(signed) ______________________________________________
(chair of the committee)
LARS GILBERTSON, PHD

________________________________________________
EDWARD BENZEL, MD
________________________________________________
CLARE RIMNAC, PHD

________________________________________________
JOSEPH MANSOUR, PHD

________________________________________________
________________________________________________
3 DECEMBER 2010
(date) _______________________
*We also certify that written approval has been obtained for any proprietary
material contained therein.
For Kendra and Elijah, Mom and Pop, and Ben, Noah and Jason -- this effort
was only possible through your years of love, patience and endless
encouragement
TABLE OF CONTENTS
TABLE OF CONTENTS ............................................................................I

LIST OF TABLES ................................................................................ III
LIST OF FIGURES ...............................................................................IV
ACKNOWLEDGEMENTS ..................................................................... VIII
ABSTRACT .......................................................................................... X
CHAPTER 1 – REAL WORLD MINOR REAR CRASHES .................................. 1
INTRODUCTION................................................................................ 1
BACKGROUND.................................................................................. 1
METHODS........................................................................................ 4
RESULTS ......................................................................................... 9
DISCUSSION OF RESULTS ............................................................... 18
LIMITATIONS ................................................................................. 23
CONCLUSIONS ............................................................................... 24
CHAPTER 2 – MINOR REAR CRASH MADYMO SIMULATION ....................... 26
INTRODUCTION.............................................................................. 26
BACKGROUND................................................................................ 26
METHODS AND MATERIALS .............................................................. 27
RESULTS ....................................................................................... 31
DISCUSSION.................................................................................. 40
LIMITATIONS ................................................................................. 42
CONCLUSIONS ............................................................................... 43
CHAPTER 3 – NECK INJURY RISK CLASSIFICATION BASED ON ORIGINAL Nij
DATASETS ........................................................................................ 45
INTRODUCTION.............................................................................. 45
BACKGROUND................................................................................ 45
NIJ DEVELOPMENT........................................................................... 47
STUDY METHODS ........................................................................... 48
STATISTICAL AND VISUAL ANALYSIS METHODS ................................. 52
RESULTS ....................................................................................... 56
DISCUSSION.................................................................................. 59
LIMITATIONS ................................................................................. 60
CONCLUSIONS ............................................................................... 63
CHAPTER 4 – ATD IMPACTS AND HEAD-NECK DYNAMICS ........................ 65
INTRODUCTION.............................................................................. 65
BACKGROUND................................................................................ 65
METHODS...................................................................................... 66
RESULTS ....................................................................................... 68
DISCUSSION.................................................................................. 78
LIMITATIONS ................................................................................. 82
CONCLUSIONS ............................................................................... 83
CHAPTER 5 – ATHLETIC PADDING INFLUENCE ON ATD HEAD-NECK IMPACT
DYNAMICS........................................................................................ 86
INTRODUCTION.............................................................................. 86
BACKGROUND................................................................................ 86
LITERATURE SURVEY ...................................................................... 87
I
METHODS...................................................................................... 88
RESULTS ....................................................................................... 96
DISCUSSION.................................................................................107
LIMITATIONS ................................................................................111
CONCLUSIONS ..............................................................................112
CHAPTER 6 – HELMET TESTING IMPACT SIMULATIONS ..........................115
INTRODUCTION.............................................................................115
BACKGROUND...............................................................................116
METHODS.....................................................................................121
RESULTS ......................................................................................129
DISCUSSION OF RESULTS ..............................................................135
LIMITATIONS ................................................................................139
CONCLUSIONS ..............................................................................139
CHAPTER 7 – IMPACT PENDULUM DESIGN, PIG SPINE TESTING, FOOTBALL
HELMET IMPACTS .............................................................................141
INTRODUCTION.............................................................................141
IMPACT PENDULUM........................................................................142
CADAVERIC IMPACTS .....................................................................153
HELMETED ATD IMPACTS................................................................167
LIMITATIONS ................................................................................177
CONCLUSIONS ..............................................................................178
SUMMARY........................................................................................181
APPENDIX........................................................................................186
BIBLIOGRAPHY.................................................................................191
II
LIST OF TABLES
Table 1-1. The 113 struck vehicle occupants with crash related diagnoses .. 9
Table 1-2. Reconstruction struck vehicle delta-V and mean acceleration
(n=105 crashes)................................................................................ 10
Table 2-1. Passenger vehicle crash tests used as inputs to MADYMO minor
rear crash simulation.......................................................................... 28
Table 2-2. Comparison of additional validation parameters ..................... 34
Table 2-3. Neck injury risk values as a percentage of threshold ............... 35
Table 2-4. Head injury risk values as percentage of threshold ................. 35
Table 2-5. BioRIDII output for pulse scaled by 40% and change from original
response .......................................................................................... 39
Table 3-1. Abbreviated Injury Scale (AIS) ............................................ 52
Table 3-2. Classification comparison for Nij (2000), Nusholtz et al. (2003)
and current study .............................................................................. 57
Table 3-3. 40% injury risk intercepts for the four parameters based on
regression curves .............................................................................. 58
Table 4-1. Test impact matrix............................................................. 68
Table 4-2. Coefficient of variation among all impact trial sets (Front =
Frontal, Obl = Oblique, Lat = Lateral) ................................................... 73
Table 5-1. Test impact matrix............................................................. 91
Table 5-2. Theoretical post-impact head velocity ................................... 94
Table 5-3. Coefficient of variation of each resultant impact parameter
corresponding to time of highest cg linear acceleration resultant .............. 99
Table 5-4. Low energy impact maxima (shaded = significant increase
p<0.05, italics = significant decrease p<0.05) ......................................104
Table 5-5. High energy impact maxima (shaded = significant increase
p<0.05, italics = significant decrease p<0.05) ......................................107
Table 6-1. Test impact matrix............................................................123
Table 6-2. Simulink simulation parameters for constant kinetic energy
comparison ......................................................................................128
Table 6-3. Simulink simulation parameters for constant momentum
comparison ......................................................................................129
Table 6-4. Additional validation parameter comparison.........................133
Table 7-1. Cadaver impact matrix ......................................................156
Table 7-2. Impact matrix for football helmet power analysis (*to avoid
facemask contact, frontal and oblique impacts used a 5° forward inclination)
......................................................................................................170
III
LIST OF FIGURES
Figure 1-1. Struck vehicle delta-V ....................................................... 10

Figure 1-2. Struck vehicle acceleration................................................. 10
Figure 1-3. Medical treatment facility for first visit post-crash (n=113 struck
vehicle occupants) ............................................................................. 12
Figure 1-4. Days until first treatment visit ............................................ 12
Figure 1-5. Clinician visited first (n=85 occupants) ................................ 13
Figure 1-6. ICD-9-CM diagnoses ......................................................... 13
Figure 1-7. AIS coding....................................................................... 13
Figure 1-8. Degenerative diagnoses..................................................... 15
Figure 1-9. AIS1 injuries.................................................................... 15
Figure 1-10. AIS1 diagnoses comparing ‘whiplash’ groups during five week
treatment period post-crash ................................................................ 16
Figure 1-11. ‘Whiplash’ in degenerative cervical (Group A) and non-cervical
degenerative occupants (Group B) ....................................................... 17
Figure 1-12. AIS1 lumbar/sacral injury rate in degenerative (Group C) and
non-lumbar/sacral degenerative occupants (Group D)............................. 18
Figure 2-1. MADYMO longitudinal vehicle crash pulse and delta-V from minor
rear crash tests ................................................................................. 28
Figure 2-2. Initial position of BioRIDII facet ATD ................................... 29
Figure 2-3. Pelvis x-axis acceleration ................................................... 32
Figure 2-4. Head x-axis acceleration.................................................... 32
Figure 2-5. Thorax x-axis acceleration ................................................. 33
Figure 2-6. Head y-axis angular acceleration ........................................ 33
Figure 2-7. Pelvis x-axis acceleration ................................................... 37
Figure 2-8. Head x-axis acceleration.................................................... 37
Figure 2-9. Thorax x-axis acceleration ................................................. 38
Figure 2-10. (a) HybridIII and (b) BioRIDII [Denton ATD] ..................... 38
Figure 3-1. Nij v. time for ‘no/low injury’ (A) and ‘injured’ (B) piglet-ATD
tests ................................................................................................ 48
Figure 3-2. ATD axial force (Fz) for paired tests with ‘no/low injury’ (A) and
‘injured’ (B) piglets ............................................................................ 49
Figure 3-3. Binning method applied to normalized force data from Piglet Test
#6368 ............................................................................................. 51
Figure 3-4. Logistic regression curves for four parameters: (a) Maximum
Normalized Load, (b) Maximum Loading Rate, (c) Cumulative Impulse, (d)
%Time Above Bin=0.9. Intercepts drawn at 40% injury risk ................... 54
Figure 3-5. Three parameter bubble scatterplot for six (6) injury parameter
combinations: (a) Maximum Loading Rate vs. Maximum Normalized Force,
(b) Cumulative Impulse vs. Maximum Normalized Force, (c) Cumulative
Impulse vs. Loading Rate, (d) Maximum Loading Rate ............................ 55
Figure 3-6. Four parameter spheroidal scatterplot of four (4) injury
parameter combinations: (a) Maximum Normalized Force vs. %Bin Time vs.
Cumulative Impulse, (b) Maximum Normalized Force vs. %Bin Time vs.
Maximum Loading Rate, (c) Maximum Normalized Force vs. Cumulative
Impulse vs. Maximum Loading Rate, (d) %Bin Time vs. Cumulative Impulse
IV
vs. Maximum Force Loading Rate. Volume intercepts drawn at 40% injury
risk, sphere radius equals AIS injury severity......................................... 56
Figure 4-1. Frontal, Oblique and Lateral impact setups ........................... 68
Figure 4-2. Resultant ATD head center of gravity (cg) linear acceleration for
each set of six (6) impact trials in the three impact directions at highest
impact energy ................................................................................... 69
Figure 4-3. Resultant ATD occipital force for each set of six (6) impact trials
in the three impact directions at highest impact energy........................... 70
Figure 4-4. Mean ATD head center of gravity (cg) linear acceleration
components from each set of six (6) impact trials in the three impact
directions at highest impact energy ...................................................... 71
Figure 4-5. Mean ATD occipital force components from each set of six (6)
impact trials in the three impact directions at highest impact energy......... 72
Figure 4-6. Resultant ATD head center of gravity (cg) linear acceleration for
the three impact directions compared with impact energy [NOTE: error bars
show +/- one standard deviation] ........................................................ 74
Figure 4-7. Resultant ATD head angular velocity ................................... 75
Figure 4-8. Resultant ATD head angular acceleration ............................. 76
Figure 4-9. Resultant ATD occipital bending moment ............................. 77
Figure 4-10. Resultant ATD occipital force ............................................ 78
Figure 5-1. Unpadded Impact Setup .................................................... 91
Figure 5-2. Bare Pendulum-Headgear .................................................. 92
Figure 5-3. MMA-Bare Head ............................................................... 92
Figure 5-4. Boxing-Bare Head............................................................. 92
Figure 5-5. Boxing-Headgear Impact Setup (High Energy Only)............... 93
Figure 5-6. Simplified elastic collision model ......................................... 94
Figure 5-7. Head cg resultant linear acceleration in high energy trials for the
four padding conditions tested ............................................................. 97
Figure 5-8. Occipital neck force resultant in high energy trials for the four
padding conditions tested ................................................................... 98
Figure 5-9. Head cg resultant linear acceleration in high and low energy
trials ...............................................................................................100
Figure 5-10. Head resultant angular velocity in high and low energy trials100
Figure 5-11. Head resultant angular acceleration in high and low energy
trials ...............................................................................................101
Figure 5-12. Resultant impact force in high and low energy trials............101
Figure 5-13. Sample of computational simulation results displaying Mises
stresses within brain from Box-Bare impacts using SIMon software..........102
Figure 6-1. Pendulum Impact setups for (A) Unhelmeted, (B) ‘Leatherhead’,
(C)-(E) NOCSAE Approved Helmets .....................................................123
Figure 6-2. Two DOF system dynamics model for SEA, Ltd. lateral pendulum
helmet impacts.................................................................................124
Figure 6-3. Unhelmeted lateral NOCSAE calibration drop test system
dynamics model ...............................................................................125
Figure 6-4. Helmeted NOCSAE lateral impact test system dynamics model126
Figure 6-5. Proposed NOCSAE linear impact with highly rigid neck
attachment ......................................................................................127
V
Figure 6-6. Linear impact used in prior NFL studies with flexible Hybrid III
neck ...............................................................................................127
Figure 6-7. Hybrid III head cg linear acceleration for the single (n=1)
pendulum head impacts.....................................................................130
Figure 6-8. Hybrid III head angular velocity for n=1 pendulum head impacts
......................................................................................................130
Figure 6-9. Hybrid III head angular acceleration for n=1 pendulum head
impacts ...........................................................................................131
Figure 6-10. Hybrid III head upper neck shear force for n=1 pendulum head
impacts ...........................................................................................131
Figure 6-11. Hybrid III head upper neck bending moment for n=1 pendulum
head impacts ...................................................................................132
Figure 6-12. Simulink simulation results compared with average response
from Hybrid III head cg linear acceleration for the three (3) helmeted
pendulum impacts ............................................................................133
Figure 6-13. Constant energy Simulink results for headform cg linear
acceleration .....................................................................................134
Figure 6-14. Constant momentum Simulink results for headform cg linear
acceleration .....................................................................................135
Figure 7-1. Examples of impact pendulums -- BioKinetics (Ottawa, Canada),
NOCSAE (Knoxville, TN), Johns Hopkins Applied Physics Lab (Baltimore, MD),
Disney Epcot Center (Orlando, FL), Texas Transportation Institute (College
Station, TX), Impact Jackets LLC (Silver Spring, MD) .............................143
Figure 7-2. SolidEdge Rendering and first version.................................144
Figure 7-3. Retrofit rendering and assembled final pendulum .................145
Figure 7-4. Cadaveric and ATD impact setups ......................................146
Figure 7-5. Electromagnets used in cadaver [500N] and ATD [1500N]
impacts ...........................................................................................147
Figure 7-6. Frame vibration analysis hammer strike locations.................148
Figure 7-7. FFT results for the four (4) frame hammer strikes. NOTE: Chart
areas enlarged from original scale of Frequency = 5000Hz, lY(f)l = 0.8 to
show locations of interest ..................................................................149
Figure 7-8. Fixture mass quantification for fastening directly to load cell and
fastening to fixture components sitting atop load cell .............................150
Figure 7-9. Fixture mass effects - headform resultant linear acceleration .151
Figure 7-10. Fixture mass effects - headform resultant angular velocity...152
Figure 7-11. Fixture mass effects - load cell resultant bending moment ...152
Figure 7-12. Cadaveric specimen preparation: Removing mandible,
Removing caudal structures and sawing through skull, Testing orientation,
Potting with cerroband liquid metal – sawbones replica ..........................155
Figure 7-13. Instrumentation and testing sequence: Affixing sensors to
mounting blocks, Initial cadaver position at 40-degrees forward tilt, National
Instruments DAQ system and Occipito-cervical disruption caused in specimen
001 after sixth impact .......................................................................157
Figure 7-14. Resultant linear acceleration results from three cadaver
specimens for the fifth impact (KE = 25.9J)..........................................158
Figure 7-15. Sagittal plane angular velocity results from three cadaver
VI
Figure 7-16. Sagittal plane angular acceleration results from three cadaver
Figure 7-17. Resultant force results from three cadaver specimens across
four impacts (KE = 4.7J, 10.3J, 14.5J, 25.9J) .......................................161
Figure 7-18. Resultant moment results from three cadaver specimens across
four impacts (KE = 4.7J, 10.3J, 14.5J, 25.9J) .......................................162
Figure 7-19. Resultant linear acceleration power analysis results from three
cadaver specimens and four instrumented spinal levels ..........................163
Figure 7-20. Angular velocity power analysis results from three cadaver
specimens and four instrumented spinal levels......................................164
Figure 7-21. Angular acceleration power analysis results from three cadaver
Figure 7-22. Force and moment power analysis results from three cadaver
Figure 7-23. Football helmet power analysis impact setup alignment, lateral
impact, oblique impact, frontal impact, helmet positioning, overall view ...169
Figure 7-24. ATD-measured linear acceleration, angular velocity, angular
acceleration and bending moment from frontal pendulum impacts at 84J
(5.0m/s) .........................................................................................171
acceleration and bending moment from oblique pendulum impacts at 84J
(5.0m/s) .........................................................................................172
acceleration and bending moment from lateral pendulum impacts at 84J
(5.0m/s) .........................................................................................173
Figure 7-27. Lateral football helmet impact power analysis (73/75 conditions
had P = 80% or higher).....................................................................174
Figure 7-28. Oblique football helmet impact power analysis (71/75
conditions had P = 80% or higher)......................................................174
Figure 7-29. Frontal football helmet impact power analysis (67/75 conditions
had P = 80% or higher).....................................................................175
Figure 7-30. NFL in-game reconstructed football helmet impact power
analysis (13/30 conditions had P = 80% or higher) ...............................176
VII
ACKNOWLEDGEMENTS
The assistance of Kay Kress, Sue Legg and Adam Ratliff in the gathering of
crash cases used in Chapter 1 is gratefully acknowledged. In Chapter 2, the
contribution of the crash test data from Gunter Siegmund and Bradley
Heinrichs of MEA Forensic Engineers and Scientists made the MADYMO model
validation possible. Additionally, Doug Morr, Brian Tanner, Elaine Peterman
and Chris Iams of SEA Limited deserve thanks for their roles in facilitating
use of SEA’s MADYMO platform as well as troubleshooting. A special thanks
to Sally Duquensel and Mutaz Shoukani of TASS-SAFE for their assistance
during MADYMO model development and for providing the BioRIDII model at
no charge. The work performed in Chapter 3 would not have been possible
without heavy assistance from Guy Nusholtz from Chrysler who shared the
original piglet-ATD datasets and helped to vet some of the analysis methods
used. Also, Dr. Serkan Inceoglu was invaluable as a sounding board to help
analyze the techniques developed. The ATD impact testing performed in
Chapters 4, 5 and 6 would not have been possible without the help from SEA
Limited, including Hank Jebode and Randy Edwards, in addition to Dr.
Vincent Miele who provided human pilot impact data in order to help calibrate
the pendulum impacts. Special thanks go to Art Chou, Brian Hoying and
Greg Mathews and Rawlings sporting goods for providing their prototype
football helmet for the pilot impact tests. The work performed in Chapter 7
requires thanking many people: Dr. Manuel Saavedra and Prasath
Mageswaran assisted in pig spine removal and preparation. Robb Colbrunn
assisted greatly in preparing the kinematic sensors and data acquisition for
VIII
testing. Mike Oliver and Dave Halstead at NOCSAE provided the headforms
used at no charge and answered many questions regarding preparation and
testing. Finally, a portion of this work was supported under the National
Institutes of Health Ruth L. Kirchstein Musculoskeletal Research T32 Training
Grant AR050959.
IX
BIOMECHANICAL ENGINEERING ANALYSES OF HEAD AND SPINE
IMPACT INJURY RISK VIA EXPERIMENTATION AND COMPUTATIONAL
SIMULATION
ABSTRACT
by
ADAM JESSE BARTSCH
Head and spine injuries, such as traumatic brain injury, skull fracture,
concussion and osteoligamentous cervical spine injury continue to be
prevalent in motor vehicle crashes, athletics and the military. Automotive
safety systems, athletic safety equipment and military personal protective
paraphernalia designs have generally focused on protection discretely
designed on a component basis -- head or spine -- but not a systems basis,
considering the head-spine linkage simultaneously. But since the cervical
spine acts as the attachment point for the head, the boundary conditions
applied to the cervical spine influence the behavior of the head. Hence, in
analyzing injury risk for the head and the spine, each structure composes
one portion of an intrinsically linked osteoligamentous system; thus injury
risk for the head and the cervical spine might be more appropriately
considered concurrently as opposed to individually.
Historically, component-based injury protection designs have utilized
head and cervical spine injury risk criteria developed from human, animal
and anthropomorphic surrogate studies. While a plethora of these prior
studies separately analyzed head injury risk via linear acceleration, Head
Injury Criterion (HIC) or Gadd Severity Index (GSI), or cervical spine injury
X
risk via axial/shear forces, bending moments or the Neck Injury Criterion
(Nij), relatively few of these studies employed a systems-based approach to
understand coupled head-cervical spine injury risk behavior.
Thus, designing for optimal head and cervical spine injury protection
may not be as trivial as separate consideration of head or spine component
injury thresholds. Therefore, through a series of six biomechanical
engineering studies that comprised the chapters of this dissertation, the work
presented here broadly investigated head and cervical spine injury protection
on a systems-based approach considering head and cervical spine injury risk
simultaneously. In Chapter 1, injury risk in inertial loading during real-world
low energy minor rear car crashes was analyzed. In Chapter 2, these minor
crashes from Chapter 1 were further investigated via use of numerical
simulation in MADYMO. While Chapters 1 and 2 explored low energy car
crash loading, Chapter 3 explored multivariate head and cervical spine injury
implications from direct head loading during frontal airbag inflation in high
energy experimental car crashes. Chapter 4 expanded the direct frontal
head impact loading analyzed in Chapter 3 to include oblique and lateral
impact loading during impact experiments with a Hybrid III anthropomorphic
test device. The low- and high-energy injury analysis methods developed in
Chapters 1 through 4 helped drive the study of multivariate injury risk in
response to experimental omnidirectional athletic head impacts in Chapter 5.
Chapter 6 further built on the high-energy athletic impacts from Chapter 5
via Matlab and Simulink simulation of helmeted impacts using a systems
dynamics approach. Finally, Chapter 7 analyzed development of an impact
XI
pendulum, pilot cadaveric injury response to direct head impact and analysis
of similar impacts in a helmeted human surrogate. The results of all of these
related studies indicated that head and cervical spine injury risk were
interrelated during direct or inertial car crash and athletic impacts.
XII
CHAPTER 1 – REAL WORLD MINOR REAR CRASHES
1 CHAPTER 1 – REAL WORLD MINOR REAR CRASHES
INTRODUCTION
In this Chapter, real world minor rear crashes in the United States and
the associated injury diagnoses are examined. These crashes, with struck
vehicle speed change of less than 15km/h, have long been implicated to cause
‘whiplash’-related head and neck injury and cost many billions of dollars yearly
in lost wages, insurance costs and medical bills. Yet, little is actually known
about the actual injuries seen in the field relative to vehicle kinematics. In this
Chapter, the engineering reconstruction method using momentum-energy-
restitution is briefly explored while much of the analysis pertains to gathering
and analyzing diagnosed injury for the vehicle occupants.
BACKGROUND
Minor rear crashes in the United States have been implicated as a
significant source of head and spine injury related to inertial and/or direct
loading, including the ubiquitous ‘whiplash’ injury. Unfortunately, there
currently exists a paucity of real world minor rear aligned crash data for crashes
with struck vehicle delta-V less than or equal to 15 km/h. Because of this lack
of real world data, surprisingly little is actually known about the prevalence of
head and spine injury, including ‘whiplash’, in these crashes. And while previous
large-scale retrospective studies from Germany39;62 and Sweden77 have
examined the relationship between minor rear crash struck vehicle delta-V and
risk of occupant ‘whiplash’, head and spine injury, no large scale studies exist
for U.S. minor rear crashes to adequately define injury risk.
Eis et al. (2005) examined the German In-Depth Accident Study (GIDAS)
and found that for single rear impact crashes involving 1,724 struck vehicle
1
occupants, 59% of occupants were uninjured and 39% sustained a minor AIS13
injury. Of the AIS1 injured occupants, injury data for crashes with engineering-
reconstructed struck vehicle delta-V less than or equal to 15 km/h were
available for 314 occupants. For these 314 occupants, 69% had AIS1 neck
injury, 16% had no neck injury and 15% had AIS1 neck injury in addition to
‘other’ injuries. Hell et al. (2002) analyzed crashes from the Gesamtverband der
Deutschen Versicherungswirtschaft (GDV) database. These authors found that
of 88 occupants alleging ‘whiplash’ injury in struck vehicles with PC-Crash
reconstructed delta-V less than or equal to 15 km/h, all had grade 0, 1, 2 or 3
‘whiplash associated disorders’ (WAD) according to the Quebec Task Force.146
Krafft et al. (2005) conducted a study of Swedish crashes pooled from more
than 60,000 vehicles equipped by Folksam Insurance with event data recorders
(EDR). From these vehicles, a total of 171 struck vehicle occupants were
involved in rear crashes with EDR recorded delta-V less than or equal to 15
km/h. Of these struck vehicle occupants claiming injury, 67% had no ‘whiplash’
injury, 26% had AIS1 ‘whiplash’ injury for less than one month post-crash and
7% had AIS1 ‘whiplash’ injury for longer than one month post-crash. Of
particular note to the current study was that in both of the minor rear crash
German studies and the Swedish study, the authors made no mention of any
claimed or diagnosed struck vehicle occupant injuries to the head, thorax,
abdomen, thoracolumbar spine, upper extremities or lower extremities.
There are few U.S. minor rear crash data to directly compare with the
aforementioned European studies since no dedicated minor crash database
collection currently exists in the U.S. Tencer et al. (2001) had a sizeable study
cohort relating engineering reconstructed struck vehicle delta-V to occupant
injury complaint frequency.152 A total of 432 struck vehicle occupants,

2
previously diagnosed with ‘whiplash’ attributed to rear crashes with peak delta-V
less than 11.3 km/h, were analyzed. All subjects were proceeding to litigation.
Immediate symptoms were considered but the specific number of days between
the crash and symptom onset was not reported. A total of 174 subjects had
AIS1 neck and upper back pain at mean delta-V of 8.0 km/h (s.d. = 2.1 km/h),
174 subjects had AIS1 neck and low back pain at mean delta-V of 7.7 km/h
(s.d. = 2.9 km/h) and 84 subjects had AIS1 neck pain and arm symptoms at
mean delta-V of 8.8 km/h (s.d. = 5.0 km/h). The authors concluded that pre-
existing lumbar degeneration was associated with (a) having arm symptoms or
(b) AIS1 low back pain with neck pain. While the entire study cohort reported
AIS1 neck pain, there were no reported head, thorax, abdominal or lower
extremity complaints.
While much can be learned from the prior studies, further U.S. minor rear
crash analysis is needed in order to develop an understanding between struck
vehicle delta-V, and occupant ‘whiplash’, head and neck injury risk. The lack of
available U.S. data is particularly startling when considering that the National
Highway Traffic Safety Administration (NHTSA) estimates that approximately
806,000 occupants sustain ‘whiplash’ and minor head injuries in motor vehicle
crashes producing modern day economic and quality of life costs over $9 billion
yearly79. Over the past six decades, much has been done to improve the
understanding of specific ‘whiplash’ and head injury mechanisms, patient
treatment protocols, and vehicle bumper, seat and head restraint designs. Yet,
even with the voluminous amount of research in the engineering and medical
fields, ‘whiplash’ and minor head injuries in rear crashes have increased since
the 1960’s.73;99 Furthermore, insincere or inflated minor crash injury claims
under the U.S. tort system have been cited as a contributor to current excess
3
health care costs of approximately $5.8 billion and increased insurance
premiums of $19 to $26 billion yearly.18
In the current study, a newly collected minor rear crash forensic
engineering data set was analyzed. Crash data were sought from forensic
engineering sources since this is one of the few sources of large numbers of
minor rear crashes in the U.S. The study intended to be a first step in
answering several fundamental questions regarding minor rear crashes and
‘whiplash’, spine and head injury in the U.S. Primarily, what are the delta-V and
mean acceleration associated with diagnosed complaints, including ‘whiplash’, in
minor rear crashes in the U.S.? Secondly, what is the complaint diagnosis
distribution in these crashes based on ICD-9-CM156 and AIS coding? Thirdly, can
currently known minor rear crash injury mechanisms explain the ICD-9-CM and
AIS coded diagnoses in these crashes? And lastly, does ICD-9-CM coding from
medical records differ when compared with AIS coding? Armed with answers to
these fundamental questions, safety engineers and clinicians alike will be able to
design vehicle systems to reduce ‘whiplash’, minor head injury and spine injury
incidence, develop improved treatment protocols to specifically target the most
frequently injured body locations in minor rear aligned crashes, improve
insurance claims processing for legitimate minor rear crash injury and more
efficiently identify and remedy potentially disingenuous minor rear crash
insurance claims.
METHODS
Real world minor U.S. rear crashes were sought via (I) a database search
and (II) a review and analysis of crash files from a forensic engineering
company. Inclusion criteria for the current study limited data to (a) U.S. real
4
world rear crashes, (b) crashes between two passenger vehicles with gross
weight less than 4,536 kg for each vehicle, (c) single contact between the front
of the striking vehicle and rear of the struck vehicle, (d) aligned contact between
the two vehicles, (e) full overlap between the two vehicles, (f) crash occurring
on a flat roadway, (g) reconstructed struck vehicle delta-V less than or equal to
15 km/h and (h) medical documentation of post-crash occupant complaint
diagnoses for visits to a medical doctor (M.D.), osteopathic doctor (D.O.) or
chiropractor (D.C.).
(I) Database search
A database search for U.S. real world minor rear crashes was conducted.
The Crash Injury Research Engineering Network (CIREN), National Automotive
Sampling System Crashworthiness Data System/General Estimates System
(NASS), Fatality Analysis Reporting System (FARS), Special Crash Investigations
(SCI) and the National Trauma Databank (NTDB v6.2) were searched for all data
available as of 2007. These databases were generally predisposed to higher
energy towaway crashes with serious injuries, or had limited information on
minor rear crashes.
(II) Forensic engineering analysis
Since the database search and aforementioned literature yielded minimal
data from which to study minor rear crashes and resulting head and spine injury
risk, real world crash files from a forensic engineering company in the U.S. (SEA
Limited, Columbus, OH) were examined for crashes occurring from 1994 to
2006. In all crash files, occupants filed at least one property damage or injury
claim with a U.S. insurance company and were potential litigants in a U.S. court.
These files included available vehicle and scene inspections, police reports,
insurance repair estimates, occupant medical records and occupant statements

5
utilized for an engineering crash reconstruction. From these files, 105 minor rear
aligned crashes from 15 states met the inclusion criteria. These crashes
involved a total of 126 occupants in the striking vehicles and 151 occupants in
the struck vehicles. Front seat struck occupants totaled 137 persons. A total of
113 struck vehicle occupants were diagnosed with complaints attributed to the
crash within five weeks post-crash by a M.D., D.O. or chiropractor. No striking
vehicle occupants reported injury complaints or sought medical treatment.
For the 113 struck vehicle occupants, medical records were examined for
complaint diagnoses within five weeks post-crash. These records were obtained
according to the U.S. Health Insurance Portability and Accountability Act
(HIPAA) requirements158. Additionally, these records were redacted such that all
occupant-specific identifiers were removed. Unique claimed complaint and
injury diagnoses were compiled for this period based on International
Classification of Diseases 9th Edition-Clinical Modification (ICD-9-CM) and
Abbreviated Injury Scale 2005 (AIS) coding requirements. A unique ICD-9-CM
diagnosis was defined as any distinctive diagnosis given to the occupant during
the five week treatment period. Thus, the occupant could be diagnosed with a
cervical sprain repeatedly during visits to different clinicians, but this diagnosis
would only count for the first visit when it was originally diagnosed. Unique AIS
diagnoses differed slightly from ICD-9-CM in that AIS codes included only acute
injuries. Pain diagnoses were recorded as AIS1 injuries. Each crash was
individually reconstructed and the struck vehicle delta-V and acceleration were
calculated. The minor crash reconstruction methodology utilized follows.
Crash reconstruction
Reconstruction methods in aligned crashes between two vehicles utilize
fundamental physical collision properties to determine vehicle delta-V and mean

6
acceleration. The impulse-momentum relationship, energy balance relationship,
normal force restitution and kinematic equations of motion for the vehicles can
be used to solve for delta-V and mean acceleration:
Impulse-momentum relationship:
FT dt  m1v11  m2 v21   m1v12  m2 v22 

t2
t1
Eq. 1-1
Energy balance relationship:
 m1v12 2  m2 v 22 2 
m1v11  m2 v 21
2 2
  w2 C2 m BEVi
2
2
 2  Eq. 1-2;   FCi dCdw  i
2
Eq. 1-3
 w 2 C 2 F dCdw 
w1 C1
w1 C1 Ci 
Kinematic equations of motion:
vi
vi 2  2a i xi Eq. 1-4; vi  vi 2  vi1 Eq. 1-5; ai  Eq. 1-6
t
Restitution:
v 22  v12
e Eq. 1-7
v11  v 21
Where:
mi  vehicle ‘i’ mass (kg)
v1i  vehicle ‘i’ pre-crash velocity (m/s, ‘i’ = 1,2)
vi 2  vehicle ‘i’ post-crash velocity (m/s, ‘i’ = 1,2)
ai  vehicle ‘i’ average acceleration (m/s2)
xi  vehicle ‘i’ linear displacement from impact area to final rest position (m)
vi  vehicle ‘i’ velocity change (m/s)
t  impact duration (s)
e  restitution coefficient for crash contact
BEVi  vehicle ‘i’ barrier equivalent velocity (m/s, ‘i’ = 1,2)
t2
 F dt  impulse integral of vehicle tire forces over contact duration for both vehicles (N-s)
t1
T
w2 C2
  F dCdw  vehicle ‘i’ energy expenditure during crash due to contact forces, crush depth
w1 C1
Ci
and vehicle width (N-m, i = 1,2)
Delta-V and average acceleration were used to quantify crash severity as
these parameters have been most often used in prior studies.39;77;138 Delta-V is
different than barrier equivalent velocity (BEV). BEV is a derived quantity
relating permanent crush deformation energy in a real world crash to the same
7
type of vehicle in a controlled barrier crash test. The BEV is equal to the barrier
impact speed and can be calculated as shown in Eq. 1-3. Given the same real
world and crash test vehicle with similar crush profiles, the real world vehicle
BEV and energy dissipation is approximately the same as the crash test vehicle.
Higher energy crash reconstruction methods are not utilized here because
there is minimal permanent crushing for both vehicles in the crash, the collisions
are treated as fairly elastic, restitution must be considered and tire force impulse
is a contributing factor.17;149
Minor crash reconstruction estimates
Often after minor rear crashes, there is minimal or no crush to one or
both vehicles. Thus, when crush is absent post-crash, one cannot determine
BEV or energy expended based on crush as is done in higher energy crashes.
One possible solution for this problem has been to assume a small crush that is
not actually present and use this value with the vehicle stiffness to calculate a
BEV29. Unfortunately, assuming a small crush amount overestimates minor
crash delta-V in vehicle-vehicle impacts by an average of 110% for delta-V from
1.8 km/h to 15 km/h. Furthermore, computer crash reconstruction programs,
such as m-SMAC, m-CRASH, WinSmash, Crash3, and HVE-EDCRASH use
algorithms based on barrier crash tests and require measurable crush in order to
determine vehicle delta-V. These programs are not validated for minor crashes
and inaccuracies of these reconstruction programs for minor crashes has been
shown previously.106
Thus, accurately estimating minor crash delta-V requires a different
reconstruction methodology than that classically used in higher energy crashes.
In minor crashes, one can still apply methods shown in Equations 1-1 through
1-6, but the vehicles are assumed to behave elastically. This means that the
8
restitution relationship in Eq. 1-7 must be considered. Published minor rear
crash data provided for vehicle weight, impact duration, restitution, crush
energy and bumper properties. These data, along with calculated and estimated
pre- and post-crash values were inserted into Equations 1-1 through 1-7 and
simultaneously solved in this system of seven equations for up to seven
unknowns.
This minor crash reconstruction method, commonly known as the
momentum-energy-restitution (MER) method, has been previously shown in
vehicle-vehicle crash tests to estimate struck vehicle delta-V for a variety of
modern bumper constructs with a RMS error of 4% to 9%.5;61;137 In tandem
with the minor crash reconstructions, the claimed occupant ICD-9-CM
complaints and AIS injury diagnoses and crash dynamics were compiled and are
given in sections (i)-(viii).
RESULTS
(i) Crash occupants
CRASH OCCUPANTS
mean age ± s.d. mean weight ± s.d. mean height ± s.d.
Male
(years) (kg) (m)
(n = 36)
39.3 ± 12.0 92.1 ± 12.9 1.76 ± 0.08
mean age ± s.d. mean weight ± s.d. mean height ± s.d.
Female (years) (kg) (m)
(n = 77)
40.3 ± 12.0 78.5 ± 22.8 1.62 ± 0.07
Table 1-1. The 113 struck vehicle occupants with crash related diagnoses
(ii) Crash summary
Each crash was reconstructed using the minor crash MER method
discussed previously. Struck vehicle reconstruction delta-V and average
acceleration results are shown in Table 1-2, Figure 1-1 and Figure 1-2. For all
105 crashes, reconstruction references provided an impact duration range of
9
0.080 to 0.200 seconds with a mean of 0.136 seconds (s.d. = 0.024 seconds).
Restitution references varied from 0.10 to 0.70 with a mean restitution of 0.25
(s.d. = 0.11).
RECONSTRUCTION SUMMARY
Delta-V (km/h) Mean acceleration (g)
mean 6.3 mean 1.4
s.d. 2.1 s.d. 0.5
max 13.7 max 3.3
min 0.8 min 0.3
Table 1-2. Reconstruction struck vehicle delta-V and mean acceleration (n=105 crashes)
50
45
40
35
Frequency
30
25
20
15
10
5
0
< 4.0 4.0-6.0 6.1-8.0 8.1-10.0 10.1- 12.1-
12.0 15.0
Delta-V (km/h)
Figure 1-1. Struck vehicle delta-V
60
50
40
Frequency
30
20
10
0
< 1.0 1.0-1.4 1.5-1.9 2.0-2.4 2.5-3.5
Mean Acceleration (g)
Figure 1-2. Struck vehicle acceleration
10
Of these 105 minor rear crashes, 52 occurred between two passenger
vehicles, 24 occurred with light truck/ sport utility vehicles (SUV) striking a
passenger vehicle, 17 occurred with a passenger vehicle striking a light
truck/SUV and 12 occurred between two light truck/SUV. Mean striking and
struck vehicle weights were 1688 kg (s.d. = 506 kg) and 1509 kg (s.d. = 300
kg), respectively. Striking vehicle front bumper composition was rigid (n = 38),
piston (n = 17), polymer (n = 11), foam (n = 26) and box (n = 3). Struck
vehicle rear bumper composition was rigid (n = 23), piston (n = 38), polymer (n
= 12), foam (n = 24) and box (n = 8).
Of all 210 vehicles involved, eight struck and two striking vehicles were
towed from the scene. This low percentage of tow-away crashes indicated that
the crash set studied here would not have been captured in U.S. towaway crash
databases. For the 75 struck vehicles with IIHS geometric head restraint
ratings, 50 were rated as ‘poor’, 12 were ‘marginal’, 12 were ‘acceptable’ and 1
was ‘good’.
There was no airbag deployment in any of the 105 crashes. Police
recorded belt restraint usage for 68 of the struck occupants seeking medical
treatment, with 67 reporting use of three-point restraints and one reporting no
restraint usage. Of the 71 striking vehicle occupants reporting restraint usage,
69 reported three-point restraint usage, one reported lap belt usage and one
reported no restraint use. Police reports noted the posted speed limit in 54 of
the crashes, with 33 crashes occurring in speed zones less than or equal to 56.3
km/h (35 mph).
11
(iii) Medical treatment
Post-crash, the 113 struck vehicle occupants first sought treatment at
emergency room (ER), urgent care facilities or medical office locations as seen
in Figure 1-3.
Male Female
50
45
40
35
Frequency
30
25
20
15
10
5
0
Office ER/Urgent Care
Figure 1-3. Medical treatment facility for first visit post-crash (n=113 struck vehicle occupants)
The time between the crash and first medical treatment visit was
available for 111 of the struck occupants. One male occupant made an initial
treatment visit to both the ER and a medical office on the same day. Any visit
occurring the same day as the crash was given a value of zero days. Figure 1-4
details the first visit medical treatment distribution. The consulting clinician for
the first treatment was available for 85 occupants and is shown in Figure 1-5
Male Female
20
Days to First Visit Post-Crash
18
16
14
12
n = 13 n = 28
10
8
6 n = 23
4 n = 48
2
0
Office ER/Urgent Care
Figure 1-4. Days until first treatment visit
12
Male Female
40
35
30
Frequency
25
20
15
10
5
0
M.D. D.O. D.C.
Figure 1-5. Clinician visited first (n=85 occupants)
(iv) Unique ICD-9-CM and AIS diagnoses
The unique ICD-9-CM and AIS diagnoses for the first medical treatment
visit and all visits are shown in Figures 1-6 and 1-7, respectively.
50%
Percent of Diagnoses
First visit (n=410 dx.)

45%
40% 5 week visits (n=761 dx.)
35%
30%
25%
20%
15%
10%
5%
0%
er
d
ip
en
FS
l
r
c
ca
ra
de
ci
ea
th
ac
om
a
vi
N
ul
x/
H
O
or
er
r/S
ho
ck
bd
Th
C
er
Ba
/S
ba
/A
w
Ex
st
m
Lo
he
Lu
er
C
pp
U
Figure 1-6. ICD-9-CM diagnoses
50%
First visit (n=256 dx.)

45%
40% 5 week visits (n=427 dx.)
35%
30%
25%
20%
15%
10%
5%
0%
er
d
ip
en
FS
l
r
c
ca
ra
de
ci
ea
th
ac
om
a
vi
N
ul
x/
H
O
or
er
r/S
ho
ck
bd
Th
C
er
Ba
/S
ba
t/A
w
Ex
m
Lo
s
he
Lu
er
C
pp
U
Figure 1-7. AIS coding
13
(v) Degenerative condition diagnoses
A total of 67 struck vehicle occupants were diagnosed within five weeks
post-crash with at least one degenerative spine condition pre-existing the crash
date. The total degenerative diagnosis distribution is shown in Figure 1-8.
These conditions were broken into five categories for the cervical, thoracic and
lumbar/sacral spine: (a) disc bulging, protrusion and herniation, (b) spinal
stenosis and narrowing, (c) vertebral spurring and osteophytes, (d)
degenerative disc disease (DDD), degenerative joint disease (DJD), radicular
and myelopathic symptoms and (e) and abnormal spinal structure such as
spondylolisthesis or abnormal spine curvature. Of these degenerative
occupants, 52, nine and 29 occupants were diagnosed with at least one cervical,
thoracic and lumbar/sacral degenerative condition, respectively. These
degenerative diagnosis occupants had AIS1 injury diagnoses attributed to the
crash during the five week treatment period. The distribution of AIS1 injuries
for these occupants is shown in Figure 1-9 as compared with the 46 non-
degenerative diagnosis occupants.
14
120
DDD/DJD
(d)
100
80
Frequency
Spurring Abnormal
60 Disc structure
(c)
(a) (e)
40
Stenosis
(b)
20
Figure 1-8. Degenerative diagnoses
50%
Degenerative Occupants (n=262 dx.)

45%
40% Non-degenerative occupants (n=135 dx.)
35%
30%
25%
20%
15%
10%
5%
0%
r
ip
en
l
d
ic
ca
ra
de
ea
ac
m
ac
vi
ul
x/
H
or
o
er
r /S
ho
bd
h
C
er
T
/S
ba
/A
w
Ex
st
m
Lo
he
Lu
er
C
pp
U
Figure 1-9. AIS1 injuries
(vi) ‘Whiplash’ diagnosis results
Occupants initially diagnosed with transient AIS1 ‘whiplash’ within one
week post-crash were analyzed. These ‘whiplash’ diagnoses were often
accompanied by other non-‘whiplash’ diagnoses to other body regions. This
resulted in three groups: a ‘whiplash’ only initial diagnosis group (n=20), a
‘whiplash’ in addition to ‘other’ initial diagnoses group (n=53) and a group with
‘other’ initial diagnoses only (n=25). Occupants who made their first treatment
15
visit beyond one week post-crash (n=15) were excluded in this analysis. For
these groups, all unique AIS1 injury diagnoses from the five week treatment
period were compiled and are shown in Figure 1-10. Non-specific AIS1 injury
diagnoses were not included in this analysis.
Percentage of Diagnoses
'Whiplash' (n=50 dx.)
60%
'Whiplash' + 'Other' (n=208 dx.)
50% 'Other' (n=87 dx.)
40%
30%
20%
10%
0%
ip
ic
l
er
en
al
ca
ea
ac
/H
r
ld
om
ac
vi
ou
Ex
or
er
r/S
bd
Th
Sh
C
er
ba
/A
x/
w
m
Lo
st
Lu
he
er
pp
C
U
Figure 1-10. AIS1 diagnoses comparing ‘whiplash’ groups during five week treatment period post-
crash
(vii) Comparing ‘whiplash’ to cervical degenerative diagnoses and

comparing lumbar/sacral injury diagnoses to lumbar/sacral
degeneration
The relationship between diagnosed ‘whiplash’ and pre-existing
degenerative cervical spine conditions was examined. The null hypothesis was
that the presence of at least one diagnosed cervical degenerative condition
(Group A, n=35 occupants) would not result in a statistically significant
difference in diagnosed AIS1 ‘whiplash’ injury rate over occupants without
cervical degeneration (Group B, n=38 occupants). The Group B diagnosed AIS1
head and cervical spine injury rate over the five week treatment period defined
the mean of the population. A paired, two-tailed, t-test comparison was made
based on a p-value less than 0.05 being significant for a difference in the
diagnosed AIS1 ‘whiplash’ injury rate in Group A occupants. The comparison
between these groups is shown in Figure 1-11. Based on the p-value of 0.98 for
this comparison, the null hypothesis was accepted.
16
3.0
p = 0.98
2.5
AIS 1 'Whiplash' Rate

Group A
2.0 Group B
(n = 35)
(n = 38)
1.5
1.0
0.5
0.0
Figure 1-11. ‘Whiplash’ in degenerative cervical (Group A) and non-cervical degenerative

occupants (Group B)
The relationship between occupants diagnosed with AIS1 lumbar/sacral
injury and occupants with pre-existing lumbar/sacral degeneration was
analyzed. The null hypothesis was that occupants with pre-existing
lumbar/sacral degenerative conditions (Group C, n=16 occupants) would have
no statistically significant difference in the diagnosed AIS1 lumbar/sacral injury
rate over occupants without lumbar/sacral degeneration (Group D, n=28
occupants). The Group D diagnosed AIS1 lumbar/sacral injury rate over the five
week treatment period defined the mean of this population. Again, a paired,
two-tailed, t-test comparison was made based on a p-value less than 0.05 being
significant for a difference in the diagnosed lumbar/sacral injury rate in Group C
occupants. The results of this comparison are shown in Figure 1-12. The null
hypothesis for this analysis was accepted based on the resulting p-value of 0.30.
17
2.5
p = 0.30
AIS 1 Lumbar/Sacral Injury Rate

2.0
Group D
1.5 Group C (n = 28)
(n = 16)
1.0
0.5
0.0
Figure 1-12. AIS1 lumbar/sacral injury rate in degenerative (Group C) and non-lumbar/sacral
degenerative occupants (Group D)
DISCUSSION OF RESULTS
The 105 minor crashes involved totals of 151 and 126 occupants in the
struck and striking vehicles, respectively. These crashes were referred to as
‘minor’ since this description was deemed to be more representative of the
severity and resulting AIS1 diagnoses in these crashes than commonly used
terms such as “low-speed” or “whiplash” crashes. Of the struck vehicle
occupants, 113 were diagnosed by a M.D., D.O. or chiropractor within five
weeks of the crash with complaints attributed to the crash. None of the striking
occupants were diagnosed with complaints attributed to the crash. The crash
reconstructions estimated a mean delta-V and acceleration of all struck vehicles
of 6.3 km/h (s.d. = 2.1 km/h) and 1.4g (s.d. = 0.5g).
Nearly ninety percent of struck vehicle occupants made their first medical
treatment visit within one week post-crash. Female and male occupants made
their first visit to emergency room and urgent care facilities approximately twice
as often as visiting a medical office location, and visited M.D. or D.O. more often
than chiropractors. This was likely because these emergency facilities were
staffed by M.D. and D.O. as opposed to chiropractors. For those occupants first
visiting a chiropractor, women had a higher percentage of visits than men. For
18
the occupants first visiting medical office locations, the time until first visit was
highly variable but these occupants averaged over eight days until their first
treatment visit post-crash.
The first medical treatment visits resulted in a total of 410 unique ICD-9-
CM and 256 unique AIS1 diagnoses attributed to the crash. For the five week
post-crash diagnosis period, 761 unique ICD-9-CM diagnoses and 427 unique
AIS diagnoses were reported. Of these AIS diagnoses, two AIS2 injuries (lumbar
fracture, meniscus tear) were reported. It was questionable whether the crash
forces were of sufficient magnitude and direction to have caused the diagnosed
AIS2 injuries. ICD-9-CM had more total diagnoses than the AIS system due to
the inclusion of pre-existing degenerative condition diagnostic codes.
The diagnosis distributions changed when comparing the first visit
diagnoses with the five week diagnosis period, ICD-9-CM and AIS spine
diagnoses decreased by 2.6% and 9.0% of the total diagnoses during the five
week period, respectively. Over this same time period, head, chest and
extremity complaints increased by totals of 2.8% and 9.8%, respectively.
Reasons for the decreasing number of spine related diagnoses could have been
due to the resolution of transient spine complaints after the first week of
treatment post-crash. The increasing proportion of non-spine transient
complaints during the post-crash treatment period requires more investigation.
The combination of non-’whiplash’ ICD-9-CM and AIS diagnoses to the
lumbar/sacral spine, thoracic spine and extremities exceeded that of head and
cervical spine ‘whiplash’ related diagnoses. Since the most at-risk body parts in
these minor rear crashes were the head and cervical spine, and based on the
results of prior studies,39;62;77;152 it was hypothesized that an overwhelming
majority of all struck occupants with complaints would have ‘whiplash’ and head
19
injury complaints. The prevalence of non-‘whiplash’ complaints was higher than
previous studies of real world ‘whiplash’ complaints.15;26;39;51;62;72;77;95;96;131;134
The diagnosis frequency for the thoracic spine, lumbar/sacral spine and
extremities has not been reported in previous minor real world rear crash
research, has not been produced in human volunteer testing at similar or more
extreme crash severities and did not correlate with accepted minor rear crash
injury mechanisms.8;11;20;26;39;62;72;77;131;134;152;162 And while some authors have
cited frontal crashes as a source of ‘whiplash’ and head injury,31;78 none of the
striking vehicle occupants sought treatment post-crash for any complaints
including ‘whiplash’ or head injury. The lack of correlation between complaint
diagnoses in the current study and prior studies may have been due to the
presence of litigation influences in the study cohort. Comparison with an
unbiased cohort is needed to fully verify or refute the ICD-9-CM and AIS
diagnosis distributions and lack of frontal crash ‘whiplash’ in the current study.
When comparing the current AIS1 injury diagnoses to prior large minor
rear crash studies, discrepancies arose. While the struck vehicle delta-V range
in the current study, less than or equal to 15 km/h, was of similar magnitude or
39;62;77;152
below prior minor rear crash studies , the number of front seat struck
vehicle occupants who claimed injury was very different. In the Krafft et al.
(2005) study, it was reported that 33% of 171 front seat occupants reported an
AIS1 injury complaint while in the Eis et al. (2005) study, 41% of 1,724 front
seat occupants reported AIS1 or greater injury. These injury rates are much
lower than the 110 of 137 (80%) front seat occupants claiming AIS1 injury in
the current study. Secondly, in the Hell et al. (2002), Krafft et al. (2005) and
Tencer et al. (2001) studies, occupants who reported ‘whiplash’ complaints
comprised 100% of injured occupants; in the Eis et al. (2005) study neck
20
complaints constituted 84% of claimed AIS1 injuries. In the current study
‘whiplash’ complaints were present in only 65% of occupants seeking medical
treatment. Finally, the large number of claimed non-‘whiplash’ diagnoses did
not correlate with any data presented in these prior studies. The reasons for
such large differences in complaint type, ‘whiplash’ prevalence and non-
‘whiplash’ complaints in the current minor rear crash cohort is not fully
understood. It is possible that the influence of litigation in this sample might
have affected the frequency, duration and type of complaint diagnoses.
For the 67 occupants diagnosed with at least one pre-existing
degenerative spine condition during the five week post-crash visit period,
degenerative disc disease (DDD) or degenerative joint disease (DJD) diagnoses
were in the majority. Similar proportions were found between disc bulging,
spurring and abnormal spine structure. When AIS1 diagnoses for these
occupants were compared to the 46 non-degenerative occupants, cervical,
lumbar/sacral and thoracic spine diagnoses predominated. It is unknown why
the percentage of lumbar/sacral and thoracic spine AIS1 injury diagnoses were
high compared with cervical diagnoses since the cervical spine was the most at-
risk spine region during these crashes.
One of the more interesting findings of this study was that when
compared with the non-degenerative occupants, pre-existing degenerative
condition occupants were not statistically significant for different injury rates to
the cervical and lumbar/sacral spine, respectively. For the cervical comparison,
when the 35 occupants in Group A and 38 occupants in Group B were compared,
cervical degeneration was not a statistically significant factor (p = 0.98) for a
difference in the AIS1 ‘whiplash’ injury rate. This result bears resemblance to
prior studies.128;167 For the lumbar/sacral comparison, 16 occupants in Group C

21
and 28 occupants in Group D were analyzed. It was found that pre-existing
lumbar/sacral degeneration was not a statistically significant factor (p = 0.30)
for a different lumbar/sacral AIS1 injury rate. This comparison indicated that
normal occupants tended toward increased AIS1 injury rates over degenerative
occupants. This finding was in contrast with that of a prior study.152
While the current study examined many facets of minor rear aligned
crashes and ‘whiplash’ injury diagnoses, the study had weaknesses. Since there
were no publicly available minor rear crash U.S. databases, and the forensic
engineering crash cases were potentially subjected to litigation biases, the
authors were not able to select an unbiased sample cohort for comparison. A
second weakness is that the struck vehicle delta-V values obtained from the
minor crash reconstruction method had some amount of error. While it would
have been desirable to conduct crash tests with exemplar vehicles and
occupants for all 105 crashes in order to validate the reconstruction results, the
scope and cost of this type of testing simply was prohibitive for this study.
Though the crash reconstructions were not validated by crash testing, the minor
crash reconstruction methodology used here has been shown to be within 10%
of the actual struck vehicle delta-V in modern vehicle crash tests. Thus, the
mean error present in the struck vehicle delta-V reconstruction estimates would
have been on the order of 0.6 km/h. This error would have had minimal effect
on consideration of occupant kinetics and kinematics in these crashes while
tending to mildly overestimate the crash severity.
The current results indicate that minor rear ‘whiplash’ crashes might
result in a large proportion of diagnoses that are unrelated to ‘whiplash’.
However, this finding does not agree with any similar publications regarding real
39;62;77;117;134
world minor crashes
22
LIMITATIONS
This study had several limitations. First, because of the lack of
large numbers of publicly available real-world minor rear crashes, the
sample size was limited to the 105 crashes studied here. Due to this small
study sample size and biased data toward litigants or potential litigants, a true
comparison with previously published data from Europe is tenuous. Second,
while the Momentum-Energy-Restitution reconstruction methodology used here
is accepted as an Industry standard, this methodology contains errors that can
only be truly quantified via vehicle crash testing. Hence, while delta-V values
calculated using this methodology have been generally shown to be within 5%
to 9% of their true values, the actual error rate for this study is unknown.
Finally, there were many psychological and socio-economic interactions that
were not studied here that have been universally acknowledged to contribute to
‘whiplash’ symptoms and diagnoses. This study did not examine psychological
profiles of struck or striking occupants. Further, while socio-economic concerns
have been posited in European studies as playing a role in ‘whiplash’ complaints
and litigation, these variables were not studied here. These interactions likely
played some type of a role in this dataset, and while they may have been
partially responsible for the discrepancy in struck/striking occupant injury
claims, any relationship remains hypothetical at best. These limitations
underscore a need for a larger, controlled, and unbiased U.S. minor crash study
is needed to examine occupant diagnoses due to minor rear crashes for delta-V
less than or equal to 15 km/h. And development of a publicly available
database modeled after NASS or CIREN for minor rear crashes would be
extremely useful to researchers and clinicians investigating ‘whiplash’ and the
23
corresponding crash dynamics when struck vehicle delta-V is less than or equal
to 15 km/h.
CONCLUSIONS
In the current study, while searching available crash databases and the
published literature, it was found that few data exist to study actual ‘whiplash’
and head injury complaints in real world U.S. minor rear aligned crashes at
delta-V less than or equal to 15 km/h. This required gathering real world
crashes from a U.S. forensic engineering company archives in an effort to study
‘whiplash’, head and spine injury diagnoses. In this forensic crash data set,
reconstructed delta-V, acceleration and occupant diagnoses were gathered.
From 105 real world minor U.S. crash reconstructions, the mean delta-V was
found to be 6.3 km/h and mean acceleration was 1.4 g. The majority of struck
vehicle delta-V and accelerations were at or below delta-V or mean acceleration
threshold ranges for AIS1 injuries found in volunteer and real-world studies.
Even with these crashes being of minor severity, a majority of struck occupants,
113 out of 151, had complaints attributed to the crash that were diagnosed by a
M.D., D.O. or chiropractor within a five week treatment period post-crash. Many
of these diagnoses failed to include head or spine injury. The prevalence of
complaints to protected and cushioned body regions, such as the lumbar/sacral
spine and extremities, has never been found in any prior study investigating real
world minor rear crashes. This finding must be validated by an unbiased,
prospective U.S. cohort minor rear crash study and merits extremely cautious
interpretation in the interim.
Based on the injury exposure risk of the head and cervical spine due to
direct loading from the head restraint and inertial loading from the head-neck
24
region, and the findings of prior minor rear crash studies, the authors expected
to discover a definitive majority of ‘whiplash’ and head injury complaints in
these crashes. In these low-energy minor rear crashes, this anticipated
definitive majority of ‘whiplash’/head injury was not found. The frequency of
diagnosed complaints to body locations other than the head and cervical spine
seen here has not been reported in real world studies performed in other
countries. It was unclear why no striking vehicle occupants reported medical
complaints attributed to the crash as ‘whiplash’ and head injuries has also been
attributed to inertial and direct impact loading in minor frontal crashes.
Additionally, while pre-existing degeneration has been posited as a risk factor
for increased cervical or lumbar/sacral injury, no significant relationship was
found between degeneration and increased rates of cervical or lumbar/sacral
spine injury in the present data set. It is a distinct possibility that the injury
complaint differences seen here when compared with prior studies could have
been influenced by litigation or potential litigation in this study cohort. It would
be highly desirable to assist ‘whiplash’ and head injury researchers to
prospectively collect unbiased minor crash and occupant diagnosis data in the
model of NASS or CIREN. While developed countries, such as Germany, Japan
and Sweden have initiated programs to investigate occupant ‘whiplash’
complaints related to real world minor rear crash severity, the U.S. has yet to
follow suit.
25
CHAPTER 2 –MINOR REAR CRASH MADYMO SIMULATION
2 CHAPTER 2 – MINOR REAR CRASH MADYMO SIMULATION
INTRODUCTION
In order to quantify the kinetics and kinematics experienced by the
injured crash occupants studied in Chapter 1, this Chapter describes the
development, validation and results from a minor crash simulation using the
MADYMO numerical simulation software. Real world minor rear crash test
vehicle acceleration pulses previously performed were used as inputs to the
simulated vehicle model. Hybrid III and BioRIDII crash test dummies were
placed in the vehicle, subjected to the minor rear crash acceleration pulse,
and virtual load cells, accelerometers and potentiometers quantified the
dynamic loading to the head-neck complex. Multiple variables were studied
to examine potential injury risk to the vehicle occupants. Finally, this
simulation method was chosen over physical experimentation as it provided a
cost-effective alternative to prohibitively expensive vehicle crash testing.
BACKGROUND
In Chapter 1 real world minor rear crashes, where the struck vehicle
longitudinal velocity change, or delta-V, is less than 15 km/h, were explored.
In that analysis, occupant injuries, though minor in nature, were reported at
a high frequency to regions other than the head and spine. To further
understand injury risk to the head, spine and these other regions in this
Chapter, estimates of occupant kinetics, kinematics and injury risk in these
crashes are evaluated through the use of the computational crash model in
MADYMO (MAthematical DYnamic MOdels, TASS-SAFE, Livonia, Michigan,
26
USA). MADYMO is a dynamic modeling package that has been extensively
used in motor vehicle crash safety design and analysis. Hence, development
and validation of a minor rear crash passenger vehicle MADYMO model based
on the real-world crashes studied in Chapter 1 allows for accurate and cost-
effective evaluation of accelerations, forces and moments acting on
occupants in real world minor rear crashes without the need to undertake
expensive and potentially dangerous volunteer or anthropomorphic test
device (ATD) crash testing.
The objectives of this study were to (1) validate a generalized minor
rear crash MADYMO model using the mean of impact data similar to the
minor rear crash tests in Chapter 1 with an instrumented BioRIDII ATD, (2)
use this validated MADYMO model to observe and compare BioRIDII and
Hybrid III 50th percentile minor rear crash responses and (3) evaluate model
kinetics and kinematics in response to a 40% reduced input pulse
approximating a 7.3 km/h delta-V minor rear crash. Particular attention is
paid to head and spine loads, with comparison to existing injury thresholds
and criteria
METHODS AND MATERIALS
The simulated minor rear crash acceleration pulse was taken from the
mean of six (6) late model passenger vehicles weighing less than 4,536 kg
involved in rear barrier minor crash tests with an instrumented BioRIDII ATD
in the driver position. These crash tests involved the same types of vehicles
and Delta-V as those 105 real-world crashes studied in Chapter 1. A
27
summary of the crash tests is shown in Table 2-1 and the specifics of these
tests have been previously published.139
TEST # YEAR MAKE/MODEL Delta-V (km/h)

1.3 1997 Chevrolet/Cavalier 14.9
1.4 1997 Chevrolet/Cavalier 14.7
7.1 2002 Dodge/Caravan 12.7
11.1 2003 Volkswagen/Jetta 10.9
11.2 2003 Volkswagen/Jetta 10.9
19.1 1998 Chevrolet/Silverado 11.2
Mean from six (6) crash tests 12.1
Table 2-1. Passenger vehicle crash tests used as inputs to MADYMO minor rear crash
simulation
From these six minor rear crash pulses, the mean vehicle x-axis, or
longitudinal, acceleration pulse seen in Figure 2-1 was calculated for use in
the MADYMO simulation for the first 225 milliseconds following barrier
contact. Also shown is the mean longitudinal velocity change of the crash
tests over this pulse duration. The mean crash test delta-V was 12.1 km/h.
Figure 2-1. MADYMO longitudinal vehicle crash pulse and delta-V from minor rear crash tests
The mean rear crash pulse was applied to a generic late model vehicle
interior model in MADYMO V6.4 with a BioRIDII ATD seated in the driver and
28
passenger seats as seen in Figure 2-2. Utilizing two BioRIDII ATD allowed
for increased parametric changes in each simulation run, comparative
analysis of BioRIDII side-by-side kinematics and likely decreased total
simulation time required during the validation process.
Figure 2-2. Initial position of BioRIDII facet ATD
(1) VALIDATION – The BioRIDII initial seating position was determined
via a 500 millisecond settling simulation. In this pre-simulation, the BioRIDII
was positioned five millimeters above the seat cushion and in front of the
seat back and allowed to move into a neutral position in response to a
gravitational acceleration field acting on the BioRIDII. The final BioRIDII
position in pre-simulation served as the initial position during validation runs.
Each validation simulation lasted 725 milliseconds. The gravitational
acceleration field was applied to the BioRIDII for the first 500 milliseconds of
this time period to fully settle any minor contact loading discontinuities due
29
to initial position. The crash test pulse acceleration field was applied to the
vehicle in the forward direction during the next 225 milliseconds. Because
the acceleration, force and moment magnitudes the BioRIDII experienced
during the forward rebound phase, after the vehicle and BioRIDII achieved a
common velocity, were small or insignificant after the 225 millisecond rear
pulse period, occupant loading during rebound was ignored.
Important model parameters such as BioRIDII initial position, seat
cushion and head restraint rigidity, seat back and head restraint joint
stiffness and seat back and head restraint orientation, and less vital
parameters such as BioRIDII contact friction and damping, were
parameterized over the course of one hundred forty-two simulation
iterations. Finite element three-point belt restraints were initially applied to
the BioRIDII driver and passenger. The belts were subsequently removed
since these restraints imparted no appreciable accelerations, forces or
moments on the BioRIDII during the simulation that affected kinematic or
kinetic response and their presence reduced computational efficiency.
The model was considered valid when important BioRIDII acceleration
outputs, involving pelvis, thorax and head x-axis linear accelerations and
head y-axis angular acceleration traces fell generally within corridors defined
by +/- one standard deviation of the real world crash test BioRIDII mean
responses. Secondary validation measures, including peak upper neck x-
and z-axis forces, y-axis bending moment, pelvis and head z-axis
accelerations, head and thorax y-axis angular velocities, FMVSS208 frontal
crash neck injury criterion, Nij, potential rear crash neck injury criterion #1,
30
Nkm, and potential rear crash neck injury criterion #2, NIC, were compared
with the crash test maxima and provided further confirmation that the model
adequately represented the BioRIDII crash test response.
(2) COMPARISON WITH HYBRIDIII ATD - After the MADYMO minor
rear crash model was validated for the 12.1 km/h delta-V minor rear crashes,
a 50th percentile HybridIII ATD replaced the BioRIDII in the passenger seat
and the simulation was re-run. The HybridIII initial position was similar to
the BioRIDII initial position. This ATD response comparison was performed
since the HybridIII is a more widely used ATD in vehicle crash testing but has
been shown to have a different response than the BioRIDII in rear crashes.
(3) REDUCED INPUT PULSE - Finally, the 12.1 km/h delta-V crash test
pulse was reduced by 40% and applied to the model for the BioRIDII to
observe the occupant kinetics and kinematics representative of a 7.3 km/h
delta-V minor rear crash. This delta-V was chosen since it was of similar
magnitude to the mean delta-V found in Chapter 1.6
RESULTS
(1) VALIDATION – MADYMO BioRIDII pelvis, head and thorax x-axis
acceleration-time curves, along with the head y-axis angular acceleration-
time response, were compared with BioRIDII crash test corridors for the 225
millisecond crash pulse time period. These accelerations can be seen in
Figures 2-3 through 2-6. The shaded areas on each figure define the
corridors surrounding +/- one standard deviation of the mean crash test
values. As can be seen in the figures, the simulated BioRIDII accelerations
31
generally fell within the crash test corridors and showed good correlation to
the BioRIDII crash test response.
Figure 2-3. Pelvis x-axis acceleration
Figure 2-4. Head x-axis acceleration
32
Figure 2-5. Thorax x-axis acceleration
Figure 2-6. Head y-axis angular acceleration
The peak simulation upper neck x- and z-axis forces, y-axis bending
moment, pelvis and head z-axis accelerations and head and thorax y-axis
angular velocities were compared with the crash test ranges in Table 2-2.
The crash test ranges encompass the average maxima, plus or minus one
standard deviation. As can be seen in Table 2-2, the simulation maxima
33
generally fell within the crash test ranges, with no major differences that had
any significant affect on occupant kinematics.
PARAMETER Test Range MADYMO

Upper neck force-x (N) 14.2 to 49.4 45.4
Upper neck force-z (N) 141 to 321 180.5
Upper neck moment-y (N-m) 4.8 to 12.2 11.2
Pelvis acceleration-z (g) -1.4 to -2.8 -1.9
Thorax acceleration-z (g) -2.0 to -4.8 -2.7
Head acceleration-z (g) 3.5 to 5.7 1.0
Head angular velocity-y (rad/s) 9.6 to 16.2 5.9
Thorax angular velocity-y (rad/s) 6.2 to 11.0 6.3
Table 2-2. Comparison of additional validation parameters
Based on the results of Table 2-2, the upper neck forces and moments
were analyzed for potential injury risk. When comparing crash test and
simulation upper neck forces and moments to accepted tolerance values, the
upper neck shear force was at least 95% below the threshold, the axial
tension force was at least 80% below the threshold and the extension
moment was at least 80% below the accepted injurious levels,
respectively.144
To further compare the crash tests and simulation outputs, neck injury
risk criteria, Nij, Nkm and NIC, were calculated as seen in Table 2-3 and
presented as a percentage of the injury threshold value. It should be noted
that the Nij has only been approved for use with the HybridIII ATD in frontal
crash tests100 and therefore the values shown here may not be especially
meaningful. The Nkm130 and NIC10 are both proposed minor rear crash injury
metrics that can be calculated using outputs from the HybridIII or BioRIDII.
34
INJURY MEASURE TEST MADYMO

FMVSS208 Frontal Neck Injury Criterion (Nij) 8% 9%
Proposed Rear Neck Injury Criterion #1 (Nkm) 21% 28%
Proposed Rear Neck Injury Criterion #2 (NIC-m2/s2) 79% 114%
Table 2-3. Neck injury risk values as a percentage of threshold
These results showed that the MADYMO simulation over predicted neck
injury risk when using these three criteria as compared with the crash tests.
Interestingly in the simulation, the Nij and Nkm were well below their
respective threshold values of 1.0. The simulation NIC exceeded the
proposed threshold of 15.0m2/s2 while the crash tests resulted in a sub-
injurious value.
Finally, head injury risk was investigated for the validated model and
compared with the crash test results. Head injury criterion (HIC), cumulative
brain strain damage measure (CSDM), dilatational brain damage measure
due to vacuum pressure (DDM) and relative brain motion damage measure
(RMDM) were calculated for both the mean crash test results and simulation
results via SIMon v3.051 Beta (SIMulated Injury Monitor, National Highway
Traffic Safety Administration, Washington, D.C., USA). These results are
shown in Table 2-4 as a percent risk of injury for each measure.
INJURY MEASURE TEST MADYMO

Head Injury Criterion (HIC) 1.9% 0.9%
Cumulative Strain Damage Measure (CSDM) 0.0% 0.0%
Dilatational Damage Measure (DDM) 0.0% 0.0%
Relative Motion Damage Measure (RMDM) 3.0% 2.1%
Table 2-4. Head injury risk values as percentage of threshold
All of the head injury results show risk well below threshold or
completely absent. But for the CSDM, DDM and ASDH results, SIMon
35
currently allows calculation of these head injury metrics only for the HybridIII
ATD family, so the calculations for the BioRIDII ATD, while appearing to be
reasonable, should be regarded as preliminary. Though calculating head
injury risk might seem trivial and unnecessary considering the minor nature
of these crashes, occupant head injury potential in minor rear crashes is
often a topic of debate in real world crashes. Therefore, it was interesting to
note that head injury risk in both the crash tests and MADYMO simulation
was essentially nonexistent.
(2) COMPARISON WITH HYBRIDIII – After validating the MADYMO
BioRIDII model, a HybridIII 50th ATD was positioned in the passenger seat
alongside the BioRIDII driver. BioRIDII and HybridIII pelvis, head and
thorax x-axis acceleration-time curves were compared for the 225
millisecond crash pulse time period and are shown in Figures 2-7, 2-8 and 2-
9, respectively. Based on the results from Figure 2-7, the HybridIII
maximum pelvis x-acceleration rose at approximately the same rate as the
BioRIDII but then started deviating near the peak value. The maximum
HybridIII pelvis x-acceleration was 6.6g as compared with 5.4g for the
BioRIDII. After reaching the peak, the HybridIII pelvis acceleration
decreased more rapidly than the BioRIDII. In examining the head x-
acceleration for the HybridIII and BioRIDII it can be seen in Figure 8 that the
HybridIII head acceleration onset was faster by six milliseconds and followed
approximately the same slope to the maximum value. The HybridIII peak
was 14.8g versus 11.4g for the BioRIDII. During deceleration, the HybridIII
declined at approximately the same rate as the BioRIDII. The thorax x-axis
36
acceleration showed the largest discrepancy between HybridIII and BioRIDII
MADYMO responses. The HybridIII had a similar peak value of 4.5g versus
4.7g for the BioRIDII but these peaks were separated by fifty milliseconds.
Additionally, the HybridIII failed to show a significant secondary peak near
160 milliseconds as displayed by the BioRIDII.
Figure 2-7. Pelvis x-axis acceleration
Figure 2-8. Head x-axis acceleration
37
Figure 2-9. Thorax x-axis acceleration
The differences in the BioRIDII and HybridIII ATD crash test response
have been well documented. These differences can mainly be attributed to
spine anthropometry, since the BioRIDII has a full spine with articulating
segments and the HybridIII has simplified spine segments only between the
thorax and skull base. This can be seen in Figure 2-10.
Figure 2-10. (a) HybridIII and (b) BioRIDII [Denton ATD]
38
(3) REDUCED INPUT PULSE – The crash test pulse was reduced by
40% of its original value for the 225 millisecond crash pulse time period in
order to represent a lower magnitude minor crash delta-V, 7.3 km/h,
approximating the mean crash from the Chapter 1 real world data set.
BioRIDII kinetics were quantified in response to this reduced pulse and are
presented in Table 2-5 as a percent change from the original BioRIDII
MADYMO output.
PARAMETER BioRIDII scaled pulse % change

Pelvis acceleration-x (g) 3.3 -39%
Pelvis acceleration-z (g) -1.7 -11%
Thorax acceleration-x (g) 1.9 -60%
Thorax acceleration-z (g) -0.5 -81%
Head acceleration-x (g) 4.2 -63%
Head acceleration-z (g) 0.8 -20%
Head angular accel.-y (rad/s2) -299 -59%
Head angular velocity-y (rad/s) 4.0 -32%
Thorax angular velocity-y (rad/s) 3.2 -49%
Upper neck force-x (N) 17.1 -62%
Upper neck force-z (N) 55.5 -69%
Upper neck moment-y (N-m) 3.4 -70%
Table 2-5. BioRIDII output for pulse scaled by 40% and change from original response
When viewing the reduced pulse simulation, the BioRIDII was seen to
have reduced motion which directly resulted in lower kinetic and kinematic
outputs for all parameters. Injury criteria were not calculated for the
reduced pulse since they would have all been reduced from their already
minimal levels. The average percent change for the twelve parameters
presented in Table 2-5 was -51%, with the largest reductions in important
parameters involving upper neck loading and head accelerations. The
reductions in vertical accelerations were less important since these values
were already fairly small in the original MADYMO simulation.
39
DISCUSSION
The generalized minor rear crash MADYMO simulation with BioRIDII
ATD has been shown to be valid when compared with six crash tests
involving passenger vehicles. While the model did not exactly fall within all
specified acceleration corridors or precisely match all kinetic and kinematic
parameters of the crash tests, the model closely estimated the crash test
response. It was also seen that longitudinal (x-axis) BioRIDII accelerations
were several times larger than those in the vertical (z-axis) direction.
The BioRIDII MADYMO response could have been affected by the lack
of an available vertical acceleration (z-axis) pulse in addition to the
longitudinal (x-axis) pulse. In the rear crash tests, the vehicle sustained
both longitudinal and vertical acceleration. And while it was obvious that
longitudinal vehicle acceleration was the most pertinent to this study,
incorporating a vertical acceleration pulse into the simulation would have
likely provided results that compared even more favorably with the real world
crashes. However, as the vertical acceleration was one-fourth the magnitude
of the horizontal acceleration, the contribution to occupant loading by the
vertical acceleration was minimal.
In the individual crash tests, BioRIDII response was sensitive to seat
parameters such as cushion stiffness, head restraint positioning or seat
frame configuration. This sensitivity was quantified by Siegmund et al.
(2005) by calculating coefficients of variance for the crash test BioRIDII
response parameters of 14% to 52%. By using passenger vehicle rear crash
tests with similar delta-V, and by creating BioRIDII response corridors for
40
these tests, the current study developed a generic MADYMO rear crash model
that would be less reliant on these variable seat parameters. While
differences would exist between individual seats and the MADYMO seat
developed in this study, this simplification was necessary in order to develop
a model that could estimate occupant kinetics and kinematics in response to
a particular crash pulse, rather than any one of the hundreds of different
passenger seats in use today.
In spite of any differences between the crash test and MADYMO
BioRIDII responses, head and neck injury risk proved to be well below any
accepted, or proposed, tolerance values. Except for the case of the proposed
NIC rear crash neck injury risk, which is still a topic of debate, the injury risk
values for the minor rear crash tests and MADYMO model were expected to
be minimal and this expectation was confirmed. It is noteworthy that Nkm
and NIC injury prediction accuracy have been questioned45 and more
research is needed to fully validate these criteria.
The MADYMO HybridIII response was shown to be different than the
BioRIDII response to the same input crash pulse. While this finding was not
itself very surprising, what was interesting to note was that the HybridIII
loading was shown to be higher than the BioRIDII as illustrated in the
increased head and pelvis accelerations. The differences in spinal
anthropometry between the two ATDs would account for much of this loading
difference. Therefore, it was important to understand that MADYMO minor
rear crash simulations that utilize the HybridIII ATD overestimate occupant
loading when compared with a more biofidelic ATD such as the BioRIDII.
41
In response to the reduced input crash pulse, the BioRIDII MADYMO
model predicted reduced accelerations, velocities, forces and moments for all
parameters studied. There was a fair amount of variance in the reduction
from parameter to parameter, but the decreased BioRIDII motion and
lessened head restraint contact were the main reasons for reductions in
head, upper neck and thorax loading parameters. Head and neck injury risk
was not calculated for the reduced pulse simulations as these risks would
have been trivial since they had to be lower than the aforementioned
minimal injury risk in response to the original pulse.
LIMITATIONS
This study had many limitations. The main limitation was a result of
the many variables that were parameterized in MADYMO. Because there
were literally millions of combinations available of seatback angle, cushion
stiffness, floor mat friction, etc. the solution presented here is validated for
only one set of these parameters. There would likely be many other possible
solutions to this generalized model that might produce slightly different
BioRIDII and Hybrid III kinematics. But since the occupant loading was
shown to be of such small magnitude, it is highly likely that any other
solution would result in similar occupant kinematics and kinetics. Another
major deficit to the current generalized model is that the validation stability
was not quantified. Thus, it was unknown exactly how stable the solution
was although anecdotally the model appeared to converge toward a stable
solution after approximately 100 iterations. The last 42 iterations involved
fine-tuning the head-restraint timing and had little influence on pelvis, thorax
42
or lower extremity kinematics or kinetics. Finally, the MADYMO model suffers
from the fact that it is a theoretical solution to a real-world crash. Ideally,
crash tests should be performed to validate the Hybrid III theoretical
performance alongside the BioRIDII. Because time and resources did not
permit for full-scale vehicle crash testing, the Hybrid III results should be
treated as theoretical until this validation can occur.
CONCLUSIONS
As discussed in Chapter 1, minor rear crashes, where struck vehicle
speed change, or delta-V, is less than 15 km/h can be a source of injury
complaints, costly repairs and litigation. In this Chapter, accurate estimates
occupant kinetics and kinematics in these minor crashes were quantified via
a generalized passenger vehicle minor rear crash MADYMO model with a
BioRIDII ATD. This numerical simulation approach was used over cost-
prohibitive crash testing and results validated based on response from six (6)
previously performed minor rear crash tests with mean 12.1 km/h delta-V.
In the crash tests and validated MADYMO model, head and neck
loading were minimal and sufficient to cause only minimal AIS1 level injury
on par with pain or sprain/strain complaints. When comparing HybridIII and
BioRIDII MADYMO responses, the HybridIII predicted slightly higher occupant
loads due to a stiffer spine structure. Thus, minor rear crash MADYMO
simulations or crash tests that use a HybridIII ATD would overpredict
occupant head and spine loading when compared with the more biofidelic
BioRIDII for the same crash pulse. When the crash pulse was reduced by
40% to represent a common real world minor rear crash delta-V of 7.3 km/h,
43
BioRIDII loading was reduced by 11% to 81% for all twelve parameters
studied. This resulted in reduced head and neck injury risk from the already
minimal risk found in the BioRIDII response to the original pulse. This
MADYMO model could be used to provide accurate estimates of occupant
head and spine kinetics and kinematics for passenger vehicle minor rear
crashes up to a delta-V of at least 12.1 km/h in lieu of expensive crash
testing.
44
CHAPTER 3 – NECK INJURY RISK CLASSIFICATION BASED ON ORIGINAL Nij DATASETS
3 CHAPTER 3 – NECK INJURY RISK CLASSIFICATION BASED ON
ORIGINAL Nij DATASETS
INTRODUCTION
As seen in Chapters 1 and 2, minor motor vehicle crashes are a
significant cause of economic and healthcare costs in the United States. And
it was shown that consideration of multiple variables regarding head and
spine injury prediction in these relatively low energy crashes may be of
benefit. In this Chapter, high energy frontal crashes are explored in a
related fashion via re-examination of the crash database utilized to develop
the first neck injury criterion, also known as Nij. While the Nij serves only to
quantify occipito-cervical injury risk due to airbag loading, it provides a
starting point from which to define the injury risk boundary conditions to the
head-neck complex in high energy automotive, military and athletic
applications. Hence, the multivariate analysis methods first examined in
Chapters 1 and 2 for low energy rear crashes are expanded upon here with
more theoretically-based graphical and empirical neck injury risk criteria for
the head and neck being loaded in tension and extension.
BACKGROUND
In an effort to reduce fatal spine and head injuries due to high energy
airbag loading in frontal crashes, the upper spine injury criterion, also known
as Nij, was incorporated into US Department of Transportation Federal Motor
Vehicle Safety Standard 208 crash testing in 2000.100 Nusholtz et al. (2003)
immediately raised concerns about using the nondimensional Nij force-
45
moment combination.113 They found axial tension was a better injury
predictor than Nij and posited that Nij injury prediction inaccuracies arose
from the use of prismatic beam bending theory, which treated the cervical
spine like a homogenous, rectangular prism.113 While force-moment beam
relationships have been validated when analyzing tibia injury risk,100 this
application of beam theory to the spine, with its non-homogeneous
composition of joints, bones, ligaments, tendons and muscles has not been
validated.113 And several authors have shown that the Hybrid III ATD neck
construction contributes to unrealistic upper cervical spine bending moments
during impact testing,1;40;74;113 the use of which in the Nij can lead to
erroneous predictions of spine and possibly head injury risk.
Taking cues from the prior work of Nusholtz et al.,113 and recognizing
that axial tensile cervical force could be quantified in a variety of ways, we
focused our attention on developing force-based spine injury metrics in this
study. Through retrospective examination of the original Nij piglet-ATD
database we developed injury metrics dependent upon normalized tensile
force, tensile force loading rate, tensile force impulse, tensile force ‘binning’
and AIS2 injury severity. This method of re-examining injury data to derive
new risk measures has precedent, as it has been done previously for head,
thoracic and spine trauma.35;113 It was hypothesized that potentially
relevant, accurate and useful spine injury risk metrics existed within the
original piglet-ATD cervical spine tensile force data. Spine injury prediction
based on these new tensile force-based methods was examined to evaluate if
these methods could accurately discriminate between injured and uninjured
46
tests. The Nij,100 as well as peak axial spine tension113 from the paired piglet-
ATD tests, served as benchmarks for these new spine injury metrics.
NIJ DEVELOPMENT
The Nij was developed from a series of fifty-eight (58) paired airbag
impacts with ten-week-old piglets and an ATD representing a 3-year-old child
performed in the 1980’s.94;127 Each of these paired tests involved an upright
anesthetized ten-week-old piglet being impacted by an airbag ventrally to
the lower, middle or upper thorax. The airbag was activated in response to
crash test sled pulses ranging from 4.2g to 14.9g and 28.6 km/h to 57.0
km/h, respectively. Airbag type, cover, folding, pressurization and actuation
time was varied during the testing. Piglet injury was rated according to
AIS1980.2 While there were varying injury severities observed, ranging from
AIS0 “no injury” to AIS5 “critical injury”, for comparison the piglets were
divided into a ‘no/low injury’ group (AIS<3) and an ‘injured’ group (AIS3+).
Following the piglet tests, the representative three-year-old child ATD was
impacted to the anterior cervical spine, jaw and upper thorax under similar
initial seating, crash pulse and airbag firing conditions. ATD upper cervical
spine axial force, FZ, and sagittal plane bending moment about the occipital
condyles, MY, were two of the measured outputs. Finally, piglet injury scores
were compared to forces and moments measured from each paired test.
From these paired tests, Nij, a non-dimensional force-moment
combination was developed to relate upper cervical spine injury risk and ATD
loading in four possible loading regimes as shown in Eq. 3-1.40
47
FZ M
N ij   Y  1.0 Eq. 3-1
F Zc MYc
‘i’ = tension or compression; ‘j’ = extension or flexion; FZ = measured axial force; MY =

measured sagittal bending moment. [For the three-year-old ATD, critical axial force, FZc, is
2120N in compression or tension, while the critical moment, MYc, is 68N-m in flexion and 27N-
m in extension, respectively.]
STUDY METHODS
Of the 58 total paired piglet-ATD airbag tests conducted in the Mertz
et al. (1982) and Prasad and Daniel (1984) studies, we used the thirty-four
tests with continuous axial cervical spine force data with minimum 100
millisecond duration. The Nij results for these ‘no/low injury’ (AIS<3, n=20)
and ‘injured’ (AIS3+, n=14) tests used in this study are shown in Figure 3-1.
The kite-shaped boundary defines theoretical critical loading limits for Nij
over the duration of the test for the three-year-old ATD. The axial cervical
spine force-time histories from the same ‘no/low injury’ and ‘injured’ tests
are shown in Figure 3-2.
Figure 3-1. Nij v. time for ‘no/low injury’ (A) and ‘injured’ (B) piglet-ATD tests
48
Figure 3-2. ATD axial force (Fz) for paired tests with ‘no/low injury’ (A) and ‘injured’ (B)
piglets
From these 34 piglet-ATD axial spine force data sets, the following
tensile force-based injury risk metrics were developed:
Risk Parameter 1: Maximum Normalized Tensile Force
For each test, the axial tensile force was normalized against the
maximal allowable tension of 1130N for the three-year-old ATD.93
Normalized compression force was not considered because the injury
mechanisms in the piglet tests were all tensile in nature.94;127 The maximum
normalized tension force for each test was calculated as shown in Eq. 3-2.
 F 
100 
Z
Maximum Normalized Tensile Force  max    Eq. 3-2
 1130N  0 
 
Risk Parameter 2: Maximum Tensile Loading Rate
Maximum tensile loading rate was hypothesized to be an important
factor related to spine injury risk because the spine is comprised of both
viscous and elastic constituents.53 This hypothesis was reinforced by the fact
that loading rate has been used as an injury prediction tool in cadaveric
49
impact testing.129 Thus, for each paired piglet-ATD test the maximum tensile
loading rate over the first 100 milliseconds was calculated using Eq. 3-3.
 dF 100 
Maximum Tensile Loading Rate  max  Z   Eq. 3-3
 dt  0 
 
Risk Parameter 3: Cumulative Tensile Impulse
It was theorized that cumulative tensile impulse would provide a
reasonable risk measure since crash-related injury has been long known to
depend upon the force accumulated over the impact duration and the piglet
cervical spine failure mechanisms were strictly tensile in nature.46;94;127
Furthermore, analyzing impulse during the first 100 milliseconds avoided
impulse accumulation over longer time durations that has been previously
shown to diminish prediction accuracy in head injury analysis.46 The
cumulative tensile impulse was analyzed over the 100 millisecond time
period as described by Eq. 3-4.
100
Cumulative Tensile Impulse 
F
0
Z dt Eq. 3-4
Risk Parameter 4: Tensile Force ‘Binning’
Tensile force ‘bins’ were created to compare the cumulative percent
time spent at each loading level. The normalized tensile force data were
discretized in increments of 0.10, up to the maximum normalized force, for
each test during the 100 millisecond time period. The time spent in each
discretized bin increment was summed and the percent time spent in each
bin for the 100 millisecond time period was computed. An example of force
50
binning can be seen in Figure 3-3, where the time spent in each bin is
represented by the portion of the pulse dwelling in each bin. In our study,
Bin0.9, with the total percent time spent above the bin with normalized force
equal to 0.9, was identified as the bin level showing the best delineation
between ‘injured’ and ‘no/low injury’ tests.
Figure 3-3. Binning method applied to normalized force data from Piglet Test #6368
Risk Parameter 5: Injury Severity
The piglets sustained a variety of injuries that were documented
during autopsy. As mentioned previously, each of these injuries were rated
according to AIS19802. This scale ranked injuries from zero to 6 based on
injury severity and threat-to-life. Table 3-1 illustrates the complete AIS
scoring system. Considering injury severity via AIS allowed for a dramatic
51
visual depiction of the various risk measures. When the simplified binary
system of ‘injured’ (AIS3+) or ‘no/low injury’ (AIS<3) was considered,
statistical methods were used to compare the two groups.
Abbreviated Injury Scale

(AIS)
AIS Injury Nij
Score Severity Designation
0 None
'no/low
1 Minor
injury'
2 Moderate
3 Serious
4 Severe
'injured'
5 Critical
6 Maximal
Table 3-1. Abbreviated Injury Scale (AIS)
STATISTICAL AND VISUAL ANALYSIS METHODS
A combination of statistical and subjective visual analysis methods
were used to examine the piglet-ATD data and determine which method(s)
best separated ‘injured’ and ‘no low injury’ tests based on the
aforementioned risk parameters. The subjective visual methods incorporated
AIS while the statistical methods grouped injury according to the ‘no/low
injury’ (AIS<3) and ‘injured’ (AIS3+) and combinations of the injury metric
parameters. Three types of graphical results were produced: (i) logistic
regression plots to compare the doubly censored tests and injury risk
threshold, (ii) three-parameter bubble scatterplot with failure envelope
determined by regression risk thresholds and (iii) four-parameter spheroidal
scatterplot with failure volume determined by regression risk thresholds.
52
(i) Binary Logistic Regression
For each group, a Kolmogorov-Smirnov normality test was first
conducted to verify that the data could be treated as a normal distribution.
The ‘no/low injury’ group D-statistic was 0.2181 and the critical value at
=0.05 was 0.2941. The ‘injured’ group D-statistic was 0.2054 and the
critical value was 0.3489. Based on the normality test, both groups were
treated as normal distributions. For each risk parameter, the binary logistic
regression fit was calculated based on binary ‘injured’ (AIS3+) and ‘no/low
injury’ (AIS<3) data. The marker radius is shown as a function of AIS
severity. Injury risk equal to 40% was used as a threshold for all parameters
because it resided between the injury risk values used for head injury and
neck injury analyses.40;63;93 Figure 3-4 shows the regression curves for the
four (4) injury risk parameters.
53
Figure 3-4. Logistic regression curves for four parameters: (a) Maximum Normalized Load,
(b) Maximum Loading Rate, (c) Cumulative Impulse, (d) %Time Above Bin=0.9. Intercepts
drawn at 40% injury risk
(ii) Bubble Scatterplot
The bubble scatterplot allowed examination of three-dimensional data
in a two-dimensional space. Data were plotted for the six (6) combinations
of two risk parameters with the marker radius defined by the AIS magnitude
as shown in Figure 3-5. Theoretical critical intercepts on the ordinate and
abscissa were created based on the aforementioned 40% injury risk
threshold from each parameter regression curve. While AIS magnitude
defined the marker radius, injury classification accuracy was solely based on
the location of the marker center.
54
Figure 3-5. Three parameter bubble scatterplot for six (6) injury parameter combinations:
(a) Maximum Loading Rate vs. Maximum Normalized Force, (b) Cumulative Impulse vs.
Maximum Normalized Force, (c) Cumulative Impulse vs. Loading Rate, (d) Maximum Loading
Rate
(iii) Spheroidal Scatterplot
The four parameter spheroidal scatterplot utilized combinations of
three risk parameters plotted in three-dimensional space. AIS defined the
radius of the resulting spherical markers but again, the injury classification
accuracy depended on marker center location, and not marker radius. The
theoretical risk volume was an expansion of the two-parameter failure
envelope into three-dimensional space with intercepts defined by the
respective 40% injury risk threshold for each parameter.
55
Figure 3-6. Four parameter spheroidal scatterplot of four (4) injury parameter combinations:
(a) Maximum Normalized Force vs. %Bin Time vs. Cumulative Impulse, (b) Maximum
Normalized Force vs. %Bin Time vs. Maximum Loading Rate, (c) Maximum Normalized Force
vs. Cumulative Impulse vs. Maximum Loading Rate, (d) %Bin Time vs. Cumulative Impulse
vs. Maximum Force Loading Rate. Volume intercepts drawn at 40% injury risk, sphere radius
equals AIS injury severity.
RESULTS
The results for each injury risk metric are summarized in Table 3-2
and compared with Nij and peak axial tension.113 Nij and peak axial tension
correctly classified ‘no/low injury’ tests at 80% and 88%, respectively. For
‘injured’ tests, Nij and peak axial tension classification accuracy was 65% and
80%, respectively.
56
% Correctly %
Injury Risk classified Correctly
Risk Measure
Method 'no/low classified
injury' 'injured'
Nij (2000) Force, Moment 80% 65%

Nusholtz et al.
Tensile Force 88% 80%
(2003)
Logistic Normalized Tensile Force 95% 100%
Regression-40% Tensile Force Binning 95% 100%
Threshold Cumulative Tensile Impulse 95% 86%
(Figure 4)
Tensile Loading Rate 90% 79%
Normalized Tensile Force vs.
95% 100%
Tensile Force Binning
95% 100%
Cumulative Tensile Impulse
Three Parameter Cumulative Tensile Impulse
95% 100%
Bubble vs. Tensile Force Binning
Scatterplot Tensile Loading Rate vs.
(Figure 5) 90% 100%
95% 93%
Tensile Loading Rate
Tensile Loading Rate vs.
90% 93%
Tensile Force Binning vs. 95% 100%
Tensile Loading Rate vs.
Four Parameter Tensile Force Binning vs. 90% 100%
Spheroidal Cumulative Tensile Impulse
Scatterplot Normalized Tensile Force vs.
(Figure 6) Tensile Loading Rate vs. 90% 100%
Tensile Loading Rate vs. 90% 100%
Table 3-2. Classification comparison for Nij (2000), Nusholtz et al. (2003) and current study
Logistic Regression
As seen in Figure 3-4, the regression curves for the four parameters
show discrimination between ‘injured’ and ‘no/low injury’ data at a 40% risk
of injury. As seen in Table 3-2, the best ‘no/low injury’ accuracy was 100%
for both normalized force and binning. The highest ‘injured’ prediction
accuracy was 95% for normalized force, binning and cumulative impulse.
57
The AIS marker radius in each plot appeared fairly chaotic in nature, with no
clear cut trend for AIS severity and parameter magnitude other than
separating between AIS3+ and AIS<3. The respective 40% threshold
intercepts are shown in Table 3-3.
40% Intercepts
Normalized Force (N/N) 1.10
Maximum Loading Rate (N/s) 1.10E+05
Cumulative Impulse (N-s) 30.1
%Bin Time above 0.9 9.85%
Table 3-3. 40% injury risk intercepts for the four parameters based on regression curves
Bubble Scatterplot
The six possible parametric combinations and resulting theoretical 2-D
40% failure envelope is shown for the bubble scatterplots in Figure 3-5.
Based on the results in Table 3-2 and considering the 40% failure envelopes,
the most accurate correct classification rate was 95% for ‘no/low injury’ tests
combinations of normalized force vs. binning, normalized force vs.
cumulative impulse, normalized force vs. maximum loading rate and
cumulative impulse vs. binning respectively. For the ‘injured’ tests,
combinations of normalized force vs. binning, normalized force vs.
cumulative impulse, cumulative impulse vs. binning and maximum loading
rate vs. binning, correctly classified 100% of tests, respectively. Again, the
AIS marker radius in each plot appeared to be unrelated to parameter
magnitude.
58
Spheroidal Scatterplot
Of the four parametric combinations seen in Figure 3-6 spheroidal
scatterplots, the normalized force vs. binning vs. cumulative impulse
combination classified ‘no/low injury’ tests based on the theoretical 3-D 40%
injury risk volume at a 95% accuracy rate. For ‘injured’ tests, all four
scatterplot combinations classified at the maximum rate of 100%. And as
mentioned in the regression and bubble plots, the AIS marker radius
appeared unrelated to injury parameter magnitude.
DISCUSSION
The novel spine injury risk methods successfully discriminated
between AIS<3 ‘no/low injury’ tests and AIS3+ ‘injured’ tests at rates as
high as 100%. The regression analyses showed that at a hypothetical 40%
injury risk threshold, the four risk metrics correctly classified ‘no/low injury’
tests at 90% to 95% and ‘injured’ tests at 79% to 100%, respectively. The
bubble scatterplots had classification accuracy between 90% and 95% for
‘no/low injury’ tests and for the ‘injured’ tests, classification accuracy ranged
from 93% to 100%. The spheroidal scatterplots produced classification
accuracy of 90% to 95% for ‘no/low injury’ tests and 100% for all ‘injured’
tests for marker centroids residing within the 3-D failure volume. Normalized
force, cumulative impulse and binning classified ‘no/low injury’ and ‘injured’
tests at the highest rates while maximum loading rate had the lowest
prediction accuracy across all three analysis methods.
59
All injury risk metrics presented correctly classified ‘injured’ tests at a
rate higher than the 65% rate for Nij and thirteen of fourteen metrics
classified ‘injured’ tests at a rate higher than 80% for maximum axial
tension. For ‘no/low injury’ tests, all risk measures classified injury at a
higher rate than the 80% seen in Nij or the 88% rate based on axial force
magnitude alone. For all analysis methods, an obvious relationship between
AIS injury severity, as designated by marker radius, and parameter
magnitude was not evident.
LIMITATIONS
Although the various spine injury prediction methods presented here
showed promise for accurately separating injured and uninjured test
subjects, this study had several limitations. The main limitation was that the
paired piglet-ATD tests may not have been representative of the human
injury response to frontal airbag deployment. Since the Nij frontal impact
tests were conducted to compare a representative three-year-old ATD with
the piglet surrogate, young human cadavers simply were not available to
conduct this type of testing. Nonetheless, the piglets had an internal and
external anthropometry vastly different from humans and in these tests the
piglets likely had a different impact response than human cadavers.
Numerous cervical spine injury impact tests have been done with human
cadavers since the Nij piglet-ATD frontal airbag tests in the early
1980’s.19;28;33;107-112;126;147;160;173;174;177 Thus, the criteria presented here likely
require further rigorous validation with these or future human cadaver tests.
Further, the original datasets were hand-digitized at 1000Hz and potential
60
aliasing effects were not quantifiable. The 40% risk threshold was chosen
here to optimize the injury classification accuracy, and although it is in the
range of similar risk thresholds for HIC and Nij, the exact threshold to use
requires further numerical justification based on human testing. Another
limitation in the current study was that in the paired piglet-ATD tests the
chaotic nature of airbag impacts resulted in intratest ATD responses that
were vastly different. In test series where ATD impacts were repeated after
a single piglet impact of the same magnitude, the peak axial tensile force
varied by up to 2060 N and the peak sagittal moment varied by up to 56.5
N-m, respectively.94;127 This intratest variability may have been one major
factor contributing to the reduced injury prediction accuracy by Nij and
provided false accuracy for the various criteria presented here. An
additional limitation was that the test sample sizes used here, n=14 ‘injured’
and n=20 ‘no/low injury’, were fairly small. It would be desirable to have a
larger sample size for both injury classifications in order to fully validate the
novel injury measures proposed here. A final limitation was that while
sagittal spine moment was not considered in the spine injury criteria
development presented here, there could be value in considering bending
moment for spine injury risk. While Nusholtz et al. (2003) have proposed
that the entire cervical spine cannot be treated as a prismatic beam as was
done with Nij, bending moment has been shown to be a reliable indicator of
intervertebral segment injury and thus may serve a crucial role in segmental
spine injury prediction.53
61
Deserving special note in our study is that the piglet-ATD Nij tests94;127
involved head-thorax airbag impacts that resulted many severe piglet brain
injuries in addition to spine injuries. In these tests, the ‘injured’ piglets
sustained mean head resultant linear acceleration, Head Injury Criterion
(HIC)63 and angular acceleration of 285g, 9232 and 10157rad/s2,
respectively. The ‘no/low injury’ piglets experienced mean head resultant
linear acceleration, HIC and angular acceleration of 149g, 2020 and
4717rad/s2, respectively. Drawing on the practice of neurosurgery as a
compass, wherein head-spine injury is treated on a linked system basis as
opposed to either ‘head’ or ‘spine’ components individually, it could have
been possible that spine injury risk in these tests was a function of head
injury risk and vice versa. Particularly interesting is that piglets sustained
brain injuries due to both linear and rotational motion but never sustained
skull fracture. Thus, while the simultaneous combination of linear and
rotational kinetics was shown to adversely affect injury prediction accuracy
via Nij, consideration of head linear and rotational kinematics might be
essential to predict injury. And the effect of combined head linear and
rotational motion on cervical spine dynamics is not readily obtainable from
the Nij datasets. Therefore, while we did not specifically consider a direct
linkage between head injury risk and spine injury risk in this study, the
question of linked head-spine injury response is a tantalizing topic that
merits further investigation.
62
CONCLUSIONS
Spine injuries continue to be a major concern for the US motoring
public, accounting for 37% of the most critical motor vehicle injuries seen in
US trauma wards. And while the frontal crash Nij has been part of US crash
testing regulations since 2000, as first posited by Nusholtz et al. (2003), its
efficacy in accurately predicting safe spine loads in motor vehicle crashes
remains questionable. Furthermore, other spine injury risk metrics are
currently using Nij as a basis for injury prediction in rear automobile crashes,
frontal train crashes, lateral airplane fuselage impacts, boating and military
protective equipment testing.27;42;56;130;157 Because of the diminished injury
prediction accuracy in the Nij, these other risk metrics may suffer in their
ability to discriminate between safe and unsafe loading levels in the human
spine.
To this end, a total of fourteen (14) novel spine injury risk graphical
and statistical metrics were developed via re-examination of the original
paired piglet-ATD Nij test database and evaluated against Nij and peak
tension force prediction accuracy. The risk metrics were based on cervical
spine tensile force as a function of time and included combinations of
normalized force, loading rate, impulse, force ‘binning’ and injury severity.
The injury risk metrics were found to correctly classify ‘injured’ tests up to
100% accuracy versus 68% for the Nij, and uninjured subjects up to 95%
accuracy versus 78% for the Nij, respectively. These new and accurate spine
injury risk metrics may prove useful in the future development of spine injury
criteria for non-frontal crash vectors where the predominant failure mode is
63
axial spine tension. Additionally, aerospace, railway, boating or military
safety applications may find it advantageous to use this type of parametric
analysis in the design of spine injury metrics.
64
CHAPTER 4 – ATD HEAD IMPACTS AND HEAD-NECK DYNAMICS
4 CHAPTER 4 – ATD IMPACTS AND HEAD-NECK DYNAMICS
INTRODUCTION
In this Chapter, head and spine injury analysis transitions from low
energy (Chapters 1 and 2) and high energy (Chapter 3) automotive-related
studies into examining high energy athletics-related injury. As seen in
Chapters 2 and 3, in vitro study of high energy head and spine injury is best
done with anthropomorphic surrogates. But the influence of the Hybrid III
crash test dummy intrinsic head-neck biofidelity in response to direct, high-
energy athletic impacts has not been quantified. Therefore, in this Chapter
in vitro experiments are performed on an isolated Hybrid III crash test
dummy in order to quantify its frontal, oblique and lateral response to direct
high-energy impact. The knowledge gained from these quantification
experiments lays the foundation for the analyses of padded and unpadded
multidirectional athletic head impacts presented in Chapter 5.
BACKGROUND
Concussive traumatic brain injury and cervical spine injury continue to
pose threats to athletes engaged in contact sports. In this study the Hybrid
III 50th percentile male anthropomorphic test device (ATD) was used to
quantify head-cervical spine response to in vitro head impact loading such as
might occur in concussive athletic impacts. The Hybrid III ATD was
originally developed and validated for US Federal Motor Vehicle Safety
Standard (FMVSS) 208 frontal crash testing.100 The ATD head-cervical spine
kinetics and kinematic response was developed and validated in the sagittal
65
plane using the pioneering studies of Mertz and Patrick.91;92 In these
studies, human volunteers and cadavers were subjected to quasistatic and
dynamic seated flexion and extension tests.
While the Hybrid III ATD was originally validated and is primarily used
in FMVSS 208 crash testing, it has also been widely used in athletic head
impact studies.7;44;84;104;166;168;171 However, in these prior studies the ATD
head-cervical spine biofidelity was not specifically studied and still remains
one potential confounding factor that may influence athletic head impact
injury metrics. Hybrid III occipito-cervical biofidelity quantification is of
particular interest because of the potential that in vitro ATD head impact
response does not accurately reproduce in vivo athlete head-cervical spine
dynamics during concussive impacts. Furthermore, the Hybrid III ATD head-
cervical spine was validated in the sagittal plane without direct head
impact,91;92 possesses non-biofidelic behavior in response to head-cervical
spine impacts,1;74 and the hollow ATD skull and absence of a deformable
brain might influence the inertial loading parameters of the head-cervical
spine complex. Thus, in this study the ATD head-cervical spine response to
direct head impact loading in the frontal, oblique and lateral directions in
response to three (3) increasing impact energy levels was quantified.
METHODS
A Hybrid III 50th Percentile ATD (Denton ATD, Rochester, MI) was
instrumented with a triaxial linear accelerometer (Model EAS3-250
Measurement Specialties, Hampton, VA) and a triaxial angular rate sensor
(Model ARS-06S, ATA Sensors, Albuquerque, NM) mounted at the head
66
center of gravity (cg). A six-channel occipito-cervical load cell (Model 1716A,
Denton ATD, Rochester, MI) measured forces and moments in three
directions. The moments measured by this supracondylar load cell were
translated inferiorly to the occipital condyles by multiplying respective x- or
y-axis shear forces by the 0.01778m moment arm. All data were collected at
5000Hz, filtered according to SAE J211143 and sign convention adhered to
SAE J1733145. Angular acceleration was calculated by differentiating the
angular velocity signal after determining proper signal content and pre-
differentiation filter requirements. As a consequence of the multiple impact
directions and energy levels tested, individual x-, y- and z-axis components
comprise voluminous data sets such that only the resultant dynamic
parameter values will be presented here.
A 3.6kg steel sphere impact pendulum was used in all impacts. This
mass was chosen to deliver similar impacts as the effective mass involved in
padded boxing punches to the head.168 The mass was hung from a 6.4mm
steel braided cable that was free to swing from a ceiling-mounted carabineer
above the ATD head. The ATD was seated in a chair and secured with tie-
down straps. Prior to testing, the lower extremities were removed. In order
to minimize inertial responses below the cervical spine, the test stand was
weighed down with approximately 3500N of sandbags. For each impact trial,
the pendulum was released from the desired swing height by cutting a nylon
string attached to the steel sphere. The swing heights were selected to be
within the range of magnitudes from prior athletic head impact
studies22;37;38;43;57;85;104;105;114;115;122;124;141;148;153;165;178 that indicated higher
67
risk of concussive traumatic brain injury and/or cervical spine injury. In this
study, a total of fifty-four (54) head impact pendulum trials were performed,
in sets of six (6) trials for each impact direction (frontal, oblique, lateral) and
impact energy level (lowest, medium, highest) as shown in Figure 4-1 and
Table 4-1. Medium and highest energy impacts were limited by the available
swing height within the testing space. The frontal and oblique impacts
targeted the maxillar-mandibular junction, while the lateral impact targeted
the temporal region.
Figure 4-1. Frontal, Oblique and Lateral impact setups
ATD Head Impact Energy Impact Momentum

Impact (J) (N-s)
Direction Low Med High Low Med High
Frontal 26.9 53.8 76.2 13.9 19.7 23.4
Oblique 26.9 50.2 88.8 13.9 19.0 25.3
Lateral 26.9 53.8 67.1 13.9 19.7 22.0
Table 4-1. Test impact matrix
RESULTS
The dynamic results are plotted against impact energy and are shown
in Figures 4-2 through 4-10 and Table 4-2, respectively. The same dynamic
data were also compared with impact momentum and showed the same
trends as the comparison with impact energy. Since impact energy was a
68
more meaningful comparative measure, comparisons with impact momentum
were not reported for the sake of brevity.
Figures 4-2 and 4-3 show a representative sample of the kinematic
(head cg linear acceleration) and kinetic (occipital force) resultant values
from each set of high energy impacts.
Figure 4-2. Resultant ATD head center of gravity (cg) linear acceleration for each set of six
(6) impact trials in the three impact directions at highest impact energy
69
Figure 4-3. Resultant ATD occipital force for each set of six (6) impact trials in the
three impact directions at highest impact energy
In addition to examining resultant kinematic and kinetic values,
component-based plots were also produced. Figures 4-4 and 4-5 show
representative plots of x-, y- and z-axis linear acceleration and occipital force
for the mean values from the high energy impact sets.
70
Figure 4-4. Mean ATD head center of gravity (cg) linear acceleration components from each
set of six (6) impact trials in the three impact directions at highest impact energy
71
Figure 4-5. Mean ATD occipital force components from each set of six (6) impact trials
in the three impact directions at highest impact energy
The respective coefficient of variation (COV) from each set of six
impacts for each impact energy and direction was quantified and is displayed
in Table 4-2. This was done to ensure each set of six impact trials was
repeatable and that using the mean value for each dynamic parameter was
justified. The COV was calculated by dividing the standard deviation from
each trial set of six impacts by the trial set mean value. The highest mean
COV was seen in occipital force with 7.7% variation among all trials and all
impact directions. Resultant head cg linear acceleration and resultant head
angular acceleration had the lowest mean COV, 4.6% and 4.7%,
respectively. Because the COV for all dynamic parameters was low, the trials
72
were determined to be very repeatable and use of the mean value from the
impact trials was verified.
PERCENT COEFFICIENT OF VARIATION

Front Front Front Obl Obl Obl Lat Lat Lat
Low Mid High Low Mid High Low Mid High
Head CG
2.9 2.5 3.9 7.4 10.1 4.5 4.7 2.1 3.6
Acceleration
Head Angular
3.5 1.8 4.8 5.2 8.7 4.7 12.8 9.5 5.5
Velocity
Head Angular
2.2 2.7 3.3 3.6 7.5 0.5 10.2 8.4 3.7
Acceleration
Occipital
2.1 2.7 7.1 5.7 8.8 6.1 9.1 5.5 7.3
Moment
Occipital Force 8.3 5.7 5.9 18.4 7.7 8.3 5.0 3.3 6.4
Table 4-2. Coefficient of variation among all impact trial sets (Front = Frontal, Obl = Oblique,
Lat = Lateral)
In Figure 4-6, resultant linear cg acceleration was highest for lateral
impacts, followed by frontal and oblique impacts, respectively. The cg
acceleration generally trended linearly upward, but with three impact energy
levels plotted for each impact location, any nonlinear response was not
obvious from this data.
73
Resultant Head cg Linear Acceleration v. Energy

250
Frontal
Oblique
Lateral
200
Acceleration (g)
150
100
50
0
0 10 20 30 40 50 60 70 80 90 100
Energy(J)
Figure 4-6. Resultant ATD head center of gravity (cg) linear acceleration for the three impact
directions compared with impact energy [NOTE: error bars show +/- one standard deviation]
For the resultant head angular velocity shown in Figure 4-7, at the
lowest energy impact level, all three directions produced similar resultant
angular velocities. These responses remained tightly grouped at the medium
energy level. At the highest energy impact level, frontal impacts had the
highest angular velocity, followed by oblique and lateral impacts,
respectively.
74
Resultant Head Angular Velocity v. Energy

30
Frontal
Oblique
25 Lateral
Angular Velocity (rad/s)
20
15
10
0
0 10 20 30 40 50 60 70 80 90 100
Energy(J)
Figure 4-7. Resultant ATD head angular velocity
The resultant head angular acceleration seen in Figure 4-8 displayed a
similar trend to the angular velocity results. At the lowest energy level, all
three directions had very similar angular acceleration magnitude. Again, a
crossover was seen in the medium impact energy level as lateral magnitude
dropped below the oblique magnitude. At the highest impact energy level,
frontal angular acceleration was the largest, followed by oblique and lateral
impacts.
75
Resultant Head Angular Acceleration v. Energy

10000
Frontal
9000 Oblique
Lateral
8000
Angular Acceleration (rad/s 2)
7000
6000
5000
4000
3000
2000
1000
0
0 10 20 30 40 50 60 70 80 90 100
Energy(J)
Figure 4-8. Resultant ATD head angular acceleration
In Figure 4-9, the resultant occipital moment had a clear trend for all
three impact energy levels studied. The frontal impact moment was always
highest, followed by oblique and lateral impacts.
76
Resultant Occipital Moment v. Energy

150
Frontal
Oblique
Lateral
100
Moment (N-m)
50
0
0 10 20 30 40 50 60 70 80 90 100
Energy(J)
Figure 4-9. Resultant ATD occipital bending moment
Finally, in Figure 4-10, the resultant occipital force was compared with
impact energy. The frontal impacts always had the highest occipital force
across all impact energy levels. The oblique and lateral impacts had very
similar results with the largest difference between these two directions
occurring at the medium energy level.
77
Resultant Occipital Force v. Energy
Frontal
2500
Oblique
Lateral
2000
Force (N)
1500
1000
500
0
0 10 20 30 40 50 60 70 80 90 100
Energy(J)
Figure 4-10. Resultant ATD occipital force
DISCUSSION
A key finding in this Chapter was that the data from the ATD impact
tests showed that impact direction played a heavy role in resulting ATD head
and occipito-cervical x-, y- and z- kinematic and kinetic components. This
was an important distinction in that by simply analyzing resultant peak
magnitudes, the directional sensitivity to impact was lost. Hence, while data
showed, for example in Figure 4-7, that lateral and oblique medium energy
impacts had similar angular velocity resultants, the rotational components
loading the head and neck were very different. This finding has implications
beyond ATD testing, and could be useful in analyzing head and neck
dynamics where impact direction, magnitude and duration all must be
78
considered when assessing injury risk; i.e.-simply reporting a resultant peak
magnitude as the only measure of injury risk as is done with HIC or GSI
might omit crucial information regarding x-, y- and z-axis contributions to
the dynamic force-time history.
The results also indicated that ATD head-cervical spine stiffness played
a role in determining the relative magnitudes of the parameters studied. The
resultant ATD cg linear acceleration was always highest for lateral impacts.
This might have been due to the lateral impact location, which was closer to
the head cg as opposed the oblique or frontal impacts. This response may
also be attributed to the ATD head-cervical spine complex being stiffer in the
lateral direction, causing less head motion during impact and allowing for the
linear acceleration to increase. The resultant ATD head angular velocity and
angular acceleration results were very similar. At the lowest impact energy
levels, both of these parameters showed little difference between impact
directions. As the impact energy increased, the frontal angular
velocity/acceleration increased faster than either the oblique or lateral
directions, respectively. It was interesting that the lateral direction showed a
slower increase in these angular responses as energy increased, perhaps also
related to the ATD head-cervical spine being stiffer in the lateral direction
with less motion as opposed to the frontal or oblique directions. When
investigating the resultant occipital moment results, it was likely that ATD
head-cervical spine stiffness again played a significant role as both the
frontal and oblique impacts had a faster increase in occipital moment than
the lateral impacts. In the lateral direction, the smaller bending moment
79
compared with the oblique/frontal impacts might have indicated that the
spine sustains more overall bending as opposed to bending mainly about the
occipito-cervical spine junction. When examining the occipital force results,
the frontal impacts always had the highest resultant force. The oblique and
lateral impacts had similar behavior at the lowest energy while oblique
impacts had a larger force magnitude at the highest energy impact level.
While no prior study known had specifically investigated Hybrid III ATD
head-cervical spine response to direct head impact in absence of axial
compression, some comparisons were made with prior human and ATD
testing. When considering extension alone, dynamic inertial human volunteer
motion studies indicated occipital bending moments up to 42N-m could be
sustained without injury.98 Dynamic cadaveric inertial and directly applied
extension studies showed onset of occipito-cervical AIS2+ ‘moderate’ injury3
to occur between 43N-m and 100N-m.71;108;111;125 For lateral motion
tolerance during dynamic inertial tests, volunteers endured resultant occipital
moments up to 55N-m without injury169 while dynamic inertial cadaveric
specimens have shown AIS2+ occipito-cervical injury at resultant moments
of 247N-m.88 And in torsional loading, volunteers endured inertial occipito-
cervical bending moments up to 20N-m without injury, while dynamic
cadaveric specimens have injured at 17N-m to 23N-m.24;97;169 Quasistatic
specimens tested in pure torsion have endured bending moments up to 14N-
m prior to failure.54 The summary of these extension, lateral and torsional
loading studies is that the occipito-cervical spine appeared to have the
highest injury and/or failure tolerance in the lateral direction, followed by
80
extension and torsional motion, respectively. The ATD lateral rotation axis in
this study was the stiffest, which tended to mimic the lateral motion
tolerance finding, and the ATD had a stiffer response to oblique loading
rather than extension loading, which differed from cadaveric and volunteer
data. However, it was important to note that the Hybrid III ATD head-
cervical spine was validated solely in the sagittal plane for flexion-extension
response, and correct occipito-cervical response was never specifically
desired or studied for this ATD.
In comparing ATD multidirectional response to prior testing,
quasistatic cadaveric spine testing at 1.0N-m or 1.5N-m bending moment
showed the occipito-cervical junction was progressively stiffer in response to
axial torsion, extension and lateral bending, respectively, with the lateral
bending being roughly twice as stiff as the extension bending response.119;120
In quasistatic volunteer direct loading tests, the maximum tolerable occipital-
cervical bending moment was approximately 2-3 times higher in the oblique
or lateral directions when compared with extension.121 Dynamic human
volunteer inertial sled tests showed that lateral tests had lower cervical
bending moments and total rotation when compared with oblique tests.170
Dynamic human volunteer direct bending tests showed the tolerable lateral
cervical spine force to be almost twice that of the tolerable oblique force.121
In a dynamic cadaveric helmeted impact study, cadavers impacting their
frontal bone had occipital accelerations 50% higher than those cadavers
impacting the temporal region.55 All of these studies, with the exception of
Got et al. (1978) indicated the occipito-cervical junction to be stiffest in
81
response to lateral loading, followed by oblique and extension loading,
respectively. These findings were mirrored in the ATD impact data based on
occipital bending moment. When examining the Got et al. (1978) study, the
cadaveric head acceleration behavior was reversed from these results with
the ATD in that in this study frontal head acceleration was approximately
50% of the lateral acceleration. In summary, while the Hybrid III ATD head-
cervical spine junction was created solely as a tool to evaluate sagittal plane
motion and injury risk in frontal crashes, the ATD appeared to have
reasonable biofidelity that mimics loading behavior in most of the prior
literature without applied axial compression.
LIMITATIONS
There were three main drawbacks to this study. First, it should be
noted again that impact location on the ATD head could have played a role in
the response. While frontal and oblique impacts struck the same elevation on
ATD head, the lateral impact struck approximately three inches superior,
closer to the cg plane. This might have affected rotational responses by
generating higher lateral occipital bending moment, head angular velocity
and angular acceleration as well as generating higher linear cg acceleration.
Unfortunately, the ATD could not have been struck in the same transverse
plane as the frontal and oblique impacts because this would have involved
direct cervical spine impact. The second drawback was that the total head-
cervical spine rotational position response could not be quantified. While
prior extension, oblique, torsional and lateral bending studies have
investigated total30;71;91;92;97;121;135;169;170;175 and
82
local16;25;34;54;83;108;111;119;120;125;136 head-cervical spine rotational position, we
could only monitor head rotation via the angular rate sensor. Thus, a true
value of head-cervical spine stiffness, with comparison between force or
moment and segmental rotation, was not possible. Finally, the ATD was
impacted numerous times over the course of a single day without a
predetermined rest time between impacts. While the effects of conducting
these impacts in a high-throughput manner versus over the course of many
days would likely be very small, nonetheless effects of repeated impacts
were not quantified. In future testing it would be necessary to address these
limitations.
CONCLUSIONS
Though well-used in low- and high-energy automotive crash
experiments, the in vitro head-cervical spine response of the Hybrid III 50th
Percentile Male Anthropomorphic Test Device (ATD) had not been
conclusively determined in response to athletic type head impacts in absence
of applied axial compression. Through a series of frontal, oblique and lateral
pendulum tests at impact energy ranging from 27J to 89J and momentum
ranging from 14N-s to 25N-s, it was determined Hybrid III ATD head-cervical
spine response to direct head impacts in absence of applied axial
compression. For all impacts, Hybrid III ATD resultant head center of gravity
linear acceleration was 52g to 247g, resultant head angular velocity was
14.0rad/s to 26.4rad/s, resultant head angular acceleration was 4083rad/s2
to 9715rad/s2, resultant occipital condyle force was 778N to 2444N and
resultant bending occipital moment was 20.6N-m to 111.3N-m, respectively.
83
When comparing head impacts at similar energy levels but different impact
vectors, ATD resultant head acceleration was highest in lateral impacts,
followed by frontal and oblique impacts, respectively. At the highest impact
energy levels, frontal resultant head angular velocity and angular
acceleration were the highest, followed by oblique and lateral impacts,
respectively. At lower impact energy levels, angular velocity and angular
acceleration had variable behavior dependent upon impact direction. Across
all impact energy levels ATD occipito-cervical bending moment was highest in
frontal impacts, followed by oblique and lateral impacts, respectively.
Resultant ATD occipital force was similar in the lateral and oblique impact
directions but highest in response to frontal impacts across all impact
energies.
Impacts producing highest ATD resultant cg linear acceleration
resulted in the lowest resultant occipito-cervical spine bending
moment/force, resultant ATD head angular velocity and angular acceleration
did not appear coupled to impact direction at lower impact energy levels
while these parameters were coupled to impact direction at higher energy
levels and when comparing the current ATD loading behavior to the majority
of prior cadaveric and volunteer studies, the trend of progressively increasing
occipito-cervical stiffness in response to extension, torsion and lateral
bending was reproduced in the Hybrid III ATD.
These were the first known data to specifically study Hybrid III ATD
head-cervical spine response to multi-directional direct head impact loading
without applied axial compression. This new understanding of Hybrid III ATD
84
head-cervical spine response to direct head impact allowed for isolation of
ATD response and head padding response in athletic impact tests.
85
CHAPTER 5 – ATHLETIC PADDING INFLUENCE ON ATD HEAD-NECK IMPACT DYNAMICS
5 CHAPTER 5 – ATHLETIC PADDING INFLUENCE ON ATD HEAD-NECK
IMPACT DYNAMICS
INTRODUCTION
In Chapter 4 the behavior of the Hybrid III ATD in response to direct
head impact was quantified. In this Chapter, the isolated Hybrid III response
is built upon via in vitro Hybrid III athletic padding impacts. Particular
attention is paid to the head-cervical spine loading behavior, and multiple
injury risk parameters quantified in Chapters 2 and 3 are broadly expanded
to include specific skull, brain and neck injury measures. Finally, theories on
the importance of considering both linear and rotational dynamic response
metrics are explored via potential deficits in current athletic testing standards
that utilize only the Gadd Severity Index. These results help to build the
foundation for the Matlab simulations in Chapter 6 that attempts to address
the most adequate impact testing scenario for athletic headgear.
BACKGROUND
Historically, linear kinematics has been the gold standard by which to
measure athletic head injury risk. But linear and rotational dynamics are
both widely acknowledged as contributors to concussive traumatic brain
injury. Current head injury metrics, namely the Gadd Severity Index (GSI)46
and Head Injury Criterion (HIC)163, are based on resultant linear head center
of gravity (cg) acceleration and are incapable of predicting rotational
concussion risk as well as ignoring any motion between the head-neck
junction. These injury metrics were developed from human cadaveric head
86
impacts and have long been correlated to skull fracture risk but not
necessarily brain injury. As HIC, GSI, resultant linear head cg acceleration
and impact duration are the most widely used and only currently-mandated
injury metrics for athletic head padding impact testing,4;101;142 in spite of the
voluminous literature pointing to rotational kinetics and kinematics being
responsible for concussion and spine injury, athletic helmets and head
padding are still certified solely by the GSI linear kinematic criterion. More
troubling is the fact that while it is commonly believed that padding reduces
athletes’ concussion and spine injury risk, there is little evidence that
confirms this hypothesis as few studies have investigated padding effects on
injury risk to the head or spine. Furthermore, no published studies have
investigated padding effects on concussion or neck injury risk in striking
sports like boxing and mixed martial arts (MMA). Therefore, while athletic
padding may be capable of reducing linear dynamic quantities like GSI, HIC,
linear acceleration and impact force, padding effects on rotationally-induced
concussion and risk for spine injury are unclear. In this study, athletic type
head impacts are delivered to a padded Hybrid III ATD in an attempt to
quantify concussion and spine injury risk for different padding conditions for
different impact directions and energy levels.
LITERATURE SURVEY
Several prior studies have investigated athletic head or hand padding
effects in sports such as American and Australian Football, boxing, soccer,
rugby and equestrian events. In these sports, padding has been investigated
in impact tests from 9J to 354J using headforms, anthropomorphic
87
surrogates, volunteers or cadavers.14;23;52;70;76;80;87;89;102;140;154;166;172 With the
exception of one American Football helmet study166 and one study of jockey
and equestrian helmets52, rotational head-neck dynamics have been
neglected from both protective athletic head padding studies. Interestingly,
many studies have reported variable linear responses, with some padding
models increasing, and other models decreasing, linear kinematic parameters
like GSI, HIC or head acceleration.13;41;89;102;140;154;166;172 The American
football helmet study showed that linear and rotational dynamics were no
different in newer helmets versus a widely used existing helmet.166 Most
importantly, several head padding studies reported that none of the models
tested provided adequate concussion protection when compared with bare
head impacts.69;89;154;172 Many studies have also indicated that padding
thickness, composition and contact friction play important roles in reducing
concussion risk.41;52;68;89;154;172 Because of these varying head impact
attenuation characteristics of athletic padding, and the lack of data on
padding effects on rotational dynamics, concussion risk and spine injury,
investigation of padding effects on rotational dynamics, concussion risk and
spine injury is still unclear.
METHODS
This study, which was part of a larger investigation into head impact
dynamics, concussion and spine injury risk, investigated the linear and
rotational dynamics of boxing and mixed martial arts (MMA) padding during a
series of fifty-four (54) pendulum impacts at low (27-29J) and high (54-58J)
energy to the lateral head of an instrumented Hybrid III Anthropomorphic
88
Test Device (ATD.) A total of five (5) padding combinations were examined:
(a) unpadded, (b) MMA glove-unpadded head, (c) boxing glove-unpadded
head, (d) unpadded pendulum-boxing headgear and (e) boxing glove-boxing
headgear. From these tests, a total of twenty-nine (29) established and
proposed head and spine injury parameters were measured or calculated,
including fifteen (15) ‘linear’, five (5) ‘rotational’ and nine (9) combined
parameters.
A Hybrid III 50th Percentile ATD was instrumented with a triaxial linear
accelerometer (Model EAS3-250 Measurement Specialties, Hampton, VA) and
triaxial angular velocity sensor (Model ARS-06S, ATA Sensors, Albuquerque,
NM) mounted at the head cg. A set of three triaxial linear accelerometers
were mounted externally on the head but this data will be presented in future
work. A six-channel upper neck load cell (Model 1716A, Denton ATD,
Rochester, MI) measured forces and moments in three directions. Moments
measured by the load cell were translated to the occipital condyles by
multiplying respective x- and y-axis shear forces by the 0.01778m moment
arm. A triaxial linear accelerometer was mounted in the ATD chest but was
later found to have minimal magnitude and will not be presented here. All
data channels were collected at 5000Hz, filtered according to SAE J211143
and sign convention adhered to SAE J1733145. Linear head jerk, angular jerk
and angular acceleration were calculated via filtered signal differentiation.
Linear head velocity was calculated via filtered signal integration. The ATD
head weighed 5.1kg with instrumentation, the cervical spine weighed 1.6kg
and the ATD torso and upper extremities weighed 21.2kg.
89
A steel sphere was precisely machined to create a 3.6kg pendulum
impactor mass. The mass was chosen to approximate the effective upper
extremity mass of volunteer boxers during similar punch testing in a prior
study.168 A steel eye hook was threaded into the flat machined face and the
impactor was balanced such that the mass moment of inertia was translated
only in the vertical direction. The same steel sphere was used in all trials.
A 6.4mm steel braided pendulum cable was hung from a free-swinging
carabiner secured to a ceiling strut located approximately 5 meters directly
above the ATD head cg. The ATD was secured to a test stand with tie-down
straps and lower extremities removed at the femur. Inertial responses below
the neck were minimized by securing the test stand with 3500N of sandbags.
The pendulum mass was raised into position via nylon fishing line with 222N
rupture strength, tied to the eye hook and routed through a ceiling-mounted
pulley aligned with the lateral ATD head surface.
In each head impact trial, the pendulum mass was raised to 0.76m or
1.52m and the nylon line was cut with a pair of scissors. The swing heights
were selected to re-create low and high energy impacts on par with boxing
punches to the lateral ATD head. These punch data will be presented in a
future publication, but produced peak ATD linear head cg accelerations of 24-
72g, angular accelerations of 2100-4400rad/s2, angular velocities of 18-
31rad/s and occipital neck moments of 25-100N-m, respectively. These
impact dynamics were on par with prior boxing studies124;141;148;165 and in the
range of dynamics proposed to cause cTBI.22;37;38;43;57;85;104;105;114;115;122;153;178
90
Tuf-Wear training headgear (0.63kg), Everlast Pro-Style boxing
training glove (0.27kg) and UFC Official MMA glove (0.18kg) were tested.
The boxing and MMA gloves were firmly affixed to the pendulum via a
combination of double sided tape on the mass and gaffers tape on the
exterior. The headgear was secured with double-sided tape on the head and
by cinching taut the chinstrap and posterior lacing. The impact conditions
are described in Table 5-1 and shown in Figure 5-1 through Figure 5-5.
Impact
Impact
Impact Setup Momentum
Energy (J)
(N-s)
Low High Low High
Unpadded 26.8 53.7 13.9 19.7
Bare Pendulum-Headgear 26.8 53.7 14.6 19.7
MMA-Bare Head 28.3 56.5 15.0 20.7
Boxing-Bare Head 28.9 57.9 13.9 21.2
Boxing-Headgear - 57.9 - 21.2
Figure 5-1. Unpadded Impact Setup
91
Figure 5-2. Bare Pendulum-Headgear
Figure 5-3. MMA-Bare Head
Figure 5-4. Boxing-Bare Head
92
Figure 5-5. Boxing-Headgear Impact Setup (High Energy Only)
A total of six (6) trials were conducted for the five impact setups and
the two impact energy levels for a total of 54 impacts. The boxing-headgear
setup was omitted in the low energy condition. Similarly, the MMA glove was
not used in headgear impacts because the boxing glove delivered a more
severe impact under similar test conditions. The multiple trials at each swing
height reduced variations due to small changes in impact location or impactor
rotation. The coefficient of variation (COV) between setup trials was 2.1% to
13.5% for linear cg resultant acceleration. Unpadded head impacts served
as controls for each setup at the two impact velocities.
Energy and Momentum Variability in Setup
Because the swing height was kept constant between setups, the
impact energy and momentum varied by 5.0% to 7.5% between the low and
high energy setups due to the changing pendulum mass. These variations
were considered when analyzing the two-tailed t-test results for the dynamic
head impact parameters.
Further test variation was found due to the headgear mass, which
increased inertial impact resistance. As a means to estimate changing
93
impactor and head mass effects on the dynamic parameters, a simplified
two-body elastic collision model was developed as shown in Figure 5-6.
Figure 5-6. Simplified elastic collision model
Theoretical
Post-Impact Vh
Impact Setup
Low High
Energy Energy
Unpadded 3.21 4.54
Bare Pendulum-Headgear 3.30 4.67
MMA-Bare Head 3.34 4.73
Boxing-Bare Head 2.99 4.23
Boxing-Headgear - 4.42
Table 5-2. Theoretical post-impact head velocity
Thus, while keeping pendulum swing height constant, adding padding
to the impactor would theoretically increase post-impact head velocity by
2.9% to 4.3%, adding padding to only the head would theoretically decrease
94
post-impact head velocity by 6.8% and adding padding to both the impactor
and head would theoretically reduce head velocity by 2.6%. These
theoretical head post-impact velocity differences from the control setup
provided a test variation range to consider when scrutinizing the dynamic
head impact parameter t-test results.
Parametric Analysis
For each setup, twenty-nine (29) relevant concussion and spine injury
parameters were analyzed. Parameters requiring more in-depth description
are presented in the Appendix. The parameters were grouped into ‘linear’,
‘rotational’ or ‘combined’ groups. All vector quantities were reported as
resultant values. The linear group consisted of fifteen (15) parameters:
resultant head cg acceleration, resultant head cg jerk, resultant head cg
delta-V, resultant upper neck force, resultant linear head cg momentum,
resultant impact force, maximum resultant impact force loading rate,
minimum resultant impact force loading rate, head kinetic energy, head
power, impact duration, Gadd Severity Index (GSI)46, Head Injury Criterion
(HIC)163, HIC duration163 and Skull Fracture Correlate (SFC)161. The
rotational group consisted of five (5) parameters: resultant head angular
velocity, resultant head angular acceleration, resultant head angular jerk,
resultant upper neck moment and resultant head angular momentum. The
combined group incorporated both linear and rotational components and
comprised the remaining nine parameters: Neck Injury Criteria (Nij),40
Generalized Acceleration Model for Brain Injury Threshold (GAMBIT),103
weighted Principal Component Score (wPCS),57 Head Impact Power (HIP),105
95
Cumulative Strain Damage Measure (CSDM-0.05, 0.10 and 0.15 levels),151
Dilatation Damage Measure (DDM)151 and Relative Motion Damage Measure
(RMDM).151 For each parameter, the maximum value, from the mean of the
six trials in each setup, was used in a two-tailed paired t-test comparison
against the unpadded impacts at either low or high energy. Significance was
defined based on a p value less than 0.05.
RESULTS
Kinematic (linear head cg acceleration) and kinetic (occipital force)
data were plotted for each of the six trials in each padding condition at high
energy and are shown in Figure 5-7 and Figure 5-8, respectively. Since all of
the impacts were laterally directed, the resultant values were analyzed and
individual x-, y- and z-axis components were not examined separately since
linear and rotational dynamics were largest in the directions of largest motion
and loading.
96
Figure 5-7. Head cg resultant linear acceleration in high energy trials for the four padding
conditions tested
97
Figure 5-8. Occipital neck force resultant in high energy trials for the four padding conditions
tested
As seen in Figures 5-7 and 5-8, there was some degree of variability
during the sequence of six repeated impacts. As a means to quantify this
variability in the data collected, Table 5-3 collates the respective coefficient
of variation (COV) of each impact condition for the set of six (6) repeated
impact trials. The COV was calculated by dividing the standard deviation
from each set of six impacts by the mean value at the time of highest linear
cg acceleration resultant.
98
Low Energy
Bare-
Quantity Control UFC-Bare Box-Bare
Headgear
Linear CG Acceleration 4.7 6.9 11.8 12.4
Angular Velocity 10.3 24.6 10.0 19.9
Angular Acceleration 10.2 22.1 21.1 13.2
Occipital Condyle Force 7.4 21.1 22.0 16.9
Occipital Condyle Moment 15.1 14.0 27.4 16.9
High Energy
Bare- UFC- Box- Box-
Quantity Control
Headgear Bare Bare Headgear
Linear CG Acceleration 2.1 4.3 13.5 9.9 37.7
Angular Velocity 12.7 29.0 10.2 9.3 33.3
Angular Acceleration 8.4 25.5 19.4 10.6 20.1
Occipital Condyle Force 3.5 5.9 10.8 9.3 23.3
Occipital Condyle Moment 7.8 10.5 20.2 7.8 9.9
Table 5-3. Coefficient of variation of each resultant impact parameter corresponding to time
of highest cg linear acceleration resultant
Thus, even though the conditions were nearly identical for each
padded impact trial set, the COV shown in Table 5-3 was seen to vary from
as low as 2.1% (no padding) to as high as 37.7% (boxing glove-boxing
headgear). Trials conducted with impactor and/or head padding saw
increased variation that indicated some difficulty in generating highly
repeatable and precise impact conditions.
Mean ATD resultant head cg linear acceleration, head angular velocity,
head angular acceleration and impact force from the five impact setups are
presented in Figures 5-9 through 5-12. The computational results of the
Simulated Injury Monitor (SIMon) software151, used to generate CSDM, DDM
and RMDM damage measures, is demonstrated in Figure 5-13.
99
Figure 5-9. Head cg resultant linear acceleration in high and low energy trials
Figure 5-10. Head resultant angular velocity in high and low energy trials
100
Figure 5-11. Head resultant angular acceleration in high and low energy trials
Figure 5-12. Resultant impact force in high and low energy trials
101
Figure 5-13. Sample of computational simulation results displaying Mises stresses within brain
from Box-Bare impacts using SIMon software
As can be seen in these figures, by padding either the impactor or
ATD, the peak resultant linear head acceleration decreased at both energy
levels. Resultant head angular velocity increased for the bare-headgear and
MMA-bare conditions at both energy levels. The boxing-headgear setup had
a similar resultant angular velocity peak as the control. Resultant head
angular acceleration increased in the MMA-bare setup and only decreased
slightly in the bare-headgear setup at high energy; in low energy impacts, all
padded setups reduced peak resultant angular acceleration. The impact
force trends followed the same pattern as resultant cg linear acceleration,
with the control impact having the highest resultant impact force and the
boxing-headgear having the greatest resultant impact force reduction.
The twenty-nine (29) dynamic head and neck impact parameters were
measured or computed for each setup by taking the mean of the maximum
102
values from the six trials. These parameters were then compared, via a two-
tailed t-test, to the unpadded control impacts for the low and high energy
setups. The results are summarized in Table 5-4 and Table 5-5. The
maximum value for each of the parameters is shown with the percentage
decrease from the unpadded control setup. Based on this t-test comparison,
parameters significantly lower at p<0.05 level are italicized and parameters
significantly higher at p<0.05 level are shaded.
Low Energy Results
For the linear dynamics t-test results, impact duration and HIC
duration increased significantly from the control, while all other parameters
significantly decreased by 33% to 86% across all setups. While the linear cg
acceleration was significantly reduced in all three setups by 59% to 64%, the
head delta-V remained relatively unchanged due to longer impact duration.
The rotational parameters were all significantly reduced in the three
setups with the exception of angular velocity, occipital neck moment and
angular momentum. The 61% decrease in angular jerk in the MMA-bare
setup was nearly significant (p=0.078.) All of the setups with impactor
padding increased peak angular velocity while the bare-headgear setup
significantly decreased angular velocity. The occipital neck moment and
angular momentum showed a similar variable response dependent upon
padding.
All of the low energy combined dynamic parameters decreased when
using padding except for the Nij and CSDM05 in the boxing-bare setup,
although the 26% reduction in boxing-bare CSDM05 was nearly significant
103
(p=0.066.) Because the DDM was zero for all setups, this parameter was not
included in the t-test comparison.
Low energy tests

Bare-
Parameter max value Unpadded UFC-Bare Box-Bare
Headgear
LINEAR PARAMETERS
Impact Duration (s) 0.0037 0.0157 0.0172 0.0136
Delta-V (m/s) 2.13 2.23 2.08 1.95
Head Accel (g) 153 54.7 54.7 62.0
Jerk (m/s3) 1.62E+06 2.51E+05 2.99E+05 3.81E+05
Neck Force (N) 778 397 478 518
Momentum (N-s) 10.8 11.3 10.6 11.2
Impact Force (N) 8380 3080 3190 3920
Max Force Load Rate (N/s) 7.16E+06 9.82E+05 1.22E+06 1.51E+06
Min Force Load Rate (N/s) -7.41E+06 -1.29E+06 -1.48E+06 -2.11E+06
KE head (J) 11.6 12.7 11.0 10.9
Power head (W) 1.01E+04 3.91E+03 3.91E+03 4.52E+03
GSI 281 51.0 42.0 56.0
HIC 250 43.0 34.0 48.0
HIC Duration (s) 0.00140 0.00360 0.00300 0.00300
SFC 117 51.0 58.0 54.0
ROTATIONAL PARAMETERS
Angular Velocity (rad/s) 15.8 16.7 18.5 12.8
Angular Accel (rad/s2) 4810 2960 2440 2450
Angular Jerk (rad/s3) 2.36E+06 9.22E+05 7.26E+05 8.83E+05
Occipital Neck Moment (N-m) 32.7 18.5 34.0 23.5
Angular Momentum (N-m-s) 0.269 0.3049 0.3236 0.2451
COMBINED PARAMETERS
Nij 0.240 0.160 0.260 0.210
GAMBIT 0.624 0.236 0.230 0.257
wPCS 87.4 31.8 29.6 33.2
HIP 9720 3930 3670 3860
CSDM05 60.6 39.7 44.8 19.5
CSDM10 10.2 2.00 1.90 0.50
CSDM15 1.60 0.00 0.00 0.00
DDM 0.00 0.00 0.00 0.00
RMDM 1.89 1.43 1.16 1.27
Table 5-4. Low energy impact maxima (shaded = significant increase p<0.05, italics =
significant decrease p<0.05)
High Energy Results
In a similar fashion to the low energy t-test results, the high energy
parameters were analyzed for significant differences that might have been
due more to the mass effect as opposed to padding properties. Again, head
104
delta-V, linear momentum and kinetic energy may have been potentially
affected by changing head or pendulum mass between setups. Both impact
duration and HIC duration significantly increased across all setups; the longer
impact duration allowed for head delta-V to remain relatively unaffected even
though cg acceleration decreased by 38% to 72%. The other linear dynamic
parameters were significantly reduced by 28% to 90% for all setups.
The high energy rotational dynamic parameters showed much greater
variation than the low energy results. These tests had largest variation in
angular velocity, occipital neck moment and angular momentum. The
padded pendulum setups significantly increased angular velocity by 22% to
41% and angular momentum by 18% to 30% while having a varying effect
on angular jerk (MMA-bare decreased 9% while boxing-bare significantly
decreased by 19%) and occipital moment (MMA-bare and bare-headgear
setups significantly decreased while boxing-bare and boxing-headgear
showed no significant differences). The boxing-bare setup had no significant
effect on angular acceleration and was nearly significant for impact moment
(p=0.082). The bare-headgear setup significantly decreased every rotational
parameter. The boxing-headgear setup significantly decreased all
parameters except angular momentum (6% increase; p<0.05), angular
velocity (1% decrease) and occipital moment (13% increase; p=0.13).
When examining the combined dynamic parameter results it was found
that GAMBIT, wPCS, HIP and DDM all significantly decreased for all setups
and RMDM was significantly decreased with the exception of a 6% increase
for the MMA-bare setup. The three CSDM results varied across setups; the
105
boxing-bare setup had significant increases of 29% to 121% and boxing-
headgear setup significantly decreased 67% to 100%, respectively. The
MMA-bare setup consistently had higher CSDM values, ranging from 22% to
102% higher than control, with the CSDM05 level being significantly higher.
In contrast, the bare-headgear setup had significant reduction in CSDM05 of
29% while the 60% (p=0.062) and 65% (p=0.061) reduction in CSDM10 and
CSDM15 were nearly significant, respectively. When examining spine injury
risk the Nij significantly decreased for the MMA-bare and boxing-headgear
setups while showing no significant decrease for either the boxing-bare or
boxing-headgear setups.
106
High energy
Bare- Box-
Parameter max value Unpadded UFC-Bare Box-Bare
Headgear Headgear
LINEAR PARAMETERS
Impact Duration (s) 0.00350 0.0114 0.0143 0.00730 0.0182
Delta-V (m/s) 2.99 3.04 3.01 2.85 2.78
Head Accel (g) 232 129 117 144 65.0
Jerk (m/s3) 2.74E+06 9.54E+05 8.48E+05 1.27E+06 3.10E+05
Neck Force (N) 1250 746 836 868 629
Momentum (N-s) 15.2 15.5 15.3 16.3 15.9
Impact Force (N) 12800 7160 6630 8900 4240
Max Force Load Rate
1.14E+07 3.96E+06 3.36E+06 5.47E+06 1.26E+06
(N/s)
Min Force Load Rate
-1.22E+07 -4.64E+06 -4.21E+06 -6.71E+06 -1.77E+06
(N/s)
KE head (J) 22.7 23.7 23.1 23.3 22.1
Power head (W) 2.14E+04 1.29E+04 1.19E+04 1.53E+04 6.76E+03
GSI 768 265 219 311 79.0
HIC 676 229 188 271 66.0
HIC Duration (s) 0.0012 0.0020 0.0022 0.0020 0.0038
SFC 200 127 112 124 60.0
ROTATIONAL PARAMETERS
Angular Velocity
18.5 22.6 26.2 16.3 18.3
(rad/s)
Angular Accel (rad/s2) 5260 5550 5240 3800 1740
Angular Jerk (rad/s3) 2.57E+06 2.32E+06 2.08E+06 1.60E+06 6.30E+05
Occipital Neck Moment
48.8 34.0 47.7 38.2 42.4
(N-m)
Angular Momentum
0.345 0.406 0.449 0.321 0.364
(N-m-s)
COMBINED PARAMETERS
Nij 0.39 0.28 0.38 0.31 0.33
GAMBIT 0.936 0.541 0.491 0.583 0.264
wPCS 173 80.9 71.3 87.8 35.8
HIP 20500 12700 11600 12700 5430
CSDM05 63.6 77.4 81.9 45.2 0.210
CSDM10 12.5 21.8 24.5 5.00 0.40
CSDM15 2.4 4.8 5.3 0.8 0.0
DDM 0.8 0.0 0.0 0.0 0.0
RMDM 2.35 2.48 2.17 2.03 1.08
Table 5-5. High energy impact maxima (shaded = significant increase p<0.05, italics =
significant decrease p<0.05)
DISCUSSION
While generally reducing linear dynamic parameters, including the
Gadd Severity Index (GSI), in both low and high energy impacts, rotational
107
head and spine injury risk significantly decreased, significantly increased or
were unaffected across the range of setups tested in this study. Prior athlete
head padding literature showed variable linear and rotational head and spine
injury risk mitigation dependent upon impact conditions.12;52;89;102;140;154;166;172
In particular, a recent study examining NFL football helmets found that of the
four new helmet models tested against an existing helmet model, none of the
new helmets consistently reduced linear or rotational head and spine injury
risk during a series of ten real-world impact recreations. It should be noted
that this prior study conducted descriptive statistical analyses and could not
be directly evaluated with the comparative statistical analysis from this
study.
It was seen in the low energy tests that padding always reduced linear
and combined dynamic parameters with the exception of the Nij and the
parameters potentially affected by the changing mass between setups.
Rotational dynamic parameters had more variable results. Specifically,
variability in angular velocity, occipital neck moment and angular momentum
indicated an impact behavior sensitive to padding compression,52;67;89;154;172
interface coefficient of friction41 or other dynamic relationship. Because the
MMA-bare and boxing-bare impacts were conducted within the envelope of
typical punch impact magnitude, the increases in angular velocity and
variability in Nij and occipital neck moment may have altered concussion and
spine injury risk for competitors participating in training or competition
without head padding. With the exception of Nij and CSDM05, a consistent
decrease in combined dynamic head injury parameters for all low energy
108
setups was seen. The RMDM results, which were a correlate for brain
bridging vein rupture, were cause for concern. For low energy setups, the
RMDM exceeded the tolerance limit of 1.0, indicating striking sport athletes
may be at risk for brain injury even with the use of the hand or head padding
tested here.
In the high energy tests, the results were more variable than observed
in the low energy tests. These tests were conducted at the higher range of
human punch magnitude, while still being below impact magnitudes found in
NFL collisions.38;165;168 Again, mass effects may have been present in head
delta-V, momentum and kinetic energy; the significance of these results may
have been different if impact momentum or energy were held constant
between setups. For the rotational parameter comparisons, several of the
more noteworthy results related to angular velocity, angular acceleration,
occipital moment and angular momentum. The significant increase in
angular velocity for the MMA-bare and boxing-bare setups may have implied
greater risk of rotational concussion and/or neck injury versus a bare-fist
impact. And while the bare-headgear condition significantly reduced angular
velocity, the real-world condition of boxing-headgear showed no change from
the control impact; this could have implied that this combination of
protective equipment used in sparring may do little to prevent potential
concussion and/or neck trauma related to angular velocity. It was intriguing
that in high energy impacts, MMA-bare occipital moment significantly
decreased from 48.8N-m to 34.0N-m, while the boxing-bare occipital
moment decreased only slightly to 47.7N-m, respectively. Thus, glove
109
selection likely played a role in resulting bending moment magnitude
endured by the cervical spine even though both gloved conditions indicating
increased brain injury risk. This variation in occipital moment between the
setups but could have been also been due to known stiffness issues with the
Hybrid III neck.1;74;113 The significant increase in the angular momentum
when using real-world impact setups (MMA-bare, boxing-bare, boxing-
headgear) was interesting. Again, this result indicated that athletes
subjected to repeated blows while using these padding combinations may not
be maximally protected against concussion or neck injury due to angular
momentum. Finally, the combined head injury parameters GAMBIT, wPCS,
HIP and DDM were significantly reduced for all padding setups when
compared with control. Spine injury, as predicted by Nij, was significantly
reduced for MMA-bare and bare-headgear but not the boxing-bare or boxing-
headgear setups. This result might have indicated that the interactions
between the boxing glove or MMA glove and head were important
considerations to minimize spine injury risk, but it must be noted that all Nij
results were well below the injury risk threshold of 1.0. The CSDM was only
significantly reduced for the boxing-headgear setup while showing an
increase for both the MMA-bare and boxing-bare setups. As CSDM is a
correlate to diffuse axonal injury (DAI) and concussion, this result was
potentially troubling. In these particular setups, the unpadded setup was
more effective at reducing DAI and concussion risk than using padded
gloves. Finally, the RMDM results indicated concussion all padding setups
exceeded the threshold of 1.0 for brain bridging vein rupture. This result
110
may be of concern for athletes who use head and hand padding tested here
but are subjected to repeated blows near these high energy impact
magnitudes.
LIMITATIONS
It is important to note that several study limitations exist. Primarily,
the contact characteristics between the ATD head, padding and pendulum in
these tests may not have been an exact replica of the friction and
deformation response in real-world striking sport head impacts. Secondly,
even though impact velocity was held constant, the changing pendulum
and/or head mass increased impact energy and momentum by up to 7.5%.
The changing mass also had a theoretical influence on head delta-V by
increasing bare head setup delta-V by up to 4.3% while decreasing headgear
setup delta-V by up to 6.8%. Therefore, the significance of several results,
including delta-V, momentum and kinetic energy, may not have been as
pronounced if energy or momentum was held constant between impacts and
may not have necessarily been solely a function of padding. Similarly, the
low energy MMA-bare angular velocity and bare-headgear angular
momentum differences could be due partially to this mass effect. Thirdly, it
was hypothesized that real-world striking sport punches may involve variable
impact mass due to muscle force and inertial effects. Thus, the constant
3.6kg pendulum effective arm mass may not have been a proper
representative of real-world fist-head impacts. Additionally, only lateral
impacts, mimicking severe hook punches from an instrumented volunteer,
were analyzed. The padding effects on dynamic parameters due to straight,
111
uppercut or oblique punches have yet to be studied. Finally, because the aim
of this study was to macroscopically compare “off-the-shelf” athletic padding
effects on concussive head impact parameter mitigation, the padding
materials used were not subjected to rigorous physical or elemental
characterization. The unquantified padding compression, morphology and
construction characteristics may have all played a role in the presence or lack
of statistical significance in any single dynamic head injury parameter.
CONCLUSIONS
The Gadd Severity Index (GSI) has been the gold standard by which to
quantify athletes’ head injury risk for nearly five decades. However, GSI
accounts only for linear acceleration of the skull while ignoring rotational
motion and upper spine loading. Rotational kinematic parameters are
universally acknowledged as a significant contributor to concussive brain
injury in athletes. And the cervical spine, which defines the loading boundary
conditions for the head, may play a role in concussion as well. In spite of the
voluminous literature pointing to rotational kinematics being partially
responsible for concussion as well as spine injury, athletic helmets and head
padding are still certified solely by the GSI linear kinematic criterion without
testing for brain injury or spine injury risk. More troubling is the fact that
while it is commonly believed that head and/or fist padding reduces athletes’
concussion and spine injury risk, there is little evidence that confirms this
hypothesis in high energy impact sports like boxing or football. Furthermore,
no published studies have investigated padding effects on concussion and
spine injury risk in striking sports like boxing and mixed martial arts (MMA).
112
This test series of fifty-four (54) low (27-29J) and high (54-58J)
energy pendulum impacts to an instrumented Hybrid III anthropomorphic
test device (ATD) showed that measured Hybrid III linear head acceleration
(55 to 144 g), angular acceleration (1741 to 5552 rad/s2), angular velocity
(16 to 26 rad/s), neck moment (19 to 49 N-m) and impact force (3083 to
8901 N) were within ranges of theoretical concussion and spine injury limits
in spite of GSI (42 to 311) being below the accepted skull fracture limit of
1200. These results indicated that head injury risk due to linear head cg
acceleration, linear head jerk, resultant impact force, resultant occipital
force, resultant impact force loading rate, head power, HIC, GSI, SFC,
GAMBIT, wPCS and HIP in striking sports was significantly (p<0.05) reduced
via head and hand boxing and MMA padding at both energy levels. Spine
injury risk predicted by Nij was below the accepted threshold in all setups but
was significantly decreased in only four of the seven setups tested. Padding
always significantly (p<0.05) increased impact duration, leading to no
change or an increase in linear head momentum and kinetic energy at both
energy levels. And in high energy impacts head angular velocity, CSDM and
RMDM rotational parameters were unaffected or increased. These
parameters were found to be in the range of predicted onset of diffuse axonal
injury (DAI) and bridging vein rupture regardless of the head and hand
padding used.
These results have implications for striking sport athletes at risk of
concussion or spine injury during training or competition. These tests
indicated that while padding protects against head injury from linear dynamic
113
parameters it may do little to decrease the risk of rotationally-induced spine
injury and concussion under lateral impacts. The multivariate parametric
analysis presented here might prove useful in determining safe impact
thresholds for striking sport athletes, future protective athletic padding
development, developing impact test standards for protective padding and/or
in comparing head injury mitigation properties of different types of athletic
padding.
114
CHAPTER 6 – HELMET TESTING IMPACT SIMULATIONS
6 CHAPTER 6 – HELMET TESTING IMPACT SIMULATIONS
INTRODUCTION
The results from the Hybrid III athletic padding impacts in Chapter 5
indicated that padding inserted between the impactor and ATD head
influenced head, neck and spine injury risk – in some cases, padding
increased head and neck injury risk. These results are particularly
interesting when considering the multiple impact test methodologies that are
used to test athletic padding – drop tests, pendulum impacts and linear
pneumatic impacts. Since these various test methodologies may influence
head, neck and spine injury risk results, the purpose of this Chapter is to (i)
collect pilot football helmet impact data and (ii) use this data to develop and
validate a numerical helmet impact model in Simulink and (iii) compare
injury risk results from five (5) impact test setups. Thus, the Chapter 5
pendulum impact test boundary conditions influence on injury risk is
compared with current helmet test methodologies used by the National
Football League (NFL) and National Operating Committee on Safety in
Athletic Equipment (NOCSAE) via simplified one- and two-degree-of-freedom
system dynamics models. Additional attention was paid to consideration of
linear acceleration (current testing standards) versus consideration of linear
and rotational kinematics (future testing standards). It was found that the
pendulum impact setup from Chapter 5 most closely resembles a linear
impactor method used by the NFL with a flexible Hybrid III ATD neck.
Furthermore, drop tests and horizontal impact tests with a rigid head-neck
115
coupling currently used by NOCSAE may induce high head accelerations while
simultaneously ignoring rotational-based concussive and spine trauma. In
Chapter 7, the numerical pendulum impact test method explored in this
Chapter is employed via a unique physical impact pendulum to analyze
cadaver and ATD injury response to direct head loads.
BACKGROUND
Historically, linear head center of gravity (cg) acceleration has been
the gold standard by which to measure athletic head injury risk, with
concussion and neck injury having no widely accepted measurement
methodology. In spite of linear acceleration being the gold standard, many
authors have attempted to unlock the relationship between inertial and direct
contact loading and understand the linear and rotational injury mechanisms.
Brief Review of Linear Kinematics and Head Injury
The rationale for using linear acceleration as the metric to predict head
injury was first postulated in the 1940’s by Gurdjian and Lissner et al.58-60;81
who attributed intracranial damage to skull deformation and pressure
gradients developed by acceleration after direct head impact. In these
studies, angular acceleration and negative internal pressure were deemed to
have little significance when examining head injury. In 1967, Ommaya et
al.116 concluded that inertial rotation alone could not produce injury levels
caused by direct head impact. This was later revised by Ommaya and
Hirsch115 in 1971 that inertially induced rotation could account for
approximately half of brain injury potential, with injury due to direct impact
comprising the other half. Gennarelli et al.48;50 showed that linear translation
116
generated only focal effects like cerebral contusions and intracerebral
hematomas while rotational effects were likely responsible for diffuse axonal
injuries. This study also suggested that linear motion likely induced brain
cavitation, which was thought responsible for blood vessel rupture due to
vacuum. In a well-known study in 1971, Unterharnscheidt159 found that
linear acceleration generally induced intracranial pressure gradients while
rotational acceleration appeared to cause shearing stress between the tissue
and skull. Finally, Ono et al.118 conducted a series of primate studies and
found no correlation between concussion and angular acceleration and
concluded concussion could be produced by linear acceleration from direct
impact.
As a result of this work, well known linear acceleration-based injury
criteria, including the Gadd Severity Index (GSI),46 Head Injury Criterion
(HIC),81 Skull Fracture Criterion (SFC),65 Society of Automotive Engineering
Injury Thresholds,144 and Generalized Brain Injury Threshold (GAMBIT)103
have been developed. But while these linear acceleration-based criteria
correlate well with skull fracture risk, it has been widely thought that
concussion and internal brain damage is a function of rotational kinematics
as well.60;75;115
Brief Review of Rotational Kinematics and Head Injury
In 1943, Holbourn66 was the first to hypothesize that inducing
rotational acceleration via inertial and/or direct head loading could cause
cerebral concussion as well as coup-countrecoup contusion via shear and
tensile strains within the brain tissue. In a study of bridging vein rupture
117
from 1975, Lowenhielm82 proposed that angular acceleration caused gliding
contusion from excessive strain in the cerebral vessels. Gennarelli and
Thibault performed a series of studies47-50;153 to investigate rotational
kinematic influence on live non-human primates and anthropomorphic
surrogates. In these studies the authors concluded that angular acceleration
contributed more than linear acceleration to the generation of concussion,
diffuse axonal tearing and acute subdural hematomas via induction of intra-
tissue shear strains. In 1995, McLean90 found in a series of fatal traffic
crashes that zero cases of brain injury occurred in absence of direct head
impact and that brain injury was not generally associated with neck injury.
In 2003, Takhounts et al.151 produced the first publicly available Finite
Element brain injury package called SIMon (Simulated Injury Monitor). This
FE package is driven by both linear and angular acceleration inputs and
returned risk values for skull fracture, diffuse axonal injury, dilatational
damage due to vacuum and acute subdural hematoma due to tissue tearing.
A finite-element based study by Zhang et al. (2004) showed that shear strain
and shear strain rate due to rotationally-induced shearing within the brain
are likely indicators of brain injury severity and concussion.178 In 2005,
Viano et al.165 studied concussive impact case studies in the National Football
League with video reconstructions, helmeted ATD impact tests and the
Wayne State Human Model Finite Element simulation. These case studies
showed that brain shear strain and shear strain rate were highest in impacts
with highest linear acceleration and angular acceleration. However, no
118
definitive relationship was defined for head and neck dynamics as a function
of type of helmet worn.
Using these rotational kinematic injury studies and the aforementioned
linear kinematic studies, head injury criteria with rotational and linear
components have been proposed such as Head Impact Power (HIP),105
Simulated Injury Monitor (SIMon),151 peak shear strain rate,178 weighted
Principle Component Score (wPCS)57 and Louis Pasteur University model
(ULP).86
Headform Development and Helmet Certification
During the second half of the 20th Century, interest mounted in athletic
helmet testing as players were sustaining severe and fatal head and neck
trauma during competition.132;133;155 In an effort to institute the first helmet
injury quantification test, while researchers at Wayne State University were
developing the HIC and GSI criteria, they also created the Wayne State
University impact headform. This headform differed from the available
impact surrogates, like the Hybrid II or Hybrid III automotive crash testing
headforms, in that it contained a glycerin-filled skull and pliable overmolded
polymer to create impedance similar to human cadaver heads. The Wayne
State headform was subsequently adopted by the newly formed NOCSAE
(National Operating Committee on Safety in Athletic Equipment) and utilized
in drop testing with athletic helmets in the 1975.64 A simple test, wherein
the helmeted headform was dropped onto an elastomeric pad of varying
thickness, was devised and the peak headform cg linear resultant
acceleration and GSI were collected. A GSI threshold of 1500 – since
119
lowered to 1200 in 1998 – was selected for Abbreviated Injury Scale (AIS)
level 3 head injury based on data from earlier cadaver testing and skull
fracture risk.46;98;163
Athletic helmet testing continues in the exact same manner to this
day, with NOCSAE being the only testing entity and all helmets being
certified solely based on skull fracture risk related to GSI. Currently, no
criterion enables quantification of helmet mitigating properties on brain, neck
or spine injury.
Review Summary
Hence, though much research has been done on head, neck and spine
injury risk for direct and inertial loading, injury parameters like the Gadd
Severity Index (GSI) and Head Injury Criterion (HIC) developed in the 1960’s
remain the only widely used head injury risk metrics. And GSI is the
exclusive metric by which NOCSAE tests all athletic helmets in headform drop
tests. But GSI and HIC were developed from a series of cadaveric impact
tests and correlate to skull fracture risk; as such, GSI and HIC have no
conclusive empirical relationship to brain injury, concussion, neck or spine
injury.
In an effort to understand direct head impact effects on brain,
concussion, neck and spine injury risk, in this Chapter pilot football helmet
impact experiments were conducted, the data from these experiments were
used to validate a pendulum impact model in the Matlab toolbox function
Simulink and this validated model was then compared to existing NOCSAE
120
drop test models as well as proposed pneumatic linear impact models with
rigid and flexible surrogate necks.
METHODS
Pilot Pendulum Impact Tests
The pilot pendulum impact tests were conducted with collaborators at
SEA Limited in Columbus, Ohio. Using five impact conditions – unhelmeted,
‘leatherhead’ and using three current NOCSAE-approved elite-level football
helmets. Lateral impacts were conducted with the impactor striking a seated
Hybrid III 50th percentile male ATD (Denton ATD, Rochester, MI) in the head.
The ATD head was instrumented with a triaxial linear accelerometer (Model
EAS3-250 Measurement Specialties, Hampton, VA) and a triaxial angular rate
sensor (Model ARS-06S, ATA Sensors, Albuquerque, NM) mounted at the
center of gravity (cg). A six-channel occipito-cervical load cell (Model 1716A,
Denton ATD, Rochester, MI) measured spine forces and moments in three
directions. The moments measured by this supracondylar load cell were
translated inferiorly to the occipital condyles by multiplying respective y-axis
shear force by the 0.01778m moment arm. All data were collected at
5000Hz, filtered according to SAE J211143 and sign convention adhered to
SAE J1733145. Angular acceleration was calculated by differentiating the
angular velocity signal after determining proper signal content and pre-
differentiation filter requirements. A single impact was conducted with each
helmeted condition at impact velocity of 6.5m/s, producing impact energy of
150J and impact momentum of 45N-s. The impact energy and momentum
121
were selected to be similar to NOCSAE rigid-neck drop testing impacts with a
9kg mass traveling at 4.2-5.5m/s (80-134J and 38-50N-s, respectively.)
The impacting mass, a helmeted Hybrid II headform, weighed 7kg and
was selected to mimic the impact impedance characteristics of the real-world
impactor used in the proposed NOCSAE linear impact standard. The striking
mass was uninstrumented. The impactor was hung from four (4) 6.4mm
steel braided cables that were free to swing from a ceiling-mounted
carabineers above the ATD head. The ATD was seated in a custom-built
plywood platform that permitted free head-neck motion in the lateral
direction and the torso secured with tie-down straps. In order to minimize
inertial responses below the cervical spine, the ATD below the waist was
weighed down with approximately 2500N of sandbags. For each impact trial,
the impactor cg was position such that the initial impact location was in the
coronal plane +/- 2cm from the head cg. The impactor was released from a
consistent swing height of 2.18m above ATD head cg by pulling the nylon
tension cable to failure. In addition to being in the range of impact velocity,
energy and momentum for NOCSAE drop tests, this swing height was
selected because it produced impacts within the concussive range of
magnitudes from prior athletic head impact studies.122;165 In this study, a
total of five (5) head impact trials were performed as shown in Figure 6-1
and Table 6-1.
122
(A) (B)
(D) (E)
(C)
Figure 6-1. Pendulum Impact setups for (A) Unhelmeted, (B) ‘Leatherhead’, (C)-(E) NOCSAE
Approved Helmets
Impact
Impactor Drop Height Impact
ATD Pendulum Impacts Momentum
Mass (kg) (m) Energy (J)
(N-s)
Lateral 7 2.18 45 150
Impact Test Simulations
After conducting the impact tests, a computational simulation model
was built in Simulink and validated against the response of the three
NOCSAE-approved helmets. This model incorporated a two-degree-of-
freedom spring-mass-damper system. The helmet was modeled as a lumped
element, with a single mass and parallel spring-damper replicating the
behavior of the helmet shell, padding and facemask. The model is shown in
Figure 6-2.
123
Figure 6-2. Two DOF system dynamics model for SEA, Ltd. lateral pendulum helmet impacts
The Hybrid III head cg linear acceleration ( x

 N ) was used as the main
validation parameter in the Simulink model against the lateral impacts, with
the peak angular velocity (  N ) being a secondary validation check. For this
model, mass, stiffness and damping parameters were all taken from values
found in literature.9;32;36;123;150;164;176
Using the validated lateral helmet impact test simulation parameters,
Simulink models were then constructed to examine (a) NOCSAE headform
calibration drop test, (b) NOCSAE helmeted headform certification drop test,
(c) horizontal pneumatic impact with helmeted headform and rigid neck and
(d) horizontal pneumatic impact with helmeted headform and flexible Hybrid
124
III neck. These models are illustrated in Figures 6-3, 6-4, 6-5 and 6-6,
respectively.
Figure 6-3. Unhelmeted lateral NOCSAE calibration drop test system dynamics model
125
Figure 6-4. Helmeted NOCSAE lateral impact test system dynamics model
126
Figure 6-5. Proposed NOCSAE linear impact with highly rigid neck attachment
Figure 6-6. Linear impact used in prior NFL studies with flexible Hybrid III neck
Each of the impact testing methods (3-NOCSAE, 1-NFL, 1-SEA) were
exercised in Simulink under constant initial energy and momentum
127
(b)
conditions. The current NOCSAE drop test calibration protocol served as the
lower bound for initial energy (105J) and momentum (38N-s). The lateral
impact tests conducted at SEA, Ltd. were the next highest initial energy
(150J) and momentum (45N-s) conditions studied. Finally, the NOCSAE
calibration initial energy and momentum were increased by a factor of two
and four, respectively, to analyze response to higher velocity impacts like
those conducted in prior NFL head trauma research.21 The inputs for each
model under these conditions are shown in Tables 6-2 and 6-3.
105J 150J 210J 420J

CONSTANT ENERGY Moving
SIMULATED TEST Mass Initial Initial Initial Initial
METHOD (kg) Velocity Velocity Velocity Velocity
(m/s) (m/s) (m/s) (m/s)
NOCSAE-
7.0 5.5 6.5 7.7 10.9
CALIBRATION
NOCSAE-HELMETED
9.0 4.8 5.8 6.8 9.6
DROP
NOCSAE-LINEAR
14.5 3.8 4.5 5.4 7.6
IMPACT RIGID NECK
NFL LINEAR IMPACT
14.5 3.8 4.5 5.4 7.6
FLEXIBLE NECK
SEA PENDULUM
7.0 5.5 6.5 7.7 10.9
IMPACTS
Table 6-2. Simulink simulation parameters for constant kinetic energy comparison
128
38N-s 45N-s 76N-s 154N-s

CONSTANT
Moving
MOMENTUM Initial Initial Initial Initial
Mass
SIMULATED TEST Velocity Velocity Velocity Velocity
(kg)
METHOD (m/s) (m/s) (m/s) (m/s)
NOCSAE-
7.0 5.5 6.4 10.9 21.9
CALIBRATION
NOCSAE-HELMETED
9.0 4.2 5.0 8.5 17.0
DROP
NOCSAE-LINEAR
14.5 2.6 3.1 5.3 10.6
IMPACT RIGID NECK
NFL LINEAR IMPACT
14.5 2.6 3.1 5.3 10.6
FLEXIBLE NECK
SEA PENDULUM
7.0 5.5 6.4 10.9 21.9
IMPACTS
Table 6-3. Simulink simulation parameters for constant momentum comparison
RESULTS
The pilot lateral pendulum impact results from testing conducted at
SEA, Ltd. for unhelmeted, ‘leatherhead’ and helmeted Hybrid III are shown in
Figures 6-7, 6-8, 6-9, 6-10 and 6-11.
129
No Helmet Leatherhead Helmet 1 Helmet 2 Helmet 3
20.0
10.0
0.0
0.490 0.500 0.510 0.520 0.530 0.540
-10.0
Linear Acceleration (g)
-20.0
-30.0
-40.0
-50.0
-60.0
-70.0
-80.0
-90.0
-100.0
time (sec)
Figure 6-7. Hybrid III head cg linear acceleration for the single (n=1) pendulum head impacts

5.0
0.0
0.490 0.500 0.510 0.520 0.530 0.540
Angular Velocity (rad/sec)
-5.0
-10.0
-15.0
-20.0
-25.0
-30.0
-35.0
time (sec)
Figure 6-8. Hybrid III head angular velocity for n=1 pendulum head impacts
130

3000
Angular Acceleration (rad/sec^2)
2000
1000
0
0.490 0.500 0.510 0.520 0.530 0.540
-1000
-2000
-3000
-4000
-5000
time (sec)
Figure 6-9. Hybrid III head angular acceleration for n=1 pendulum head impacts
700
600
500
Neck Force (N)
400
300
200
100
0
0.490 0.500 0.510 0.520 0.530 0.540
-100
time (sec)
Figure 6-10. Hybrid III head upper neck shear force for n=1 pendulum head impacts
131
40
30
Neck Moment (N-m)
20
10
0
0.490 0.500 0.510 0.520 0.530 0.540
-10
-20
-30
time (sec)
Figure 6-11. Hybrid III head upper neck bending moment for n=1 pendulum head impacts
The Simulink impact test simulation results are shown in comparison
with the average head cg linear acceleration response from the three lateral
helmet impact tests with NOCSAE-approved helmets in Figure 6-12.
132
TEST AVERAGE SIMULATION
30.0
20.0
10.0
0.0
acceleration (g)
0 0.01 0.02 0.03 0.04 0.05

-10.0
-20.0
-30.0
-40.0
-50.0
-60.0
-70.0
time (sec)
Figure 6-12. Simulink simulation results compared with average response from Hybrid III
head cg linear acceleration for the three (3) helmeted pendulum impacts
The secondary simulation validation parameter - maximum head
angular velocity - is shown in Table 6-4.
Maximum
Angular
ATD Head Impact Direction
Velocity
(rad/s)
Lateral Impact Test
-26.2
Average
Simulation -30.5
Table 6-4. Additional validation parameter comparison
After the lateral impact simulation was validated against the data
collected in SEA pendulum impact tests, each of the aforementioned impact
test methodologies (NOCSAE drop test calibration, NOCSAE helmeted drop
test, NOCSAE horizontal impact test with rigid neck, NFL horizontal impact
test with flexible neck and SEA pendulum impact tests) was exercised in
133
Simulink under constant energy and momentum conditions. The resulting
lateral headform cg linear acceleration for each of the impact conditions is
shown in Figure 6-13 and Figure 6-14.
Figure 6-13. Constant energy Simulink results for headform cg linear acceleration
134
Figure 6-14. Constant momentum Simulink results for headform cg linear acceleration
DISCUSSION OF RESULTS
The pilot pendulum impact tests were conducted in order to gather in
vitro baseline impact data for the subsequent Simulink validation. Though
the resulting data should be viewed with caution, as only a single impact was
conducted for each of the unhelmeted, ‘leatherhead’ and three (3) helmeted
conditions at impact energy and momentum equal to 150J and 45N-s,
respectively, some interesting preliminary results were notable.
135
The average helmeted impact resulted in peak linear cg head
acceleration (-62.3g) that was lower than either the unhelmeted (-92.1g) or
‘leatherhead’ (-77.1g) impacts. This result was unsurprising, as the modern
helmets have undergone several design iterations to optimize protection
against skull fracture and the NOCSAE drop test measurement of resultant
linear headform cg acceleration. However, it was intriguing that the World
War I era ‘leatherhead’, having a large degree of desiccation and being
tested as-is with zero reconditioning, reduced the linear acceleration
approximately half as well as the modern helmets.
The ATD rotational responses were more interesting. The angular
velocity results indicated that in this limited lateral impact dataset that the
helmeted impacts (-26.2rad/s) and the ‘leatherhead’ (-26.6rad/s) had similar
protection improvement over the unhelmeted impact (-32.6rad/s). When
examining the ATD head angular acceleration for these pilot impacts, the
helmeted impacts (-4445rad/s2) and ‘leatherhead’ (-4543rad/s2) were both
slightly higher than the unhelmeted impact (-4332rad/s2). As concussion
risk has been linked to both linear and angular head kinematics,75 the fact
that in this limited dataset modern NOCSAE-approved helmets failed to
protect against rotational concussion injury better than the ‘leatherhead’ was
surprising. This preliminary result could indicate that the NOCSAE-approved
helmets have been optimized to reduce linear acceleration but have not yet
been designed to mitigate rotational kinematics.
The loading at the occipito-cervical (head-neck) junction showed that
the helmeted impacts (526N) reduced peak lateral shear force better than
136
the ‘leatherhead’ (569N) when compared with the unhelmeted impact
(642N). One possible explanation for this reduction could have been that the
lower head acceleration for the helmeted and ‘leatherhead’ scenarios
contributed to lower dynamic force acting at the skull base. The mean
helmeted neck moment in lateral bending for the helmeted impacts was
slightly higher (34.5N-m) than the ‘leatherhead’ (31.8N-m) and the
unhelmeted (32.8N-m) impacts. This result mimics the preliminary
rotational kinematic results and may indicate that helmets have been
optimized to reduce linear kinetics but have yet to reduce rotational kinetics.
After analyzing the pilot pendulum impact test data, a simplified two
degree-of-freedom system dynamics simulation was developed and
implemented in Simulink. This model incorporated both linear and rotational
spring, damper and mass/inertia elements. The simulation was validated
against the linear cg acceleration and showed very good correlation for the
initial impact loading rate, peak acceleration and unloading rate. The
simulation angular velocity (-30.5rad/s) was slightly higher than the helmet
test maximum of -26.2rad/s.
After validating the lateral pendulum tests at 150J and 45N-s, the
pendulum model was exercised in Simulink with the NOCSAE calibration
model, NOCSAE drop-test model, NOCSAE linear impact-rigid neck model
and the NFL linear impact-rigid neck model at constant energy (105J, 150J,
210J and 420J) and constant momentum (38N-s, 45N-s, 76N-s, 154N-s).
For the constant energy simulations, the headform linear acceleration was
137
seen to always be highest for the unhelmeted NOCSAE calibration test,
followed by the NOCSAE helmeted drop test; the NOCSAE and NFL linear
impact tests, as well as the SEA pendulum tests were fairly tightly grouped
together regardless of the initial energy. As expected, the unhelmeted
calibration test also had the lowest impact duration, due to the hard contact
between headform and MEP pad. The linear impacts (NOCSAE, NFL and SEA)
all had similar impact durations and all were longer than the helmeted drop
test. Of note was that the linear impacts with deformable neck always had
longer impact durations and slightly lower linear acceleration peak. Thus,
when considering constant energy impacts, the varying mass for each of the
five impact conditions studied will influence the resulting linear cg
acceleration for the headform.
The constant momentum results showed similar trends in that the
calibration and helmeted NOCSAE drop tests always had highest linear
acceleration. However, it was seen that for the higher momentum that the
helmeted NOCSAE drop test had longer contact duration than the three linear
impact conditions studied. This result may point to some deficiency in the
elastomeric MEP pad model and likely requires further investigation. Again,
the linear impacts with hybrid III neck model showed slightly lower pulse
width but in the cases of higher momentum showed some difference in peak
acceleration when comparing the NFL to the SEA models. Physical testing
was needed to bear out these results but the constant momentum results
indicate that the impact condition and impactor mass had heavy influence on
the resulting impact pulse magnitude, shape and duration.
138
LIMITATIONS
The simulation work done in this Chapter had limitations. Primarily,
the models analyzed here were very simplified in comparison to variable real-
world impact testing. The simple 1DOF (NOCSAE drop test, NFL linear
impactor) and 2DOF (NFL with flexible neck, Cleveland Clinic) models studied
here lay the foundation for more thorough simulation analysis and parametric
study. Hence, while the helmeted parameters were verified via ATD-helmet
lateral impact experimentation, in order to truly develop robust simulation
models, omnidirectional impacts and multi-DOF simulations should be
validated against complex real-world data for head and neck dynamics.
Another limitation was that system damping characteristics were estimated
and show some resonance in the simulation results. Because this resonance
was not seen in the lateral ATD impact data, system damping requires more
investigation.
CONCLUSIONS
The results from the physical helmeted impact testing indicated that
much more investigation is needed to explore football helmet linear and
rotational dynamics behavior. While only a few lateral impacts were
collected here, it would be of high interest in future work to examine the
linear and rotational dynamic performance of all eleven (11) current
NOCSAE-approved adult football helmets in multiple impact locations, impact
velocities and with different end conditions like constrained neck, whole body
ATD, ATD on the moving sled, etc.
139
The Simulink model validated for the pilot helmet impacts was
compared with four other impact conditions. These models showed that the
difference in mass and impact orientation led to widely differing linear
acceleration responses. Because a main goal of NOCSAE and laboratory
impact tests have been to make the impacts as realistic as possible, the
simulation results indicated that drop testing helmeted headforms without a
neck did not produce a real-world impact scenario when compared with the
NFL linear impact simulation modeling boundary conditions at the base of the
spine. Further, it appeared from the simulation results that it would be
difficult to compare drop testing headform kinematics with linear impact
kinematics since the impact conditions, including impactor mass, energy and
momentum, had great influence over the resulting headform dynamics.
While much more physical testing would be needed to validate all the
Simulink results, the results presented in this Chapter indicated that impact
test method, impactor mass, impact energy and impact momentum all
merited careful consideration when developing a protocol to study dynamic
head and neck injury response. Ultimately, any horizontal impact method
could be used to deliver a suitable ‘real-world’ impact to a test device.
Therefore, based on the results from Chapter 6, and for ease of portability,
versatility and cost, a unique pendulum impactor designed and constructed
out of extruded aluminum sections was developed for use in cadaver and
ATD impacts in Chapter 7 versus more expensive and less-versatile
pneumatically-driven linear impactor.
140
CHAPTER 7 – IMPACT PENDULUM DESIGN, PIG SPINE TESTING, FOOTBALL HELMET IMPACTS
7 CHAPTER 7 – IMPACT PENDULUM DESIGN, PIG SPINE TESTING,
FOOTBALL HELMET IMPACTS
INTRODUCTION
The Chapter 6 pilot helmet impacts and Simulink simulations found
that a real-world horizontal impact to an ATD head-neck surrogate can be
delivered via pendulum or pneumatics-based impactor, but a headform-only
drop testing setup was less desirable. In this Chapter, the design,
development, validation and preliminary testing from a pendulum-based
impact system is detailed. This pendulum impactor is rigorously designed
and validated with the express goal of enabling future highly controlled
cadaveric and ATD impact tests that allow analysis of a full range of head
impact conditions at impact velocities of ~6m/s and under.
The first portion of the Chapter describes the pendulum engineering
design, revisions and quantification of fixture hardware effects on test
conditions. The second portion describes use of the pendulum impactor for
pilot cadaveric head impact experiments as well as dynamic and statistical
analysis that were conducted in design-of-experiment support of an
anticipated future study using large-scale cadaveric testing. Finally,
helmeted ATD head impacts are conducted and similar statistical analyses
are discussed in support of anticipated future helmet impact tests and the
relationship between head, neck and spine injury risk during direct head
impacts with varying impact energy and momentum.
141
The work completed in this Chapter, including frame vibratory Fourier
Transform analysis, load cell impedance investigation and power analysis of
pilot head impacts, provides the necessary foundation in design-of-
experiment to pursue rigorous and full-fledged studies into cadaveric and
ATD head, neck and spine injury in the future. Furthermore, this Chapter
begins to expand on a key finding from Chapter 4 in that by simply reporting
a resultant peak magnitude as the only measure of injury risk -- as is done
with HIC or GSI -- might omit crucial information regarding x-, y- and z-axis
contributions to the dynamic force-time history. Hence, multi-directional
helmeted impacts conducted in this Chapter also take into consideration the
components contributing to each resultant value for head and neck injury.
IMPACT PENDULUM
DESIGN
Impact pendulums have been used for several decades to deliver
highly controlled and repeatable impacts to cadaveric tissue as well as
human surrogate test devices. Chosen in the current study over other well-
known impact systems like the linear pneumatic impactor or drop test
system, the current impact pendulum design was driven by a desire for
simplicity, ease-of-use, cost effectiveness and ability to conduct impacts with
high throughput. Several designs were considered based on existing
pendulum devices currently in use around the world. Figure 7-1 shows some
examples.
142
Figure 7-1. Examples of impact pendulums -- BioKinetics (Ottawa, Canada), NOCSAE

(Knoxville, TN), Johns Hopkins Applied Physics Lab (Baltimore, MD), Disney Epcot Center
(Orlando, FL), Texas Transportation Institute (College Station, TX), Impact Jackets LLC (Silver
Spring, MD)
The required impact velocity was predetermined to be approximately 6
m/s in order to replicate lower energy impacts seen in youth head impacts as
143
well as osteoligamentous cadaver testing. A simple A-frame construction
was chosen in order to make the pendulum construction simple and allow for
portability. Available ceiling clearance dictated that the pendulum overall
height be less than ten feet tall, and the width and depth were chosen to
accommodate test specimens and personnel. Extruded aluminum sections
were created in SolidEdge V2.0 (Siemens PLM Software, Plano, TX) and
assembled to meet the aforementioned dimensional restrictions as shown in
Figure 7-2. This assembly was forwarded to a manufacturer (80/20 Inc.,
Columbia City, IN) for production. After the aluminum sections were
prepared and delivered, the pendulum was assembled as shown in Figure 7-
2.
Figure 7-2. SolidEdge Rendering and first version
After initial assembly, the A-frame was found to have less rigidity than
desired. The frame was retrofit with lateral and longitudinal stabilizers and
144
supplemental extruded aluminum sections were ordered and assembled as
shown in Figure 7-3.
Figure 7-3. Retrofit rendering and assembled final pendulum
After finalizing the assembly of the pendulum, the impact fixtures were
similarly designed and ordered. Images of the cadaveric and ATD impact
setups are shown in Figure 7-4.
145
Figure 7-4. Cadaveric and ATD impact setups
The test fixtures themselves were designed to accommodate both the
cadaveric or ATD head-neck preparations using interchangeable fasteners
and components. The fixtures allow for 360-degree rotation, zero to ninety
degree tilt and submillimeter indexing in the x- and y-shear force directions.
For cadaveric specimens, a specially machined 3.6kg steel impactor - used in
Chapters 4 and 5 - was used. Helmeted ATD impacts used a NOCSAE
medium headform secured inside a Riddell VSR-4 large adult helmet. A
surface contact DC electromagnet capable of 500N (cadaver) or 1500N (ATD)
pulling force was used to raise and hold the impactor mass steady prior to
impact as shown in Figure 7-5.
146
Figure 7-5. Electromagnets used in cadaver [500N] and ATD [1500N] impacts
For the cadaveric pig spine impacts, a two-cable system was used to
prevent impactor rotation and for ATD impacts, a three-cable system was
used to both prevent impactor rotation as well as impactor vertical
alignment.
VIBRATION ISOLATION
In order to understand how spurious inputs from vibrations within the
frame itself might influence impact test results, a series of hammer strikes
against the frame were conducted to attempt to isolate these inputs. The
frame was struck in a number of locations as shown in Figure 7-6 and the
six-channel load cell (ATI Omega IP65, ATI Industrial Automation, Apex, NC)
collected data for the vibration duration. A control data set, with load cell
acquiring data in a static environment, was collected as well.
147
Figure 7-6. Frame vibration analysis hammer strike locations
After collecting the hammer strike frame vibration data, a fast Fourier
transform (FFT) algorithm was written and implemented in Matlab v2009a
(Mathworks, Natick, MA) using the measured load cell vertical force strain
gage. Before conducting the FFT, the data were filtered according to impact
data collection standard SAE J211/1.143 The FFT results are shown in Figure
7-7, with each hammer strike compared against the FFT control data.
148
Single-Sided Amplitude Spectrum of Load Cell Strain Gage2(t) Single-Sided Amplitude Spectrum of Load Cell Strain Gage2(t)
0.05 0.05
Top Hammer Strike Left Top Hammer Strike
0.045 Baseline 0.045 Baseline
0.04 0.04
0.035 0.035
0.03 0.03
|Y(f)|
|Y(f)|
0.025 0.025
0.02 0.02
0.015 0.015
0.01 0.01
0.005 0.005
0 0
0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500
Frequency (Hz) Frequency (Hz)
Single-Sided Amplitude Spectrum of Load Cell Strain Gage2(t) Single-Sided Amplitude Spectrum of Load Cell Strain Gage2(t)
0.05 0.05
Left Middle Hammer Strike Left Lower Hammer Strike
0.045 Baseline 0.045 Baseline
0.04 0.04
0.035 0.035
0.03 0.03
|Y(f)|
|Y(f)|
0.025 0.025
0.02 0.02
0.015 0.015
0.01 0.01
0.005 0.005
0 0
0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500
Frequency (Hz) Frequency (Hz)
Figure 7-7. FFT results for the four (4) frame hammer strikes. NOTE: Chart areas enlarged
from original scale of Frequency = 5000Hz, lY(f)l = 0.8 to show locations of interest
Based on the FFT results, each hammer strike produced negligible
vibratory response when compared with the baseline control data. The
highest magnitude vibrations were seen for the hammer strike to the left
side-bottom of the frame at 23Hz, 120Hz and 205Hz, but these magnitudes
occurred at less than 1% of the total signal amplitude. Vibrations stemming
from the pendulum support location (top hammer strike) showed signal
amplitude on par with the baseline control data. With the FFT analysis
complete, confidence was established that collected impact data was isolated
from any vibratory response of the frame itself.
149
FIXTURE MASS
The fixture mass affixed to the top of the load cell, in between the load
cell and base of cadaveric spine, had a inertial mass effect on load cell
readings that was unknown. In order to quantify this effect, as seen in
Figure 7-8, load cell data was collected under two conditions: (1) with a
surrogate mass fastened to the load cell and (2) with the same surrogate
mass attached to the fixture sitting atop the load cell. Pendulum impacts
were conducted with the same energy and headform impact location for both
conditions. The load cell forces and moments were compared for these two
conditions to quantify the fixture impedance differences.
Figure 7-8. Fixture mass quantification for fastening directly to load cell and fastening to
fixture components sitting atop load cell
A total of six (6) impacts were conducted with the 3.6kg impactor and
impact speed of 3.43m/s (impact energy = 21.2J, impact momentum =
150
12.3N-s) for the two mounting conditions. In addition to the six-channel load
cell, the ATD headform was instrumented with three (3) uniaxial 500g linear
accelerometers (Model M62 Measurement Specialties, Hampton, VA) in an
orthogonal arrangement at the headform center of gravity and three (3)
uniaxial 210 rad/s angular velocity sensors (DTS ARS12K, Seal Beach, CA).
The mean ATD response was analyzed for each data channel. For purposes
of brevity, only the resultant head center of gravity acceleration, head
resultant angular velocity and resultant bending moment load cell data is
presented here.
140
Mounted to Chuck
Mounted Directly
120
100
Linear acceleration (g)
80
60
40
20
0
0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Figure 7-9. Fixture mass effects - headform resultant linear acceleration
151
20
Mounted to Chuck
18 Mounted Directly
16
Angular Velocity (rad/s) 14
12
10
0
0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Figure 7-10. Fixture mass effects - headform resultant angular velocity
60
Mounted to Chuck
Mounted Directly
50
Bending Moment (N-m)
40
30
20
10
0
0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Figure 7-11. Fixture mass effects - load cell resultant bending moment
152
From Figures 7-9 and 7-10 it can be seen that the headform
kinematics were affected minimally regardless of the mounting condition.
But the resultant bending moment displayed in Figure 7-11 showed that the
mounting to the chuck produced higher bending moments for the impact
duration. Because of the additional 13.1cm moment arm between the chuck-
mounting and load cell, the shear force contributions were responsible for
this increased bending moment. This offset was then qualitatively taken into
account when analyzing moments developed during pig cadaveric impacts.
CADAVERIC PIG SPINE IMPACTS
BACKGROUND
Debilitating or fatal head, neck and spine injuries are prevalent in
motor vehicle crashes and athletics. Historically, design of automotive safety
systems and athletic personal protective equipment has generally focused on
protection discretely designed on biomechanical studies formulated on a
component basis (head or neck), but not a systems basis (connected head-
neck). But the head and neck are a mechanically linked osteoligamentous
structure that behaves as a function of inputs to the head as well as
boundary conditions supplied by the skull base and neck. Component-based
injury protection designs, such as those used in athletics (Gadd Severity
Index-GSI), military (Head Injury Criterion-HIC, peak brain cavitation
pressure) or crash injury (Neck Injury Criteria-Nij), have utilized head and
spine injury risk criteria developed from impact testing related to the
automotive industry. While a plethora of studies have analyzed head (HIC,
GSI) or neck injury risk (Nij) separately, relatively few studies have taken a
153
systems-based approach to understand coupled head-neck injury risk. Thus,
head-neck injury protection during impact may not be as trivial as separate
consideration of head or neck injury thresholds.
In this portion of Chapter 7, a series of cadaveric head impact tests
were conducted with the express intent of (i) developing a reliable method to
study head and osteoligamentous cervical spine linear and rotational dynamic
response to repeated anterior-posterior sub-injurious athletic head impacts,
(ii) collect dynamic impact data from at least four (4) instrumented spinal
column levels in at least three (3) cadaveric specimens, (iii) analyze these
data to ascertain if an increase in established head injury risk criteria (HIC,
GSI, linear acceleration, angular acceleration, angular velocity) precedes an
increase in spine injury criterion (neck forces, neck moments, Nij) during the
repeated impact sequence, (iv) Determine if rotational dynamics (angular
acceleration, angular velocity, occipital moment) increase at a different rate
than linear impact parameters (head acceleration, impact force, occipital
shear force), and (v) gather sufficient data to conduct a sampling power
analysis to determine the necessary number and type of specimens for future
experiments.
SPECIMEN PREPARATION AND INSTRUMENTATION
A total of three (3) fresh-frozen porcine cadaveric head-cervical spine
specimens were procured by the Spine Research Laboratory at Cleveland
Clinic and were tested in vitro. The specimens were stripped of soft tissue
and non-osteoligamentous structures. Cranially, the skull base was prepared
by removing the mandible and sawing through the skull along a line
154
comprising the occipital protuberance dorsally and the hyoid bone ventrally.
Any remaining brain and spinal cord tissue was removed from the superior
spinal column and an aluminum striking plate was affixed to the skull based
with 1” wood screws. Caudally, the specimen was prepared by removing
spinal segments from the third thoracic vertebrate inferiorly. The specimen
was potted using a low-melting point cerrobend alloy poured around
vertebral bodies C7-T2 and allowed to harden until reaching room
temperature. The specimen was kept moist with saline solution and at room
temperature while the potting material hardened completely. Figure 7-12
shows an overview of the specimen preparation procedure.
Figure 7-12. Cadaveric specimen preparation: Removing mandible, Removing caudal

structures and sawing through skull, Testing orientation, Potting with cerrobend liquid metal –
sawbones replica
155
The skull base, C1, C3 and C5 vertebral bodies were prepared for
instrumentation by insertion of a 1.3mm threaded surgical pin to which was
affixed a specially machined 1cm x 1cm x 1.5cm sensor mounting block
made from a dice blank. For each mounting block, three (3) 500g uniaxial
linear accelerometers were attached with the exception of C5 which only had
room for two accelerometers. Additionally, a single uniaxial angular velocity
sensor was mounted on each mounting block to monitor sagittal plane
rotation. The ATI six-axis load cell measured reaction forces and moments
at the base of the cervical spine. All data were collected via National
Instruments PXI-1042 module (National Instruments Corp, Austin, TX) and
filtered in Matlab according to SAE J211/1 at 10kHz. The spine sign
convention adhered to SAE J1733.145
After instrumentation was affixed, pre-impact the specimen was
positioned at a 40-degree anterior tilt, necessary to approximate a neutral
posture and prevent pre-impact extension. Each specimen was exercised
through a flexion-extension range of motion for ten (10) cycles to normalize
viscous tissue elements. The aforementioned 3.6kg steel pendulum
impacted the aluminum interface plate with impact velocity, energy and
momentum detailed in Table 7-1 for a total of six (6) impacts per specimen.
Swing Height Impact Velocity Energy Momentum

Test
(m) (m/s) (J) (N-s)
1 0.033 0.80 1.2 2.9
2 0.134 1.62 4.7 5.8
3 0.293 2.40 10.3 8.6
4 0.410 2.84 14.5 10.2
5 0.732 3.79 25.9 13.6
6 1.063 4.57 37.5 16.4
Table 7-1. Cadaver impact matrix
156
The sixth impact caused injury in each of the three specimens tested.
In between impacts, the specimen was inspected visually and also subjected
to manual flexion-extension manipulation to ensure no obvious
osteoligamentous injury. Injury was documented post-test via dissection and
autopsy. Figure 7-13 illustrates the instrumentation and testing sequence.
Figure 7-13. Instrumentation and testing sequence: Affixing sensors to mounting blocks,
Initial cadaver position at 40-degrees forward tilt, National Instruments DAQ system and
Occipito-cervical disruption caused in specimen 001 after sixth impact
Due to the voluminous data collected, a respective sample of linear
acceleration, angular velocity, force and bending moment test results are
shown in Figures 7-14 through 7-18, respectively. The linear acceleration
and angular velocity data are displayed from the four instrumented spinal
levels during the 25.9J impact. The force and moment data are shown for
157
the 4.7J, 10.3J, 14.5J and 25.9J impacts, respectively. Data from the 37.5J
impact are not shown here due to the resulting macroscopic injury in all
three specimens at this impact energy level.
Skull Base C1
400 400

CAD01 CAD01
CAD02 CAD02
300 CAD03 300 CAD03
200 200
100 100
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
C3 C5
400 400

CAD01 CAD01
CAD02 CAD02
300 CAD03 300 CAD03
200 200
100 100
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
Figure 7-14. Resultant linear acceleration results from three cadaver specimens for the fifth
impact (KE = 25.9J)
158
Skull Base C1
200 200

100 100
0 0
CAD01 CAD01
CAD02 CAD02
CAD03 CAD03
-100 -100
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
C3 C5
200 200

100 100
0 0
CAD01 CAD01
CAD02 CAD02
CAD03 CAD03
-100 -100
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
Figure 7-15. Sagittal plane angular velocity results from three cadaver specimens for
the fifth impact (KE = 25.9J)
159
5 5
x 10 Skull Base x 10 C1

1 1
0.5 0.5
0 0
-0.5 CAD01
-0.5 CAD01
CAD02 CAD02
CAD03 CAD03
-1 -1
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
5 5
x 10 C3 x 10 C5

1 1
0.5 0.5
0 0
-0.5 CAD01
-0.5 CAD01
CAD02 CAD02
CAD03 CAD03
-1 -1
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
Figure 7-16. Sagittal plane angular acceleration results from three cadaver specimens for the
fifth impact (KE = 25.9J)
160
4.7J 10.3J
500 500
CAD01 CAD01
400 CAD02 400 CAD02

CAD03 CAD03
Force (N)
Force (N)
300 300
200 200
100 100
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
14.5J 25.9J
500 500
CAD01 CAD01
400 CAD02 400 CAD02

CAD03 CAD03
Force (N)
Force (N)
300 300
200 200
100 100
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
Figure 7-17. Resultant force results from three cadaver specimens across four impacts (KE =
4.7J, 10.3J, 14.5J, 25.9J)
161
4.7J 10.3J
CAD01 CAD01
100 CAD02 100 CAD02
Moment (N-m)
Moment (N-m)
CAD03 CAD03
50 50
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
14.5J 25.9J
CAD01 CAD01
100 CAD02 100 CAD02

Moment (N-m)
Moment (N-m)
CAD03 CAD03
50 50
0 0
0 0.05 0.1 0 0.05 0.1
time (s) time (s)
Figure 7-18. Resultant moment results from three cadaver specimens across four impacts (KE
= 4.7J, 10.3J, 14.5J, 25.9J)
Using these cadaveric impact tests as pilot data, a power analysis was
conducted to ascertain the number of specimens needed in a full-scale
experiment. Hence, while there were obvious anthropometrical differences
between the porcine specimens tested here and the desired future testing in
a human cadaver head-neck, it was hoped that this power analysis would
provide a reasonable estimate on sample size necessary while avoiding any
unnecessary destructive testing on human cadaveric tissue.
The power analysis used a significance level  = 0.05 and
benchmarked 80% power as the minimum required. Because so many
variables were measured, the maximum of the resultant value from each of
these variables, including linear acceleration, angular velocity, force and
162
bending moment were analyzed. The null hypothesis (Ho) was that for each
parameter collected, the individual test resulted in a maximum value equal to
that of the mean of maximum values. The alternative hypothesis (Ha) was
that deviation of at least 10% from the mean of maximum sample values
occurred. In short, the analysis determined what necessary sample size
would be needed to accurately detect a 10% change in the peak value of
any measured parameter at least 80% of the time with a 5% test of
significance. This analysis also requires that the allowable probability of a
Type II error is 20% or less. The power analysis plots are show in bar
format in Figures 7-19 through 7-22.
Acceleration
Necessary samples for Power = 0.80
100
50
37.5
25.9 C5
14.5 C3
10.3
C1
4.7
Occiput
Energy (J) 1.2
Figure 7-19. Resultant linear acceleration power analysis results from three cadaver
specimens and four instrumented spinal levels
163
Angular Velocity

200
150
100
50
37.5
25.9 C5
14.5 C3
10.3
C1
4.7
Occiput
Energy (J) 1.2
Figure 7-20. Angular velocity power analysis results from three cadaver specimens and four
instrumented spinal levels
164
Angular Acceleration

200
150
100
50
37.5
25.9 C5
14.5 C3
10.3
C1
4.7
Occiput
Energy (J) 1.2
Figure 7-21. Angular acceleration power analysis results from three cadaver specimens and
four instrumented spinal levels
165
Force/Moment

200
150
100
50
37.5
25.9
14.5
10.3
Moment
4.7
Energy (J) Force
1.2
Figure 7-22. Force and moment power analysis results from three cadaver specimens and
four instrumented spinal levels
As seen in the previous figures, for rotational data like angular velocity
and angular acceleration, the required samples were generally higher than
linear acceleration or the measured force or moment from the load cell.
Based on this pilot set of cadaveric impacts, and because of the variability in
measured test data, the mean number of specimens necessary to achieve
80% power across all parameters was n > 62. It is likely infeasible to
conduct a set of cadaveric experiments with such a large sample size,
therefore, it would be advisable to conduct experiments on individual spinal
components – such as a functional spinal unit – versus the highly
unconstrained and variable spinal kinematic chain studied here.
166
HELMETED ATD IMPACTS
After conducting the cadaveric impacts and pilot test power analysis, a
similar set of experiments were conducted with helmeted ATD. This was
done because there is high need for tightly controlled ATD direct head impact
experimentation to compare with results from future cadaveric studies, as
well as a need to garner fundamental information regarding the reliability of
data taken from helmeted impacts.
In these impacts, a helmeted NOCSAE medium headform was attached
to a Hybrid III 50th percentile ATD neck. The base of the neck was bolted
directly to the six-channel force/moment transducer mentioned previously. A
total of three (3) linear accelerometers like those used in the cadaveric study
were attached to the mounting location near the headform center of gravity
and a custom mount was crafted to enable attachment of three (3) angular
velocity sensors near the headform ‘mouth’. Because of small errors in
machining the angular velocity sensor mounting, a planar rotation transform
was implemented to transform the measured angular velocity into the linear
accelerometer local reference frame. The force/moment transducer defined
the global reference frame and was stationary throughout impact. A second
NOCSAE medium headform was placed inside a Riddell VSR-4 football helmet
with polycarbonate shell, vinyl nitrile foam and steel facemask. The total
striking mass, including headform and helmet, was 6.7kg. This helmet was
positioned such that the center of gravity of the second headform was level
with the stationary headform at the moment of impact. The headform and
helmet were hung in an axial ‘spearing’ position from three light steel cables
167
attached to the impact frame cross members so that at impact the crown of
the headform inside the striking helmet was oriented normal to the struck
headform. A winch and electromagnet assembly raised the striking
headform-helmet mass to the desired swing height. In between each
impact, the swing height was confirmed and the struck and striking helmets
were returned to their initial position via use of an Optotrak Certus three-
dimensional motion monitoring system. Video was taken at 30fps for each
impact and each helmet was impacted six (6) times consecutively during
each impact condition. Approximately 90-120 seconds rest time was
provided in between impacts per NOCSAE standard.101 Figure 7-23 illustrates
the impact setup.
168
Figure 7-23. Football helmet power analysis impact setup alignment, lateral impact, oblique
impact, frontal impact, helmet positioning, overall view
The impact velocity was selected to mimic lower velocity youth football
impacts, well below that of prior NFL testing at 6-12m/s.165 Three impact
directions were analyzed based on the most common sites of concussive
169
injury from in-game analysis.122 While there were eleven (11) NOCSAE-
approved helmets available to test, a representative sampling of these
helmets was chosen due to time constraints. Hence, one modern helmet was
chosen from each of the four current helmet manufacturers and an authentic
game-worn ‘leatherhead’ helmet was tested as well. The ‘leatherhead’ was
included based on pilot testing results found in Chapter 6. The impact matrix
is shown in Table 7-2.
Swing Impact
Energy Momentum
HELMET MODELS TESTED ORIENTATION* Height Velocity
(J) (N-s)
(m) (m/s)
RIDDELL VSR-4, SCHUTT AIR
LATERAL,
ADVANTAGE, XENITH X1,
OBLIQUE, 0.20 2.0 13 13
ADAMS A2000,
FRONTAL
LEATHERHEAD
LATERAL,
OBLIQUE, 0.62 3.5 41 23
ADAMS A2000,
FRONTAL
LEATHERHEAD
LATERAL,
OBLIQUE, 1.28 5.0 84 33
ADAMS A2000,
FRONTAL
LEATHERHEAD
Table 7-2. Impact matrix for football helmet power analysis (*to avoid facemask contact,
frontal and oblique impacts used a 5° forward inclination)
For purposes of brevity, the 5.0m/s impacts are presented here in
Figures 7-24 through 7-26.
170
70 25
60
20
50

Linear Acceleration (g))
15
40
30
10
20
5
10
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
120 7000
6000
100

5000
FRONTAL: 5.0 m/s
80
4000 MODERN HELMETS

60 LEATHERHEAD
3000
40
2000
20
1000
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
Figure 7-24. ATD-measured linear acceleration, angular velocity, angular acceleration and
bending moment from frontal pendulum impacts at 84J (5.0m/s)
171
70 25
60
20
50

15
40
30
10
20
5
10
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
120 7000
6000
100

5000
OBLIQUE: 5.0 m/s
80
4000 MODERN HELMETS

60
LEATHERHEAD
3000
40
2000
20
1000
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
bending moment from oblique pendulum impacts at 84J (5.0m/s)
172
70 25
60
20
50

15
40
30
10
20
5
10
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
120 7000
6000
100

5000
80
LATERAL: 5.0 m/s
4000
MODERN HELMETS
60
LEATHERHEAD
3000
40
2000
20
1000
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (s) Time (s)
bending moment from lateral pendulum impacts at 84J (5.0m/s)
After collecting and collating the data from the 270 individual impact, a
power analysis was done for each helmet during each of the 45 impact
conditions (3 impact velocities, 3 impact locations, 5 helmets). For this
analysis, the maximum resultant linear acceleration, angular velocity,
angular acceleration, force and moment from the individual struck NOCSAE
headform impacts were compared to the mean of each set of six impacts for
each condition. The null hypothesis (Ho = mean of the maximum values
from each set of six impacts), alternate hypothesis (Ha deviates 10% from
Ho), significance level (a = 0.05) and power (P = 80%; Type II error risk =
20%) remained the same as was used in the cadaveric analysis. The football
helmet testing power analysis is shown in Figures 27 through 29.
173
2.0 m/s 3.5 m/s
Calculated Power
Calculated Power
0.6
0.6
0.4
0.4
0.2
Moment 0.2
0 Moment
0
Force
01 Force
01
02 Ang Accel
02 Ang Accel
09 Ang Vel
09 Ang Vel
16
Lin Accel 16
Helmet LB Lin Accel
Helmet LB
Calculated Power
0.6
0.4
0.2
Moment
0
Force
01
02 Ang Accel
09 Ang Vel
5.0 m/s 16
Lin Accel
Helmet LB
Figure 7-27. Lateral football helmet impact power analysis (73/75 conditions had P = 80% or
higher)
2.0 m/s 3.5 m/s

Calculated Power
Calculated Power
0.6 0.6
0.4 0.4
0.2 0.2
Moment Moment
0 0
Force Force
01 01
02 Ang Accel 02 Ang Accel
09 Ang Vel 09 Ang Vel

16 16
Lin Accel Lin Accel
Helmet LB Helmet LB
Calculated Power
0.6
0.4
0.2
Moment
0
Force
01
02 Ang Accel
09
5.0 m/s 16
Ang Vel
Lin Accel
Helmet LB
Figure 7-28. Oblique football helmet impact power analysis (71/75 conditions had P = 80% or
higher)
174
2.0 m/s 3.5 m/s
Calculated Power
Calculated Power
0.6 0.6
0.4 0.4
0.2 0.2
Moment Moment
0 0
Force Force
01 01
02 Ang Accel 02 Ang Accel
09 Ang Vel 09 Ang Vel

16 16
Lin Accel Lin Accel
Helmet LB Helmet LB
Calculated Power
0.6
0.4
0.2
Moment
0
Force
01
02 Ang Accel
09
5.0 m/s 16
Ang Vel
Lin Accel
Helmet LB
Figure 7-29. Frontal football helmet impact power analysis (67/75 conditions had P = 80% or
higher)
Thus, across all of the 45 impact conditions, and the five parameters
analyzed for this power analysis, a total of 210 out of 225 (93%) analysis
metrics achieved 80% power or greater. Of the 15 metrics failing to achieve
80% power, the majority occurred in frontal impacts with slight
facemask/upper boss contact.
As a comparison, the NFL test power analysis was done and is shown
in Figure 7-30.
175
NFL Power Analysis
Calculated Study Power 0.8
0.6
0.4
0.2
6.3
8.1
8.8
9.1
9.4
9.5
9.6 Lin Accel
10.3 Ang Accel
Impact Velocity (m/s) 10.7 GSI
10.8
Figure 7-30. NFL in-game reconstructed football helmet impact power analysis (13/30
conditions had P = 80% or higher)
Thus, the higher energy NFL tests achieved 80% power in 13 out of 30
(43%) metrics; the average number of impacts per condition needed to
achieve 80% power was n=7 for shell-to-shell and n=13 for facemask
impacts. This was contrasted with the lower energy groups of n=6 shell-to-
shell impacts studied here; power was 80% or higher for 5m/s impacts in 71
out of 75 (95%) cases. This indicated it was very much possible to conduct
higher energy helmeted impacts with high repeatability as long as the
impacts were done in a tightly controlled manner and a suitably powerful
sample size was gathered. Furthermore, this analysis might have provided a
rationale that some of the NFL impacts lacked sufficient power and required
additional impacts in order to render results statistically significant.
176
LIMITATIONS
There were several limitations in both the pig spine and helmeted ATD
impacts. For the pig spine testing, controlling specimen orientation proved to
be very challenging. Pre-impact orientation was held relatively constant but
post impact re-positioning required that the specimen occiput balance
precariously before the next impact. Oftentimes, the specimen fell out of
balance and tilted to one side which required aborting the impact and re-
positioning the occiput. Another issue with pig spine testing was that
precisely reproducing sensor insertion position and orientation was
impossible due to slight variations from specimen to specimen. It would
have been desirable to develop a jig that would hold the pig spine in place
and permit repeatable sensor insertion. Potting the specimen proved
challenging as well, and although the specimens were all prepared similarly,
slight variations in the cervico-thoracic potting material height could have
influenced the resulting caudal cervical injury in two of three specimens.
Finally, in the pig spine testing, there was no means to arrest the impactor
post-impact. On occasion, the impactor struck the specimen initially as
intended and then grazed the specimen on the backswing. Future
experiments should develop a catch mechanism to stop the impactor post-
impact.
For the helmeted ATD impacts the main limitation was that only a
single helmet of each type was tested. The intra-helmet impact attenuation
effects remain unquantified. Hence, the results shown here could have had
some artifact related to the specific helmet tested and that the true helmet
177
performance might have been different if a representative sampling of each
make and model were tested in a similar manner. Also, the striking helmet
was always the same Riddell VSR-4, and the potential performance
degradation in this striking helmet was unknown. In the future, a protocol
for the impacting interface lifespan should be developed in order to avoid any
hysteresis effects or degradation in the impact surface polycarbonate.
CONCLUSIONS
Impactor Design
The gravity-fed pendulum design drew on analysis from Chapter 6 and
was chosen for simplicity, high-throughput, versatility and low cost.
Ultimately, a few design iterations were needed to arrive at the final
structure. The test fixtures mounted inside the structure were designed for
maximum versatility and be swapped for either cadaveric or ATD tests.
Finally, the impactor vibration response and inertial effect of the fixture mass
on the load cell were studied. Frame vibration during impact contributed
negligible artifact to the sensor signals. The added fixture mass on top of the
load cell induced higher measured bending moments due to a larger moment
arm between load cell sensing face and perpendicular shear force. This
result played an important role in quantifying end condition loading for future
cadaveric experiments and required a certain scale factor being applied to
measured bending moments.
178
Pilot Cadaveric Experiments
The pilot cadaver experiments were conducted as a means to quantify
experiment power and attempt to determine the necessary specimens
needed in future experiments. A method to prepare, clamp, instrument and
impact the cadaver specimens, consisting of occiput through the second
thoracic vertebrae, was developed. While each specimen was injured at the
highest energy impact, little else can be ascertained from these experiments
due to the chaotic nature of the osteoligamentous specimen. The power
analysis indicated that for measuring linear and angular kinematics, a future
similar experiment may require in excess of 60 specimens to achieve
necessary statistical power. Hence, future work may benefit more from a
focus on functional spine unit testing in order to isolate the many variables
uncontrolled here. But these pilot experiments still did much to support the
theory that head and spine protection should be considered on a system,
rather than component, basis. As impact energy increased, linear and
rotational head and neck kinematics increased. Qualitatively, the peak head
linear acceleration as measured by sensors mounted to the skull base
increased at roughly the same rate as peak head angular velocity. When
examining the base of the spine, the shear force and bending moments
measured by the load cell seemed to also increase at approximately the
same rates. Thus, these data gave initial qualitative insight as to the benefit
of prioritizing head or neck protection versus protecting the system as a
whole. Thus, head and spine injury risk may be dependent upon end
conditions applied at the base of the neck as well as inputs to the head.
179
ATD Helmeted Impacts
ATD helmeted impacts were conducted with a modified NOCSAE
headform attached to a Hybrid III neck. A total of three impact orientations
(lateral, oblique, frontal) were chosen to go along with three impact
velocities (2.0m/s, 3.5m/s, 5.0m/s). Four modern helmets and one
‘leatherhead’ helmet were subjected to a total of n=6 impacts for each
condition and the power was calculated based on the collected data. From all
impacts and all data metrics analyzed, it was found that for these lower
energy impacts that 93% of the metrics achieved power of at least 80%.
This was contrasted with the high energy NFL impacts wherein 44% of
metrics achieved the necessary power. And while not specifically analyzed
here, the impact protection performance provided by the leatherhead helmet
merited further investigation. While not always performing on par with
modern helmets, in many instances the leatherhead performed equally well
or outperformed the modern helmets when comparing linear and rotational
kinematic and kinetic response.
180
SUMMARY
8 SUMMARY
In the first Chapter, inertial loading to the head and spine was
analyzed through a first-of-its-kind study on minor rear crashes and
‘whiplash’. In this small sampling of 105 crashes that were reconstructed
using physical principles of momentum, energy and restitution, key findings
were that nearly 80% of struck occupants visited a medical professional post-
crash while none of the striking occupants did the same, crash Delta-V for
struck vehicles was generally 4km/h to 8km/h, on par with sitting in a chair
or lying in bed and occupants with spinal degeneration were not at
significantly higher risk at sustaining injury than those without degeneration.
Finally, much was still unknown regarding quantification of the loading
experienced by the occupants in these crashes, so a means to estimate these
loads was pursued in Chapter 2.
Chapter 2 involved the application of real-world minor rear crash data
to a computational simulation model of a vehicle and BioRIDII crash test
dummy in the software MADYMO. The model was validated with BioRIDII
against real world data and the full body loading was quantified via virtual
sensors within the model. Injury risks for all body regions were all less than
10% of the appropriate injury tolerance data with the exception of the
proposed ‘whiplash’ neck protection criterion (Nkm). When these injury data
were compared with the diagnosed injuries in Chapter 1, it was found that
the dynamic loading in these crashes was insufficient to cause AIS2+ injury.
The model was further exercised with HybridIII 50th Percentile ATD and it
was found generally that the Hybrid III loading was slightly higher than
181
SUMMARY
BioRIDII subjected to the same exposure. Hence, it might be useful in rear
crash tests or simulations that use Hybrid III to consider that the loading
experienced by a more biofidelic surrogate – e.g., BioRIDII – would be lower
than that measured by HybridIII. Finally, the simulation showed low impact
head contact with the head restraint. As a means to study head-neck injury
further, direct contact injury was studied in Chapter 3 via re-analysis of the
original neck injury criterion test datasets.
In Chapter 3, the analysis of head-neck injury risk in rear crashes with
low energy inertial and direct impact loading to high energy frontal airbag
impacts was performed. The Nij, or neck injury criterion, was established in
2000 by the US Department of Transportation as a standard means to
quantify neck injury risk due to airbag contact. In this chapter, novel neck
axial force-based methods to predict neck injury were explored in an effort to
correctly discriminate between injured and uninjured subjects. Through use
of fundamental physical principles involving impulse, loading rate and
normalized force along with industry standards for injury rating through the
Abbreviated Injury Scale, a total of 14 potential injury metrics were created
with accuracy as high 95% in uninjured specimens and 100% for injured
specimens. This was contrasted with Nij accuracy of 78% and 65% for
uninjured and injured specimens, respectively.
The head-neck injury analysis from Chapter 3 was carried into Chapter
4 via extensive use of anthropomorphic test device (ATD) tests. And while
examining direct head impact from a swinging pendulum, as opposed to
airbag tests in Chapter 3, the data collected in Chapter 4 helped greatly to
182
SUMMARY
quantify the Hybrid III ATD response to frontal, oblique and lateral impacts.
These impacts were of the type seen in automotive crashes as well as
athletics, but the ATD response had never been quantified to direct impact in
absence of compression. It was found that the lateral direction provided the
stiffest response, followed by the oblique and frontal directions respectively.
This was an important result as the Hybrid III was originally only validated
for frontal impact loading but seemed to provide reasonable biofidelity in
response to off-axis loading. Armed with this knowledge, athletic head
impacts were conducted with the Hybrid III in Chapter 5.
Chapter 5 was a true transition from the motor vehicle crash-centric
Chapters 1-3 into studying direct head impacts in athletics. In this Chapter,
pugilistic head strikes were recreated using a weighted pendulum and several
protective padding conditions were studied. Based on the results from this
chapter, it was seen that certain head and neck kinematic and kinetic impact
responses were always minimized by use of padding on both the impactor
and ATD. But heavier padding on the impactor, such as using a boxing glove
versus a mixed-martial-arts mitt, could increase kinematics and kinetics
when compared with lighter padding. And in certain instances, use of
impactor padding against a bare headform caused comparable or higher
angular velocity/acceleration for the head as a control impact without any
padding. These results were also analyzed via the SIMon finite element brain
model and some brain injury risk parameters when using padding actually
increased diffuse axonal injury (DAI) risk over a zero-padding condition.
Essentially, head and neck injury protection due to direct head impact was
183
SUMMARY
found to be a complicated dynamic problem, and required further
investigation in Chapters 6 via simulation and Chapter 7 via experimentation.
Chapter 6 saw use of Simulink and Matlab to validate a computational
system dynamics model versus real-world pilot helmeted head impact data.
Many impacting systems were analyzed with the express requirements of the
system being simple, able to accommodate a head-neck specimen (cadaveric
and ATD), cost effective and able to conduct experiments with high
throughput. Currently used drop systems and pneumatic linear impactors
were modeled in an effort to understand the intrinsic differences between the
systems. The results were that a simple gravity-fed pendulum could provide
similar versatility and capability as systems currently used by NOCSAE and
the NFL, but cost less and provided more fidelity at lower impact velocities.
This result was crucial in the development of a gravity fed head-neck impact
pendulum in Chapter 7.
In Chapter 7 all of the head-neck injury concepts were tied together.
The pendulum impactor system analyzed numerically in Chapter 6, was built,
validated and put through cadaveric and crash dummy testing. The
cadaveric data, though taken from a very chaotic system, pointed to
continued need to investigate head and spine injury concomitantly. And
these data also hinted that the boundary conditions at the base of the neck,
as well as direct inputs to the head, influence head-neck loading. The helmet
tests and power analysis indicated that n=6 impacts done under controlled
conditions at impact velocity from 2.0m/s to 5.0m/s produced data with
184
SUMMARY
sufficient power 93% of the time. Of note was that prior high energy NFL
helmet testing achieved the same in 44% of cases.
The findings from this dissertation could prove useful in aiding design
of personal athletic protection equipment, automotive injury mitigation tools
or in injury criteria to prevent impact injury. The potential significance of this
work was a clearer understanding of the injury mechanisms and impact
dynamics responsible for head and cervical spine injury related to direct head
impact or inertial loading as well as the creation of novel injury metrics for
the prediction of dynamic head and cervical spine injury.
185
APPENDIX
9 APPENDIX
A-1 Impact Duration:
Impact duration was selected based on the length of the loading and unloading pulse after
initial contact as based on impact force.
A-2 Resultant Head CG Delta-V:
Delta - VR t    a~R t dt
A-3 Resultant Head CG Acceleration:
~
aR t   ~aCGX t 2  ~aCGY t 2  ~aCGZ t 2
where
~
aCGX t   measured head cg acceleration along x - axis
~
aCGY t   measured head cg acceleration along y - axis
~
aCGZ t   measured head cg acceleration along z - axis
A-4 Resultant Head CG Jerk:
~
jR t  
d ~
dt

aR t  
A-5 Resultant Occipital Force:
~ ~ 2 2 2
FoccR t    F t    F~occY t    F~occZ t 
 occX 
where
~
FoccX t   measured occipital force along x - axis
~
FoccY t   measured occipital force along y - axis
~
FoccZ t   measured occipital force along z - axis
A-6 Resultant Linear Head CG Momentum:
~
pR t   mheadDelta - VR t 
where
mhead  5.08kg
A-7 Resultant Impact Force:
186
APPENDIX
~ ~ 2 2 2
FIR t    F t    F~IY t    F~IZ t 
 IX 
where
~ ~ ~
FIX t   mheadaCGX t   FoccX t 
~ ~ ~
FIY t   mheadaCGY t   FoccY t 
~ ~ ~
FIZ t   mheadaCGZ t   FoccZ t 
A-8 Maximum Resultant Impact Force Loading Rate:
d ~
Maximum Resultant Impact Force Loading Rate  max   F t 
 IR 
 dt
A-9 Minimum Resultant Impact Force Loading Rate:
d ~
Minimum Resultant Impact Force Loading Rate  min  F t 
 IR 
 dt
A-10 Head Kinetic Energy:
1
KEhead t   mhead Delta - VR t 
2
2
A-11 Head Power:
d
P t   KEhead t 
dt
A-12 Gadd Severity Index (GSI):
T
~
 a t 
2.5
GSI  R dt
0
where
~
aR t  in g
0 : T   essential impact duration, chosen to be 15 msec
A-13 Head Injury Criterion (HIC):
2.5
 t2 
1 ~
HIC  t2  t1  aR t dt 
 t2  t1  
 t1 
where
~
aR t  in g
t1 : t2   time period, in msec, where HIC is maximized
A-14 HIC Duration:
HIC Duration  t2  t1
A-15 Skull Fracture Correlate (SFC):
187
APPENDIX
 maxDelta - VR t   1
SFC   
 HIC Duration  g
m
g  9.81 2
s
A-16 Resultant Head Angular Velocity:
~ t  
 R ~X t 2  ~Y t 2  ~Z t 2
where
~ t   measured head angular velocity about x - axis
 X
~ t   measured head angular velocity about y - axis
Y
~ t   measured head angular velocity about z - axis
 Z
A-17 Resultant Head Angular Acceleration:
~ t  
 R ~X t 2  ~Y t 2  ~Z t 2
where
~ t   measured head angular acceleration about x - axis
 X
~ t   measured head angular acceleration about y - axis
Y
~ t   measured head angular acceleration about z - axis
 Z
A-18 Resultant Head Angular Jerk:
 R t  
d ~
dt

 R t 
A-19 Resultant Occipital Moment:
~ ~ 2 ~ 2 2
MoccR t   t    M ~
occY t    MoccZ t 
 M  
 occX    
where
~
MoccX t   measured occipital moment about x - axis
~
MoccY t   measured occipital moment about y - axis
~
MoccZ t   measured occipital moment about z - axis
A-20 Resultant Head Angular Momentum:
R t  
~ I X ~X t 2  IY ~Y t 2  IZ~Z t 2
where
I X  head mass moment of inertia about x - axis  0.0204kg - m2
IY  head mass moment of inertia about y - axis  0.0211kg - m2
I Z  head mass moment of inertia about z - axis  0.0143kg - m2
The combined group comprised the remaining eight parameters: Neck Injury Criterion (Nij)40,
Generalized Acceleration Model for Brain Injury Threshold (GAMBIT)103, weighted Principal
Component Score (wPCS)57, Head Impact Power (HIP)105, Cumulative Strain Damage Measure
188
APPENDIX
(CSDM-0.05, 0.10 and 0.15 levels)151, Dilatation Damage Measure (DDM)151 and Relative
Motion Damage Measure (RMDM)151
A-21 Neck Injury Criterion (Nij)
~
F
Nij  max  occZ
t    F~occY t   d  M
~ 
occX  
 FZcrit  M Xcrit 
  
where
FZcrit  critical occipital axial force  6806N
M Xcrit  critical occipital moment  135N  m
d  0.01778m
A-22 Generalized Acceleration Model for Brain Injury Threshold (GAMBIT):
1
 ~ 2.5 ~ t  2.5  2.5
aR t   
GAMBIT t    ~   ~R  
 
 aC    C 
  
where
~
aC  critical head linear acceleration  250g
~
  critical head angular acceleration  25,000rad/s2
C
A-23 Weighted Principal Component Score (wPCS):
  GSI  GSI m   HIC  HIC m   

 k    k HIC    
 GSI  GSI sd   HIC sd  
    
wPCS  k lat  10    2
  max ~
 k LIN 
aR t  
a 
m  
 k ROT 
max  R 
~ t   
m 



  a    
  sd   sd  
where
k lat  1.0 for lateral impact
k GSI  0.4718, GSI m  25.83, GSI sd  75.20
k HIC  0.4720, HIC m  17.65, HIC sd  51.95
k LIN  0.4336, am  26.24g, asd  21.77g
k ROT  0.2164,  m  1537rad / s 2 ,  sd  1432rad / s 2 ,
A-24 Head Impact Power (HIP):
~ ~ ~ ~ ~
HIP  mhead aCGX t  aCGX t dt  aCGY t 
   a t dt  a
CGY CGZ t  ~
aCGZ t dt  


~ t  
~ t   I 
~ t  
~ t   I 
~ t 
  t 
~
I  X X  X Y Y  Y Z Z Z
A-25 Cumulative Strain Damage Measure (CSDM05):
CSDM 05  % Brain volume exceeding 5% principal strain as determined by SIMon Finite Element Model
CSDM10  % Brain volume exceeding 10% principal strain as determined by SIMon Finite Element Model
189
APPENDIX
t
CSDM15  % Brain volume exceeding 15% principal strain as determined by SIMon Finite Element Model
A-28 Dilatation Damage Measure (DDM):
DDM  % Brain volume experiencing negative pressure less than - 101.4 kPa as determined by SIMon Finite Elemen
A-29 Relative Motion Damage Measure (RMDM):
 t 
RMDM 
 F t ,  t 
where
 t   bridging vein strain as determined by SIMon Finite Element Model
 F t ,  t   bridging vein failure strain at given strain rate as determined by SIMon Finite Element Model
190
BIBLIOGRAPHY
10 BIBLIOGRAPHY
1. Agaram V, Kang J, Nusholtz GS et al. Hybrid III dummy neck response
to airbag loading. Seventeenth International Technical Conference on
Experimental Safety Vehicles Paper No. 469, 1-10. 2001. Amsterdam,
The Netherlands, National Highway Traffic Safety Administration.
2. Association for the Advancement of Automotive Medicine. Abbreviated
injury scale 1980. 1980. Barrington, Illinois, USA, Association for the
Advancement of Automotive Medicine.
3. Association for the Advancement of Automotive Medicine. Abbreviated
injury scale 2005. Gennarelli, T. A. and Wodzin, E. 2005. Barrington,
Illinois, USA, Association for the Advancement of Automotive Medicine.
4. ASTM International. Standard test methods for equipment and
procedures used in evaluating the performance characteristics of
protective headgear. F1446-01a, 1-10. 2009. West Conshohocken, PA.
5. Bailey MN, Wong BC, Lawrence JM. Data and methods for estimating
the severity of minor impacts. Society of Automotive Engineers World
Congress (SAE Technical Paper 950352), 139-175. 1995.
6. Bartsch AJ, Gilbertson LG, Prakash V et al. Minor crashes and
'whiplash' in the United States. Annu.Proc.Assoc.Adv.Automot.Med.
2008;52:117-28.
191
BIBLIOGRAPHY
7. Beckwith JG, Chu JJ, Greenwald RM. Validation of a noninvasive
system for measuring head acceleration for use during boxing
competition. J.Appl.Biomech. 2007;23:238-44.
8. Berglund A, Alfredsson L, Jensen I et al. Occupant- and crash-related
factors associated with the risk of whiplash injury. Ann.Epidemiol.
2003;13:66-72.
9. Beusenberg MC, Shewchenko N, Newman JA et al. Head, Neck and
Body Coupling in Reconstructions of Helmeted Head Impacts.
International Research Council on Biomechanics of Impact (IRCOBI).
2001. Isle of Man (UK).
10. Bostrom O, Svensson M, Aldman B et al. A new neck injury criterion
candidate - based on injury findings in the cervical spinal ganglia after
experimental neck extension trauma. International Research Council
on the Biomechanics of Injury (IRCOBI) Annual Conference. 123-136.
1996. Dublin, Ireland.
11. Brault JR, Siegmund GP, Wheeler JB. Cervical muscle response during
whiplash: evidence of a lengthening muscle contraction.
Clin.Biomech.(Bristol., Avon.) 2000;15:426-35.
12. Broglio SP, Ju YY, Broglio MD et al. The Efficacy of Soccer Headgear.
J.Athl.Train. 2003;38:220-4.
J.Athl.Train. 2003;38:220-4.
192
BIBLIOGRAPHY
J.Athl.Train. 2003;38:220-4.
15. Bunketorp L, Stener-Victorin E, Carlsson J. Neck pain and disability
following motor vehicle accidents--a cohort study. Eur.Spine J.
2005;14:84-9.
16. Camacho DL, Nightingale RW, Robinette JJ et al. Experimental
flexibility measurements for the development of a computational head-
neck model validated for near-vertex head impact. Stapp Car Crash
Journal 1997;41st Proceedings:473-86.
17. Campbell KL. Energy basis for collision severity. Society of Automotive
Engineers World Congress (SAE Technical Paper 740565). 1974.
18. Carroll S, Abrahamse A, Vaiana M. The costs of excessive medical
claims for automobile personal injuries. 1995. Santa Monica,
California, RAND Institute for Civil Justice.
19. Carter JW, Ku GS, Nuckley DJ et al. Tolerance of the cervical spine to
eccentric axial compression. Stapp Car Crash J. 2002;46:441-59.
20. Cassidy JD, Carroll LJ, Cote P et al. Effect of eliminating compensation
for pain and suffering on the outcome of insurance claims for whiplash
injury. N.Engl.J.Med. 2000;342:1179-86.
193
BIBLIOGRAPHY
21. Casson IR, Pellman EJ, Viano DC. Concussion in the National Football
League: an overview for neurologists. Phys.Med.Rehabil.Clin.N.Am.
2009;20:195-214, x.
22. Casson IR, Pellman EJ, Viano DC. Concussion in the National Football
League: an overview for neurologists. Phys.Med.Rehabil.Clin.N.Am.
2009;20:195-214, x.
23. Caswell SV, Deivert RG. Lacrosse Helmet Designs and the Effects of
Impact Forces. J.Athl.Train. 2002;37:164-71.
24. Chang H, Gilbertson LG, Goel VK et al. Dynamic response of the
occipito-atlanto-axial (C0-C1-C2) complex in right axial rotation.
J.Orthop.Res. 1992;10:446-53.
25. Chang H, Gilbertson LG, Goel VK et al. Dynamic response of the
occipito-atlanto-axial (C0-C1-C2) complex in right axial rotation.
J.Orthop.Res. 1992;10:446-53.
26. Chapline JF, Ferguson SA, Lillis RP et al. Neck pain and head restraint
position relative to the driver's head in rear-end collisions.
Accid.Anal.Prev. 2000;32:287-97.
27. Chichester C, Bass CD, Boggess B et al. A test methodology for
assessing demining personal protective equipment (PPE). Mine I, 1-74.
2001. United States Army-CECOM.
194
BIBLIOGRAPHY
28. Ching RP, Nuckley DJ, Hertsted SM et al. Tensile mechanics of the
developing cervical spine. Stapp Car Crash J. 2001;45:329-36.
29. Cipriani AL, Bayan FP, Woodhouse ML et al. Low speed collinear impact
severity: a comparison between full scale testing and analytical
prediction tools with restitution analysis. Society of Automotive
Engineers World Congress (SAE Technical Paper 2002-01-0540), 23-
37. 2002.
30. Clemens HJ, Burow K. Experimental investigation on injury
mechanisms of cervical spine at frontal and rear-front vehicle impacts.
Stapp Car Crash Journal 1972;16th Proceedings:76-104.
31. Croft AC, Haneline MT, Freeman MD. Low speed frontal crashes and
low speed rear crashes: is there a differential risk for injury?
Annu.Proc.Assoc.Adv.Automot.Med. 2002;46:79-91.
32. De Jager M, Sauren A, Thunnissen J et al. Discrete-parameter head-
neck models for impact. Frontiers in Head and Neck Trauma.
1998:454-68.
33. Dibb AT, Nightingale RW, Chancey VC et al. Comparative structural
neck responses of the THOR-NT, Hybrid III, and human in combined
tension-bending and pure bending. Stapp Car Crash J. 2006;50:567-
81.
34. Dibb AT, Nightingale RW, Chancey VC et al. Comparative structural
neck responses of the THOR-NT, Hybrid III, and human in combined
195
BIBLIOGRAPHY
tension-bending and pure bending. Stapp Car Crash J. 2006;50:567-
81.
35. DiDomenico L., Nusholtz G. Risk curve boundaries. Traffic.Inj.Prev.
2005;6:86-94.
36. Doeblin EO. Mechanical Engineering 687 Course Notes. The Ohio State
University Department of Mechanical Engineering. 2000.
37. Duma SM. Analysis of field data regarding concussive head impact in
football. North American Congress on Biomechanics (NACOB) Head
Injury Symposium. 2008.
38. Duma SM, Manoogian SJ, Bussone WR et al. Analysis of real-time head
accelerations in collegiate football players. Clin.J.Sport Med.
2005;15:3-8.
39. Eis V, Sferco R, Fay P. A Detailed Analysis of the Characteristics of
European Rear Impacts. The 19th Technical Conference on the
Enhanced Safety of Vehicles (ESV) . 2005. Washington, D.C., USA,
National Highway Traffic Safety Administration.
40. Eppinger R, Sun E, Kuppa S et al. Supplement: development of
improved injury criteria for the assessment of advanced automotive
restraint systems-II. SES-1-SA-11. 2000. Washington, D.C., National
Highway Traffic Safety Administration.
196
BIBLIOGRAPHY
41. Finan JD, Nightingale RW, Myers BS. The influence of reduced friction
on head injury metrics in helmeted head impacts. Traffic.Inj.Prev.
2008;9:483-8.
42. Forbes PA, van Rooij L, Philippens M et al. Occupant Safety in Freefall
Lifeboats: Full Scale Drop Testing with Instrumented Dummy.
International Conference of Crashworthiness (ICRASH). 2008. Kyoto,
Japan.
43. Frechede B, McIntosh AS. Numerical reconstruction of real-life
concussive football impacts. Med.Sci.Sports Exerc. 2009;41:390-8.
44. Funk JR, Cormier JM, Bain CE et al. Validation and application of a
methodology to calculate head accelerations and neck loading in
soccer ball impacts. 2009 SAE World Congress Paper Number 2009-
01-0251. 2009. Society of Automotive Engineers.
45. Funk JR, Cormier JM, Bain CE et al. An evaluation of various neck
injury criteria in vigorous activities. International Research Council on
the Biomechanics of Injury (IRCOBI) Annual Conference. 233-248.
2007. Maastricht, The Netherlands.
46. Gadd CW. Use of weighted-impulse criterion for establishing injury
hazard. Stapp.Car.Crash.J.-Proceedings of 10th Stapp Car Crash
Conference 1966;SAE Paper Number 660793:164-74.
47. Gennarelli TA. Mechanisms of brain injury. J.Emerg.Med. 1993;11
Suppl 1:5-11.
197
BIBLIOGRAPHY
48. Gennarelli TA, Adams JH, Graham DI. Acceleration induced head injury
in the monkey.I. The model, its mechanical and physiological
correlates. Acta Neuropathol.Suppl 1981;7:23-5.
49. Gennarelli TA, Thibault LE. Biomechanics of acute subdural hematoma.
J.Trauma 1982;22:680-6.
50. Gennarelli TA, Thibault LE, Adams JH et al. Diffuse axonal injury and
traumatic coma in the primate. Ann.Neurol. 1982;12:564-74.
51. Gibson T, Bogduk N, MacPherson J et al. Crash characteristics of
whiplash associated chronic neck pain. Journal of Musculoskeletal Pain
2000;8:87-95.
52. Gilchrist A, Mills NJ. Protection of the side of the head. Accid.Anal.Prev.
1996;28:525-35.
53. Goel VK, Weinstein JN. Biomechanics of the Spine: Clinical and
Surgical Perspective. Boca Raton, FL: CRC Press, 1990.
54. Goel VK, Winterbottom JM, Schulte KR et al. Ligamentous laxity across
C0-C1-C2 complex. Axial torque-rotation characteristics until failure.
Spine (Phila Pa 1976.) 1990;15:990-6.
55. Got C, Patel A, Fayon A et al. Results of experimental head impacts on
cadavers: The various data obtained and their relations to some
measured physical parameters. Stapp Car Crash Journal 1978;22nd
Proceedings:441-82.
198
BIBLIOGRAPHY
56. Green T, Barth T. Injury evaluation and comparison of lateral impacts
when using conventional and inflatable restraints. SAFE Symposium.
1-16. 2006. Reno, NV.
57. Greenwald RM, Gwin JT, Chu JJ et al. Head impact severity measures
for evaluating mild traumatic brain injury risk exposure. Neurosurgery
2008;62:789-98.
58. Gurdjian ES, Lissner HR, Hodgson VR et al. Mechanism of head injury.
Clin.Neurosurg. 1964;12:112-28.
59. Gurdjian ES, Lissner HR, WEBSTER JE. The mechanism of production
of linear skull fracture; further studies on deformation of the skull by
the stresscoat technique. Surg.Gynecol.Obstet. 1947;85:195-210.
60. Gurdjian ES, Webster J, Lissner H. Observations on the mechanism of
brain concussion, contusion, and laceration. Surg.Gynecol.Obstet.
1955;101:680-90.
61. Happer AJ, Hughes MC, Peck MD et al. Practical analysis methodology
for low speed vehicle collisions involving vehicles with modern bumper
systems. Society of Automotive Engineers World Congress (SAE
Technical Paper 2003-01-0492). 2003.
62. Hell W, Schick S, Langwieder K et al. Biomechanics of cervical spine
injuries in rear end car impacts: influence of car seats and possible
evaluation criteria. Traffic Inj.Prev. 2002;3:127-40.
199
BIBLIOGRAPHY
63. Hertz E. A note on the head injury criterion (HIC) as a predictor of the
risk of skull fracture. 37th Annual Proceedings of the Association for
the Advancement of Automotive Medicine (1993) . 1993.
64. Hodgson VR. National Operating Committee on Standards for Athletic
Equipment football helmet certification program. Med.Sci.Sports
1975;7:225-32.
65. Hodgson VR, Thomas LM. Breaking strength of the skull vs. impact
surface curvature. DOT HS 800 583. 1971. United States Department
of Transportation.
66. Holbourn AHS. Mechanics of head injuries. Lancet 1943;ii:438-41.
67. Hrysomallis C. Impact energy attentuation of protective football
headgear against a yielding surface. J.Sci.Med.Sport 2004;7:156-64.
71. Hu AS, Bean SP, Zimmerman RM. Response of belted dummy and
cadaver to rear impact. Stapp Car Crash Journal 1977;21st
Proceedings:589-625.
200
BIBLIOGRAPHY
72. Jakobsson L, Lundell B, Norin H et al. WHIPS--Volvo's Whiplash
Protection Study. Accid.Anal.Prev. 2000;32:307-19.
73. Kahane CJ. An Evaluation of Head Restraints Federal Motor Vehicle
Safety Standard 202. DOT HS 806 108. 1982. National Highway
Traffic Safety Administration.
74. Kang J, Agaram V, Nusholtz GS et al. Air bag loading on in-position
Hybrid III dummy neck. Society of Automotive Engineers 2001 World
Congress SAE Paper Number 2001-01-0179. 2001. Warrendale, PA,
Society of Automotive Engineers.
75. King AI, Yang KH, Zhang L et al. Is head injury caused by linear or
angular acceleration? International Research Council on the
Biomechanics of Injury. 1-12. 2003. Lisbon, Portugal.
76. Knouse CL, Gould TE, Caswell SV et al. Efficacy of Rugby Headgear in
Attenuating Repetitive Linear Impact Forces. J.Athl.Train.
2003;38:330-5.
77. Krafft M, Kullgren A, Malm S et al. Influence of crash severity on
various whiplash injury symptoms: a study based on real-life rear-end
crashes with recorded crash pulses. The 19th Technical Conference on
the Enhanced Safety of Vehicles (ESV) . 2005. Washington, D.C., USA,
National Highway Traffic Safety Administration.
78. Kumar S, Ferrari R, Narayan Y. Cervical muscle response to head
rotation in whiplash-type left lateral impacts. Spine 2005;30:536-41.
201
BIBLIOGRAPHY
79. Kuppa SM. Injury criteria and anthropomorphic test devices for
whiplash injury assessment. 2004. National Highway Traffic Safety
Administration.
80. Lewis LM, Naunheim R, Standeven J et al. Do football helmets reduce
acceleration of impact in blunt head injuries? Acad.Emerg.Med.
2001;8:604-9.
81. Lissner H, Lebow M, Evans F. Experimental studies on the relation
between acceleration and intracranial pressure changes in man.
Surg.Gynecol.Obstet. 1960;111:329-38.
82. Lowenhielm P. Mathematical simulation of gliding contusions.
J.Biomech. 1975;8:351-6.
83. Lysell E. Motion in the cervical spine. An experimental study on
autopsy specimens. Acta Orthop.Scand. 1969;Suppl.
84. Manoogian S, McNeely D, Duma S et al. Head acceleration is less than
10 percent of helmet acceleration in football impacts.
Biomed.Sci.Instrum. 2006;42:383-8.
85. Margulies SS, Thibault LE. A proposed tolerance criterion for diffuse
axonal injury in man. Journal of Biomechanics 1992;25:917-23.
86. Marjoux D, Baumgartner D, Deck C et al. Head injury prediction
capability of the HIC, HIP, SIMon and ULP criteria. Accid.Anal.Prev.
2008;40:1135-48.
202
BIBLIOGRAPHY
87. McIntosh A, McCrory P, Finch CF. Performance enhanced headgear: a
scientific approach to the development of protective headgear.
Br.J.Sports Med. 2004;38:46-9.
88. McIntosh AS, Kallieris D, Frechede B. Neck injury tolerance under
inertial loads in side impacts. Accid.Anal.Prev. 2007;39:326-33.
89. McIntosh AS, McCrory P. Impact energy attenuation performance of
football headgear. Br.J.Sports Med. 2000;34:337-41.
90. McLean AJ. Brain injury without head impact? J.Neurotrauma
1995;12:621-5.
91. Mertz HJ, Patrick LM. Strength and response of the human neck.
Stapp.Car.Crash.J. 1971;15:207-55.
92. Mertz HJ, Patrick LM. Investigation of the kinematics and kinetics of
whiplash. Stapp.Car.Crash.J. 1967;11:267-317.
93. Mertz HJ, Prasad P. Improved neck injury risk curves for tension and
extension moment measurements of crash dummies.
Stapp.Car.Crash.J. 2000;44:59-75.
94. Mertz HJ, Weber DA. Interpretations of the impact responses of a 3-
year-old child dummy relative to child injury potential. Ninth
International Technical Conference on Experimental Safety Vehicles
SAE Paper Number 826048. 1982. Warrendale, PA, Society of
Automotive Engineers.
203
BIBLIOGRAPHY
95. Minton R, Murray P, Stephenson W et al. Whiplash injury--are current
head restraints doing their job? Accid.Anal.Prev. 2000;32:177-85.
96. Moss RT, Bardas AM, Hughes MC et al. Injury symptom risk curves for
occupants involved in rear end low speed motor vehicle collisions.
Society of Automotive Engineers. Society of Automotive Engineers
2005 World Congress (Scientific Paper Number 2005-01-0296). 2005.
Warrendale, PA, USA.
97. Myers BS, McElhaney JH, Doherty BJ et al. The role of torsion in
cervical spine trauma. Spine (Phila Pa 1976.) 1991;16:870-4.
98. Nahum AM, Gatts JD, Gadd CW et al. Impact tolerance of the skull and
face. Stapp Car Crash Journal 1968;12th Proceedings:302-16.
99. National Highway Traffic Safety Administration. 49 CFR Part 571,
Federal Motor Vehicle Safety Standards Final rule-Head Restraints.
Docket No. NHTSA-2004-19807, RIN 2127-AH09. 2004. Department
of Transportation.
100. National Highway Traffic Safety Administration. Federal Motor Vehicle
Safety Standards 208-Occupant Crash Protection. 49CFR571.208.
2008.
101. National Operating Committee on Standards for Athletic Equipment
(NOCSAE). Standard test method and equipment used in evaluating
the performance characteristics of protective headgear/equipment.
NOCSAE DOC (ND) 001-08M08B, 1-26. 2008. Overland Park, KS.
204
BIBLIOGRAPHY
102. Naunheim RS, Ryden A, Standeven J et al. Does soccer headgear
attenuate the impact when heading a soccer ball? Acad.Emerg.Med.
2003;10:85-90.
103. Newman JA. A generalized model for brain injury threshold (GAMBIT).
International Conference on the Biomechanics of Impact (IRCOBI).
1986.
104. Newman JA, Beusenberg MC, Shewchenko N et al. Verification of
biomechanical methods employed in a comprehensive study of mild
traumatic brain injury and the effectiveness of American football
helmets. J.Biomech. 2005;38:1469-81.
105. Newman JA, Shewchenko N, Welbourne E. A proposed new
biomechanical head injury assessment function-the maximum power
index. Society of Automotive Engineers. 44th Stapp Car Crash
Conference. 2000-01-SC16. 2000.
106. Niehoff P, Gabler HC. The accuracy of winsmash delta-v estimates: the
influence of vehicle type, stiffness, and impact mode.
107. Nightingale RW, Camacho DL, Armstrong AJ et al. Inertial properties
and loading rates affect buckling modes and injury mechanisms in the
cervical spine. J.Biomech. 2000;33:191-7.
205
BIBLIOGRAPHY
108. Nightingale RW, Carol C, V, Ottaviano D et al. Flexion and extension
structural properties and strengths for male cervical spine segments.
J.Biomech. 2007;40:535-42.
109. Nightingale RW, Chancey VC, Luck JF et al. The human cervical spine
in tension: effects of frame and fixation compliance on structural
responses. Traffic.Inj.Prev. 2004;5:151-5.
110. Nightingale RW, McElhaney JH, Richardson WJ et al. Dynamic
responses of the head and cervical spine to axial impact loading.
J.Biomech. 1996;29:307-18.
111. Nightingale RW, Winkelstein BA, Knaub KE et al. Comparative
strengths and structural properties of the upper and lower cervical
spine in flexion and extension. J.Biomech. 2002;35:725-32.
112. Nuckley DJ, Van Nausdle JA, Eck MP et al. Neural space and
biomechanical integrity of the developing cervical spine in
compression. Spine 2007;32:E181-E187.
113. Nusholtz GS, Di DL, Shi Y et al. Studies of neck injury criteria based on
existing biomechanical test data. Accid.Anal.Prev. 2003;35:777-86.
114. Ommaya AK, Goldsmith W, Thibault L. Biomechanics and
neuropathology of adult and paediatric head injury. Br.J.Neurosurg.
2002;16:220-42.
206
BIBLIOGRAPHY
115. Ommaya AK, Hirsch AE. Tolerances for cerebral concussion from head
impact and whiplash in primates. J.Biomech. 1971;4:13-21.
116. Ommaya AK, Hirsch AE, Martinez JL. The role of whiplash in cerebral
concussion. Proc.10th Stapp Car Crash Conference SAE Paper No.
660804. 1966.
117. Ono K, Kanno M. Influences of the physical parameters on the risk to
neck injuries in low impact speed rear-end collisions. Accid.Anal.Prev.
1996;28:493-9.
118. Ono K, Kikuchi A, Nakamura M et al. Human Head Tolerance to
Sagittal Impact: Reliable Estimation deduced from Experimental Head
Injury using Subhuman Primates and Human Cadaver Skulls.
Proc.24th Stapp Car Crash Conference (SAE Paper No. 801302.).
1980.
119. Panjabi M, Dvorak J, Duranceau J et al. Three-dimensional movements
of the upper cervical spine. Spine (Phila Pa 1976.) 1988;13:726-30.
120. Panjabi MM, Crisco JJ, Vasavada A et al. Mechanical properties of the
human cervical spine as shown by three-dimensional load-
displacement curves. Spine (Phila Pa 1976.) 2001;26:2692-700.
121. Patrick LM, Chou CC. Response of the human neck in flexion,
extension and lateral flexion. Vehicle Research Institute Report No.
VRI-7.3, 1-172. 1976. Society of Automotive Engineers.
207
BIBLIOGRAPHY
122. Pellman EJ, Viano DC, Tucker AM et al. Concussion in professional
football: reconstruction of game impacts and injuries. Neurosurgery
2003;53:799-812.
123. Pellman EJ, Viano DC, Withnall C et al. Concussion in professional
football: helmet testing to assess impact performance--part 11.
Neurosurgery 2006;58:78-96.
124. Pincemaille Y, Trosseille X, Mack P et al. Some new data related to
human tolerance obtained from volunteer boxers. 33rd Stapp Car
Crash Conference. (SAE Paper 892435), 177-190. 1989. Warrendale,
PA, Society of Automotive Engineers.
125. Pintar FA, Yoganandan N, Baisden J. Characterizing occipital condyle
loads under high-speed head rotation. Stapp Car Crash J. 2005;49:33-
47.
126. Pintar FA, Yoganandan N, Pesigan M et al. Cervical vertebral strain
measurements under axial and eccentric loading. J.Biomech.Eng
1995;117:474-8.
127. Prasad P, Daniel R. A biomechanical analysis of head, neck and torso
injuries to child surrogates due to sudden torso acceleration. 28th
Stapp Car Crash Conference Journal SAE Paper Number 841656. 1984.
Warrendale, PA, Society of Automotive Engineers.
208
BIBLIOGRAPHY
128. Ronnen HR, de Korte PJ, Brink PR et al. Acute whiplash injury: is there
a role for MR imaging?--a prospective study of 100 patients. Radiology
1996;201:93-6.
129. Rupp JD, Reed MP, Van Ee CA et al. The tolerance of the human hip to
dynamic knee loading. Stapp Car Crash J. 2002;46:211-28.
130. Schmitt KU, Muser M, Niederer P. A new neck injury criterion candidate
for rear-end collisions taking into account shear forces and bending
moments. Society of Automotive Engineers. 17th International
Technical Conference on the Enhanced Safety of Vehicles (ESV). 1-9.
2001. Amsterdam, The Netherlands.
131. Schmitt KU, Muser M, Vetter D et al. Biomechanical assessment of soft
tissue neck injuries in cases with long sick leave times. Traffic.Inj.Prev.
2003;4:162-8.
132. Schneider RC. Serious and fatal neurosurgical football injuries.
Clin.Neurosurg. 1964;12:226-36.
133. Schneider RC, REIFEL E, CRISLER HO et al. Serious and fatal football
injuries involving the head and spinal cord. JAMA 1961;177:362-7.
134. Schuller E, Eisenmenger W, Beier G. Whiplash injury in low speed car
accidents: Assessment of biomechanical cervical spine loading and
injury prevention in a forensic sample. Journal of Musculoskeletal Pain
2000;8:55-67.
209
BIBLIOGRAPHY
135. Seeman MR, Muzzy WH, Lustick LS. Comparison of human and hybrid
III head and neck dynamic response. Stapp Car Crash Journal
1986;30th Proceedings:291-311.
136. Shea M, Edwards WT, White AA et al. Variations of stiffness and
strength along the human cervical spine. J.Biomech. 1991;24:95-107.
137. Siegmund GP, King DJ, Montgomery DT. Using barrier impact data to
determine speed change in aligned, low-speed vehicle-to-vehicle
collisions. Society of Automotive Engineers World Congress (Paper
#960887), 147-167. 1996.
138. Siegmund GP, Heinrichs BE, Chimich DD et al. The effect of collision
pulse properties on seven proposed whiplash injury criteria.
Accid.Anal.Prev. 2005;37:275-85.
139. Siegmund GP, Heinrichs BE, Chimich DD et al. Variability in vehicle
and dummy responses in rear-end collisions. Traffic.Inj.Prev.
2005;6:267-77.
140. Smith PK, Hamill J. The effect of punching glove type and skill level on
momentum-transfer. Journal of Human Movement Studies
1986;12:153-61.
141. Smith TA, Bishop PJ, Wells RP. Three dimensional analysis of linear
and angular accelerations of the head experienced in boxing.
International Research Council on Biomechanics of Injury (IRCOBI).
271-286. 1988. Bergish-Gladbach, Germany.
210
BIBLIOGRAPHY
142. Snell Memorial Foundation I. Standard for protective headgear for use
in non-motorized sports. N94. 1994. Sacramento, CA.
143. Society of Automotive Engineers. SAE Recommended Practice:
Instrumentation for Impact Test-Part 1-Electronic Instrumentation-SAE
J211/1 Dec03. 2003. Warrendale, Pennsylvania, USA, Society of
Automotive Engineers.
144. Society of Automotive Engineers. Human Tolerance to Impact
Conditions as Related to Motor Vehicle Design-SAE J885. 2003.
Warrendale, Pennsylvania, USA, Society of Automotive Engineers.
145. Society of Automotive Engineers. SAE Information Report: Sign
Convention for Vehicle Crash Testing-SAE J1733 Dec94. 1994.
Warrendale, Pennsylvania, USA, Society of Automotive Engineers.
146. Spitzer WO, Skovron ML, Salmi LR et al. Scientific monograph of the
Quebec Task Force on Whiplash-Associated Disorders: redefining
"whiplash" and its management. Spine 1995;20:1S-73S.
147. Stemper BD, Yoganandan N, Gennarelli TA et al. Localized cervical
facet joint kinematics under physiological and whiplash loading.
J.Neurosurg.Spine 2005;3:471-6.
148. Stojsih S, Boitano M, Wilhelm M et al. A prospective study of punch
biomechanics and cognitive function for amateur boxers. Br.J.Sports
Med. 2008.
211
BIBLIOGRAPHY
149. Strother CE, Woolley RL, James MB et al. Crush energy in accident
reconstruction. Society of Automotive Engineers World Congress (SAE
Technical Paper 860371). 1986.
150. Sulzer J, Kamalakkannan SB, Wiechel JF et al. Simplified MADYMO
Model of the IHRA Head-form Impactor. Society of Automotive
Engineers Digital Human Modeling Conference. (SAE Paper 06DHM-
68). 2006.
151. Takhounts EG, Eppinger RH, Campbell JQ et al. On the Development of
the SIMon Finite Element Head Model. Stapp Car Crash J.
2003;47:107-33.
152. Tencer AF, Mirza S, Cummings P. Do "whiplash" victims with neck pain
differ from those with neck pain and other symptoms?
153. Thibault LE, Gennarelli TA. Brain injury: An analysis of neural and
neurovascular trauma in the nonhuman primate. 34th Annual Meeting
of the Association for the Advancement of Automotive Medicine
(AAAM). 337-351. 1990. Scottsdale, AZ.
154. Tierney RT, Higgins M, Caswell SV et al. Sex differences in head
acceleration during heading while wearing soccer headgear.
J.Athl.Train. 2008;43:578-84.
212
BIBLIOGRAPHY
155. Torg JS, Quedenfeld TC, Thieler ER, III et al. Collision with spring-
loaded football tackling and blocking dummies. Report of near-fetal
and fatal injuries. JAMA 1976;236:1270-1.
156. U.S.Department of Health EaW. A reason for visit classification for
ambulatory care. DHEW Publication No. (PHS) 79-1352. 1979.
Hyattsville, MD, Public Health Service, National Center for Health
Statistics.
157. United Kingdom Rail Safety and Standards Board. Improving the
design of seats and tables, and evaluating restraints to minimise
passenger injuries (T201)-Review of two-point passenger restraints.
R&D Programme: Research Brief, 1-4. 2008.
158. United States Department of Health and Human Services. Health
Insurance Portability and Accountability Act of 1996. Public Law 104-
191([H.R. 3103]). 8-21-1996.
159. Unterharnscheidt FJ. Translational versus rotational acceleration:
animal experiments with measured input. Scand.J.Rehabil.Med.
1972;4:24-6.
160. Van Ee CA, Nightingale RW, Camacho DL et al. Tensile properties of
the human muscular and ligamentous cervical spine. Stapp Car Crash
J. 2000;44:85-102.
213
BIBLIOGRAPHY
161. Vander VM, Stuhmiller J, Ho K et al. Statistically and biomechanically
based criterion for impact-induced skull fracture.
162. Vassiliou T, Kaluza G, Putzke C et al. Physical therapy and active
exercises--an adequate treatment for prevention of late whiplash
syndrome? Randomized controlled trial in 200 patients. Pain
2006;124:69-76.
163. Versace J. A review of the severity index. Proceedings of the 15th
Stapp Car Crash Conference (SAE Technical Paper 710881), 771-796.
1971. Warrendale, PA, Society of Automotive Engineers.
164. Vetter L, Vanderby R, Broutman LJ. Influence of Materials and
Structure on Performance of a Football Helmet. Polymer Engineering
and Science 1987;27:1113-20.
165. Viano DC, Casson IR, Pellman EJ et al. Concussion in professional
football: comparison with boxing head impacts--part 10. Neurosurgery
2005;57:1154-72.
166. Viano DC, Pellman EJ, Withnall C et al. Concussion in professional
football: performance of newer helmets in reconstructed game
impacts--Part 13. Neurosurgery 2006;59:591-606.
167. Voyvodic F, Dolinis J, Moore VM et al. MRI of car occupants with
whiplash injury. Neuroradiology 1997;39:35-40.
214
BIBLIOGRAPHY
168. Walilko TJ, Viano DC, Bir CA. Biomechanics of the head for Olympic
boxer punches to the face. Br.J.Sports Med. 2005;39:710-9.
169. Wismans J, Spenny CH. Performance requirements for mechanical
necks in lateral flexion. Stapp Car Crash Journal 1983;27th
Proceedings:137-48.
170. Wismans J, van Oorschot H, Woltring HJ. Omni-directional human
head-neck response. Stapp Car Crash Journal 1986;30th
Proceedings:313-31.
171. Withnall C, Shewchenko N, Gittens R et al. Biomechanical investigation
of head impacts in football. Br.J.Sports Med. 2005;39 Suppl 1:i49-i57.
172. Withnall C, Shewchenko N, Wonnacott M et al. Effectiveness of
headgear in football. Br.J.Sports Med. 2005;39 Suppl 1:i40-i48.
173. Yoganandan N, Pintar F, Butler J et al. Dynamic response of human
cervical spine ligaments. Spine 1989;14:1102-10.
174. Yoganandan N, Pintar FA, Maiman DJ et al. Human head-neck
biomechanics under axial tension. Med.Eng Phys. 1996;18:289-94.
175. Yoganandan N, Pintar FA, Stemper BD et al. Biomechanics of human
occupants in simulated rear crashes: documentation of neck injuries
and comparison of injury criteria. Stapp Car Crash J. 2000;44:189-
204.
215
BIBLIOGRAPHY
176. Yoganandan N, Pintar FA, Zhang J et al. Upper neck forces and
moments and cranial angular accelerations in lateral impact.
Ann.Biomed.Eng 2008;36:406-14.
177. Yoganandan N, Sances A, Jr., Pintar F. Biomechanical evaluation of the
axial compressive responses of the human cadaveric and manikin
necks. J.Biomech.Eng 1989;111:250-5.
178. Zhang L, Yang KH, King AI. A proposed injury threshold for mild
traumatic brain injury. J.Biomech.Eng 2004;126:226-36.
216

Case1291318455 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Case1291318455 PDF

Uploaded by

Copyright:

Available Formats

BIOMECHANICAL ENGINEERING ANALYSES OF HEAD AND SPINE IMPACT

INJURY RISK VIA EXPERIMENTATION AND COMPUTATIONAL SIMULATION

ADAM JESSE BARTSCH

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Department of Mechanical and Aerospace Engineering

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

ADAM JESSE BARTSCH

VIKAS PRAKASH, PHD

LARS GILBERTSON, PHD

CLARE RIMNAC, PHD

JOSEPH MANSOUR, PHD

TABLE OF CONTENTS ............................................................................I

Figure 1-1. Struck vehicle delta-V ....................................................... 10

crash cases used in Chapter 1 is gratefully acknowledged. In Chapter 2, the

validation possible. Additionally, Doug Morr, Brian Tanner, Elaine Peterman

use of SEA’s MADYMO platform as well as troubleshooting. A special thanks

to Sally Duquensel and Mutaz Shoukani of TASS-SAFE for their assistance

analyze the techniques developed. The ATD impact testing performed in

Limited, including Hank Jebode and Randy Edwards, in addition to Dr.

requires thanking many people: Dr. Manuel Saavedra and Prasath

Mageswaran assisted in pig spine removal and preparation. Robb Colbrunn

used at no charge and answered many questions regarding preparation and

Institutes of Health Ruth L. Kirchstein Musculoskeletal Research T32 Training

ADAM JESSE BARTSCH

concussion and osteoligamentous cervical spine injury continue to be

prevalent in motor vehicle crashes, athletics and the military. Automotive

safety systems, athletic safety equipment and military personal protective

paraphernalia designs have generally focused on protection discretely

designed on a component basis -- head or spine -- but not a systems basis,

considering the head-spine linkage simultaneously. But since the cervical

one portion of an intrinsically linked osteoligamentous system; thus injury

considered concurrently as opposed to individually.

Historically, component-based injury protection designs have utilized

and anthropomorphic surrogate studies. While a plethora of these prior

(Nij), relatively few of these studies employed a systems-based approach to

understand coupled head-cervical spine injury risk behavior.

may not be as trivial as separate consideration of head or spine component

injury thresholds. Therefore, through a series of six biomechanical

on a systems-based approach considering head and cervical spine injury risk

simultaneously. In Chapter 1, injury risk in inertial loading during real-world

crashes from Chapter 1 were further investigated via use of numerical

simulation in MADYMO. While Chapters 1 and 2 explored low energy car

energy experimental car crashes. Chapter 4 expanded the direct frontal

head impact loading analyzed in Chapter 3 to include oblique and lateral

impact loading during impact experiments with a Hybrid III anthropomorphic

Chapters 1 through 4 helped drive the study of multivariate injury risk in

response to experimental omnidirectional athletic head impacts in Chapter 5.

Chapter 6 further built on the high-energy athletic impacts from Chapter 5

via Matlab and Simulink simulation of helmeted impacts using a systems

dynamics approach. Finally, Chapter 7 analyzed development of an impact

of similar impacts in a helmeted human surrogate. The results of all of these

interrelated during direct or inertial car crash and athletic impacts.

1 CHAPTER 1 – REAL WORLD MINOR REAR CRASHES

Chapter, the engineering reconstruction method using momentum-energy-

restitution is briefly explored while much of the analysis pertains to gathering

and analyzing diagnosed injury for the vehicle occupants.

Minor rear crashes in the United States have been implicated as a

loading, including the ubiquitous ‘whiplash’ injury. Unfortunately, there

large-scale retrospective studies from Germany39;62 and Sweden77 have

for U.S. minor rear crashes to adequately define injury risk.