Traffic Injury Prevention, 4:297–314, 2003

Copyright  C 2003 Taylor & Francis Inc.

ISSN: 1538-9588 print / 1538-957X online
DOI: 10.1080/15389580390216235

Muscle Tetanus and Loading

Condition Effects on the Elastic and
Viscous Characteristics of the Thorax

RICHARD KENT, CAMERON R. BASS, WILLIAM WOODS,

CHRISTOPHER SHERWOOD, N. J. MADELEY, and ROBERT SALZAR
University of Virginia, Charlottesville, Virginia, USA

YUICHI KITAGAWA
Nissan Motor Company, Ltd., Yokosuka, Japan

Thoracic deformation under an applied load is an established indicator of injury risk, but the force required to achieve an
injurious level of deformation currently is not understood adequately. This article evaluates how two potentially important
factors, loading condition and muscle tensing, affect the structural response of the dynamically loaded thorax. Structural
models of two human cadaver thoraxes and two porcine thoraxes were used to quantify the effects. The human cadavers, which
represent anthropometric extremes, were subjected to anterior loading from (1) a 5.1-cm-wide belt oriented diagonally (i.e.,
seatbelt-like loading), (2) a 15.2-cm-diameter rigid hub, and (3) a 20.3-cm-wide belt oriented laterally (i.e., a distributed load).
A structural model having the mathematical formulation of a quasilinear viscoelastic material model was used to model the
elastic and viscous response, with ramp-hold tests used to determine the model coefficients. The effect of thoracic musculature
was assessed using similar ramp-hold tests on the porcine subjects, each with and without forced muscle contraction. Even
maximally contracted thoracic musculature is shown to have a minimal effect on the response, with similar elastic and viscous
characteristics exhibited by each subject regardless of muscle tone. The elastic response is shown to be approximately a
factor of three stiffer for diagonal belt loading and for this distributed loading condition than for the hub loading, indicating
that the response is influenced most by the particular anatomical structures that are engaged and, secondarily, by the area
of load application. Specifically, shoulder involvement is shown to have a strong influence. The force relaxation is found to
be pronounced, but insensitive to the loading condition, with long-time force relaxation coefficients (G∞ ) in the range of 0.1
to 0.3. The findings of this study provide restraint-specific guidelines for the force-deflection characteristics of both physical
and computational thoracic models.

Keywords Muscle Tension; Restraint Systems; Thorax; Thoracic Characteristics; Thoracic Models

BACKGROUND
Thoracic deformation in response to an applied anterior force,
often expressed as midsternal chest deflection, is an established
indicator of injury risk (e.g., Kent, Crandall et al., 2001; Kroell
et al., 1971, 1974; Nahum et al., 1975; Viano, 1978). A max-
imum allowable value is specified in Federal Motor Vehicle
Safety Standard 208 and the National Highway Traffic Safety
Administration (NHTSA) has published risk functions show-
ing a clear increase in injury risk as sternal deflection increases
Received 12 February 2003; accepted 12 March 2003.
Address correspondence to Richard Kent, 1011 Linden Avenue, Charlottes-
ville, VA 22902, USA. E-mail: rwk3c@virginia.edu
297
(Eppinger et al., 1999). Rib fractures begin to occur when mid-
sternal chest deflection reaches approximately 20% of the
initial chest depth (Nahum et al., 1975); the rib cage loses
stability at approximately 32% (Viano, 1978); internal organs
sustain serious crushing injuries at approximately 40% (Viano,
1978); and the posterior surface of the sternum contacts the an-
terior surface of the thoracic spine at approximately 50–60%
(Kent, Crandall et al., 2001; Verriest & Chapon, 1985). Of par-
ticular importance for restraint design and occupant protec-
tion in a crash is the force-time history that can be applied
to the chest without exceeding an injurious level of def-
lection. The aim of this article is to investigate two potentially
important factors influencing the force-deflection characteristics
of the thorax: the loading condition and the presence of muscle
tensing.
298 R. KENT ET AL.

Effect of Loading Condition Muscle Contraction in a Postmortem Subject

Each skeletal muscle fiber is a cylindrical cell with several
The force required to generate an injurious level of deflec-
oval-shaped nuclei arranged beneath a plasma membrane called
tion is strongly dependent on the load distribution on the chest
the sarcolemma. Each fiber contains many parallel rod-like my-
and on the particular anatomical structures that bear the force.
ofibrils that extend the length of the cell. Myofibrils consist of a
For example, Patrick et al. (1965) found that a 3.3 kN hub load
chain of smaller contractile elements called sarcomeres, which
to the sternum resulted in minor trauma in a series of cadaver
contain myocin and actin molecules connected by cross-bridge
sled tests. When the force was distributed over the shoulders and
attachments. Shortening of the sarcomeres decreases the length
chest, an 8.8 kN force was required to generate similar trauma.
of the myofibrils and causes a muscle contraction. The mecha-
Bierman et al. (1946) tested volunteers with a drop device that
nism by which sarcomeres shorten, while still subject to some de-
loaded a four-point belt harness 490 cm2 in area. Painful re-
bate, is reasonably well explained by the sliding filament theory
actions and some minor injuries occurred when loads exceeded
of Huxley (1957). The sliding filament action is accomplished
8.9 kN. When the load area was increased to 1,006 cm2 , however,
when an electrical stimulus (action potential) propagates along
loads up to 13.3 kN were tolerated without injury. In a series of
the sarcolemma, generating a transient rise in intracellular cal-
human volunteer tests, Fayon et al. (1975) observed that tho-
cium ion levels. Calcium ions expose binding sites on the actin
racic response was approximately 38% stiffer under diagonal
and myosin cross bridges, while reserves of adenosine triphos-
belt loading than under hub loading.
phate (ATP) provide energy to “cock” the myocin head. As the
With the increasing presence of air bags in the vehicle fleet,
sequence of binding-releasing and cocking-relaxing repeats, the
the increasing rates of seat belt use, and the current and future
myosin pulls the actin toward the center of the sarcomere caus-
presence of advanced restraint concepts such as force-limiting
ing the sarcomere to shorten. This sliding of the actin filaments
belts and four-point harnesses, it is imperative that both physical
continues while the calcium signal and ATP are present. In the
and computational thoracic models be able to predict injury risk
living, the action potential required to generate the calcium sig-
changes in response to a change in loading condition. In order to
nal is provided by nerve stimulus. The muscles can also contract
predict thoracic injury potential in an automotive compliance,
in a postmortem, prerigor subject, as long as ATP is still present,
assessment, or consumer information test, it is necessary that
through the use of an externally generated action potential. In a
the human model used in the assessment (currently a Hybrid
vigorously exercising living human, ATP stored in the working
III dummy) has the thoracic force-deflection characteristics of
muscles is depleted in approximately 6 s (Marieb, 1992). ATP is
a human and, equally importantly, that the model be able to
then generated by the body either aerobically or anaerobically.
respond to a change in loading condition in a manner similar to
Obviously in a postmortem subject, aerobic production of ATP
a human.
is not practicable. ATP can be produced anaerobically, however,
through the process of glycolisis and the formation of lactic
acid. In a living human, stored ATP and anaerobic production
Effect of Muscle Tensing of ATP can support strenuous muscle activity for over a minute
(Marieb, 1992), so it is reasonable to presume that full muscle
Another factor to consider in the design or evaluation of a tho- contraction (tetanus) can be sustained for at least this long in a
racic model is the effect of muscle tone on the global thoracic limited number of muscle groups using a postmortem, prerigor
response. Researchers from the NHTSA and other groups have subject.
estimated that one-half to two-thirds of crash-involved drivers
may be tensing prior to impact (Ore, 1992; Petit et al., 1998),
and it is unknown how this may affect thoracic response. Quan- METHODS
tifying the effect of muscle tone is difficult because, while it
may be possible to stimulate muscle contraction electrically in Two human cadavers and two porcine subjects were used in
a postmortem subject, human cadavers typically cannot be ob- this study. The human cadavers were used to determine the elas-
tained, screened, and prepared for testing prior to the onset of tic and viscous characteristics of the thorax and to quantify how
rigor mortis. Muscle tensing has been simulated mechanically they changed when the loading condition was varied. The swine,
for simpler systems (e.g., Funk, Crandall et al., 2001) presented which were available within seconds after death, were used to
a series of lower extremity tests in which achilles tension was evaluate the effect of muscle tensing on the thoracic response.
simulated using a clamp-and-cable system), but mechanical sim- Ramp-hold tests and quasilinear viscoelastic formulations were
ulation of the complex thoracic musculature is not practicable. used in all cases to define the characteristics. The testing and the
As a result, attempts have been made to use noninjurious human model development are described below.
volunteer tests to estimate the effect of muscle tensing (King &
Mertz, 1973), but these tests have not been performed to poten- Human Cadaver Tests
tially injurious levels of chest deflection. Porcine tests presented
here are intended to augment the human volunteer studies by ap- Two human cadavers (Table I) were subjected to each of three
proaching injurious levels. loading conditions: diagonal belt loading, distributed loading,
 CHEST BIOMECHANICS 299

Table I Description of cadaver subjects

Identification number 145 147

Age at death (years)/gender 54/Male 63/Female
Weight (kg) 87.7 45.0
Stature (cm) 191.6 160.6
Supine chest depth (4th rib/8th rib) (cm) 21.0/23.5 18.7/21.5
Supine chest breadth (4th rib/8th rib) (cm) 39.2/34.0 26.3/27.3
Supine chest circumference (4th rib/8th rib) (cm) 109.9/100.4 80.3/80.0
Seated chest circumference (4th rib/8th rib) (cm) 106.5/102.0 80.1/80.9
Cause of death Glioblastoma multiforme End-stage respiratory failure (NFS)
Bone quality assessment (Osteogram of left femoral neck) Not measured for younger subject T = −1.35, Z = +0.12
Preservation method Freezing Freezing

Note: All measurements taken with unpressurized pulmonary and cardiovascular systems.

and hub loading. A hydraulic master slave cylinder arrangement
with a high-speed material testing machine (Instron model 8874,
Canton, Massachusetts) was used to generate sinusoidal pre-
conditioning displacements, ramp-hold displacements for model
development, and sinusoidal displacements of different frequen-
cies for model evaluation (Figure 1). The configuration is capa-
ble of generating up to a 17 cm ramp at approximately 1.5 m/s.
Diagonal belt loading and distributed loading were performed
via cable belt systems. The 5-cm-wide diagonal belt passed over
the right shoulder and crossed the anterior thorax approximately
45◦ from the sagittal plane. The belt engaged the clavicle at
approximately the proximal third, crossed the midline approx-
imately midsternally, and exited the body laterally at approxi-
mately the superior-inferior location of the ninth rib (Figure 2).
For distributed loading, a 20.3-cm-wide lateral belt loaded the
area between approximately the second rib and the seventh rib

Figure 1 Loading frame used for cadaveric thoracic characterization tests.

300
Figure 2 Schematic depiction of loading conditions used in cadaver tests.
(Figure 2). For all diagonal belt and distributed loading tests,
a cable spacer was used to control the location at which the
belt disengaged the thorax (Figure 1). This cable spacer was a
piece of square aluminum tube that maintained the loading ca-
bles at a constant distance apart once they passed through the
slots on the loading table. The length of the cable spacer was
adjusted such that the angle from the horizontal to the cables
was approximately 85◦ for the distributed loading belt and ap-
proximately 70◦ for the diagonal belt. The hub load was applied
R. KENT ET AL.
with a 15.2-cm-diameter steel circular plate intended to mimic
the loading surface described by Kroell et al. (1971, 1974). The
hub edges were filleted to reduce edge stresses. A frame with a
bearing track was used with the hub condition to ensure anterior-
posterior loading and to prevent the hub from rotating due to
moments induced during loading (Figure 1).
Age, size, gender, and cause-of-death criteria were used to
select the cadavers for testing. The causes of death for both sub-
jects are unlikely to have degraded tissue properties significantly
premortem (Table I). The larger, younger male subject (subject
145) and the smaller, older female subject (subject 147) were
chosen to represent reasonable upper and lower anthropometric
extremes. The unembalmed cadavers were preserved by freez-
ing prior to testing. To model the in vivo condition as much
as practicable, the cadavers’ cardiovascular and pulmonary sys-
tems were pressurized to typical mean arterial pressure (10 kPa)
and mean inspiration pressure (10 kPa) immediately prior to test-
ing using a heta starch and dye solution and air. Cardiovascular
pressurization was accomplished via the right common carotid
artery, with a balloon catheter occluding the descending aorta at
approximately the level of the twelveth thoracic vertebra. Pul-
monary pressurization was accomplished via a tracheostomy. To
facilitate handling, the subjects’ lower extremities were disar-
ticulated at the hip joint and the upper extremities were removed
at the humerus midshaft using care to maintain the integrity of
the shoulder joint. All cadaveric testing and handling procedures
were approved by the University of Virginia (UVA) Institutional
Review Board.
To maximize the applicability of the structural models (i.e.,
to approach injury levels), but still minimize thoracic response
changes due to tissue damage, tests were designed to approach,
but not exceed, rib fracture threshold for all tests except a final,
injurious test. As discussed above, 20% deflection is approxi-
mately the level at which rib fractures initiate in cadaver tests,
but this threshold varies widely depending on the subject’s age,
gender, size, bone condition, and the presence of superficial soft
tissues. For the large male subject, a nominal sternal deflection
level of 4.4 cm (20% of unloaded supine chest depth) was chosen
as the displacement level for the noninjurious tests (Table II).
Unlike the large male subject, the small female subject had very
little superficial soft tissue. As a result, a nominal sternal deflec-
tion level of 3.0 cm (15%) was used. The lack of rib fractures
was verified audibly during loading and via palpation after each
loading cycle. The order in which the various loading condi-
tions were tested was randomized to minimize the effect of test
order and timing. Furthermore, prior to each ramp-hold test, a
10-cycle, 1 Hz sinusoid having the same magnitude deflection
was used to precondition the thorax and further minimize the
importance of test order. Following the ramp-hold tests, 0.5 Hz
and 3 Hz sinusoid tests at the same magnitude deflection were
performed to assess the model’s ability to predict force response
over a range of sternal deflection rates. At the conclusion of all
model development and validation tests, a 40% ramp-hold test
was performed. For subject 145 this injurious test was performed
with the hub condition, and for subject 147 it was performed with
the diagonal belt condition.
 CHEST BIOMECHANICS 301

Table II Description of cadaver tests

Midsternal potentiometer
Test no. Subject Loading condition Max. Input Disp. distance below sternal notch Test type Notes

41 147 Hub 3.0 cm a Precondition

42 147 Hub 3.0 cm a Step-hold
43 147 Hub 3.0 cm a Model validation
44 147 Distributed 3.0 cm 16.7 cm Precondition
45 147 Distributed 3.0 cm 16.7 cm Step-hold
46 147 Distributed 3.0 cm 16.7 cm Model validation
49 147 Diagonal belt 3.0 cm 6.0 cm Precondition Fracture identified via palpation on rib 3L,
near superior edge of hub
50 147 Diagonal belt 3.0 cm 6.0 cm Step-hold
51 147 Diagonal belt 3.0 cm 6.0 cm Model validation
52 147 Diagonal belt 8.0 cm 6.0 cm Step-hold with injury One fracture identified prior to this test,
multiple fractures after
53 145 Diagonal belt 4.5 cm 9.5 cm Precondition
54 145 Diagonal belt 4.5 cm 9.5 cm Step-hold
55 145 Diagonal belt 4.5 cm 9.5 cm Model validation
56 145 Distributed 4.5 cm 13.2 cm Precondition
57 145 Distributed 4.5 cm 13.2 cm Step-hold
58 145 Distributed 4.5 cm 13.2 cm Model validation
61 145 Hub 4.5 cm a Precondition
62 145 Hub 4.5 cm a Step-hold
63 145 Hub 4.5 cm a Model validation
64 145 Hub 8.9 cm a Step-hold with injury Palpation identified no fractures prior
to this test
a Motion of entire hub measured.

Thoracic instrumentation in the cadaver tests included load Institutional Animal Care and Use Committee. The swine were
cells measuring the cable tension and hub forces, load cells procured at the conclusion of an independent respiratory study
mounted posteriorly to measure the force generated by the (the nature of the respiratory study was such that the thoracic
deforming thorax, and string potentiometers measuring anterior- structural characteristics were unaffected) and were euthanized
posterior displacement of the belts (i.e., external deflection) at immediately prior to biomechanical testing. Subjects had been
three locations. One potentiometer was mounted midsternally intubated and ventilated for the respiratory study, and remained
in the middle of the belt for both the distributed condition and so for the subsequent thoracic testing. Thoracic anthropometry
the diagonal belt condition. For the distributed loading, a second was measured with the subjects at full inhalation (approximately
potentiometer measured displacement 8 cm to the subject’s left 1 kPa tracheal pressure) (Table III) and all tests were initiated at
and 8 cm inferior to the midsternal potentiometer and a third po- this point. The airway was occluded during thoracic loading, so
tentiometer measured displacement 8 cm to the subject’s right the volume of air in the lungs remained constant.
and 8 cm superior to the midsternal potentiometer (Figure 2). For A custom loading frame was designed for thoracic charac-
the diagonal belt condition, a second potentiometer was located terization of the porcine subjects. The apparatus is capable of
12.8 cm superior on the belt midline and a third was located generating a 5 cm ramp in approximately 40 msec (1.25 m/s)
12.8 cm inferior on the belt midline. A single potentiometer and then holding that displacement indefinitely via a ratcheting
measured displacement of the hub. Two 59 gage chestbands mechanism (Figure 3). The anesthetized living subjects were
tracked external thoracic deformation at the lateral levels of the positioned supine in the loading apparatus and an 8.9 cm diam-
fourth and eighth ribs. Piezo-resistive catheter pressure trans- eter rigid hub was positioned with the center of the hub on the
ducers (Millar model SPR-524, Houston, Texas) were inserted
into the aortic arch and into the trachea. Data were sampled at Table III Description of porcine subjects
5 kHz for the ramp portion of the test, at 0.5 kHz for the hold por-
Identification number 1 2
tion of the test, and at 1.0 kHz for the sinusoidal preconditioning Age (months)/gender 4/Male 4/Male
and model validation tests. Weight (kg) 20.5 25.0
Proximal tail to distal snout (cm) 85 91
Supine chest depth (midsternum/bottom 17.5/17.9 20.0/20.0
Porcine Tests of sternum) (cm)
Supine chest breadth (midsternum/bottom 17.2/19.0 20.0/22.3
Two postmortem porcine subjects (Sus scrofa), obtained of sternum) (cm)
through the UVA Department of Comparative Medicine, were Supine chest circumference (midsternum/bottom 56.5/56.5 63.8/63.8
of sternum) (cm)
used to evaluate the effect of muscle tetanus on the global tho-
racic response. All test procedures were approved by the UVA Note: All measurements taken with ventilated subject at maximum inhalation.
302
Figure 3 Porcine test methodology and electrode placement.
midline midway between the xiphisternum (i.e., the caudal end
of the xiphoid cartilage) and the manubrium sterni (Figure 3).
The subjects were then euthanized using a solution of pentobar-
bital, a barbiturate that affects the central nervous system and was
therefore assumed to have no affect on the muscles’ response
to an external stimulus. Immediately following death, tetanus
of the thoracoabdominal musculature was achieved via 12 elec-
trodes (6 pairs) positioned as described in Table IV. Immediately
following contraction, the ramp displacement was applied to the
chest and held for 60 s. Tetanus was maintained throughout the
hold portion of the test. Two tests were performed on each sub-
ject, one with the muscles contracted and one without (Table V).
The order of testing was reversed from the first subject (tonic
Table IV Description of electrode placement (see Figure 3)
Electrode pair (bilateral)
A1—Overlying pectoral muscle body
A2—Midway along the costal margin
B1—Dorsal to the midcoronal plane at level of the xiphisternum
B2—Midcoronal plane on the costal margin
C1—Paramedian overlying epigastrium
C2—Paramedian ventral abdominal wall at level of the iliac crest
R. KENT ET AL.
then atonic) to the second subject (atonic then tonic) to separate
the effect of potential tissue damage in the first test from the
effect of muscle tone.
The force applied anteriorly was measured using a piezo-
resistive load transducer positioned between the hub and its sup-
port. An accelerometer mounted on the hub was used for inertial
compensation. The posterior reaction force was measured using
eight piezo-electric load crystals positioned below the posterior
loading platform. Transducers were inserted into the trachea to
measure pulmonary pressure, through the carotid artery or the
femoral artery into the aortic arch to measure arterial pressure,
and through the jugular vein or the femoral vein into the inferior
vena cava to measure venous pressure. Data were sampled at
1 kHz.
Data Analysis and Model Development
The posteriorly measured force and the midsternal chest de-
flection were used in the development of quasilinear viscoelas-
tic (QLV) structural models of the human and porcine thoraxes.
The posterior force was used in the QLV model development
because the formulation chosen does not include consideration
of the accelerated sternal mass (i.e., there is no inertial term in
the model).
Development of QLV Structural Formulation
Depending on the characteristics of a material, the form of a
viscoelastic model used to describe it can include multiple spring
and dashpot elements arranged singly, in series, or in parallel.
The elastic and viscous characteristics of biological solids have
often been described using a model consisting of an elastic ele-
ment in parallel with one or more Maxwell elements (Figure 4),
where the characteristics of the springs and the dashpots are
material specific (or, in this case, the structural characteristics
are loading condition specific). This model form will be used
in the current study, but the model will be used to describe the
relationship between the applied force, F, and the resulting mid-
sternal chest deflection, c, rather than a stress-strain description
of a material.
The force-deflection response of this model can be described
using a differential equation for the force in terms of the time
derivatives of the deflection. Experimental determination of the
model coefficients, however, is facilitated by a hereditary inte-
gral description of the model’s response to an applied step in
Primary muscle groups activated
Pectoralis, intercostals, ventral serratus, rectus thoracis
Latissimus dorsi, dorsal serratus, intercostals, thoracic trapezius
External oblique abdominus, internal oblique abdominus, transverse abdominus,
rectus abdominus
 CHEST BIOMECHANICS 303

Table V Porcine test description

Test no. Subject Muscle state Step magnitude Time after death Notes

31 1 Tensed 5.07 cm 53 s No hard tissue injury found via palpation after test 3 1, but thorax depth
32 1 Relaxed 3.87 cm <600 s decreased by 1.2 cm. Subject ventilated for approximately 5 min between test 3 1
and test 3 2, but original chest depth could not be obtained
33 2 Relaxed 3.78 cm 94 s After test 3 3, decrease in thoracic depth observed, but less pronounced than the
34 2 Tensed 3.87 cm 520 s decrease observed for subject 1

deflection. The hereditary integral formulation results from the describes the response of a QLV structure:
rule of linear superposition (see Flugge, 1967). For the thorax,

consider the case of several displacements of different magni- t
∂Felastic [c(τ )] ∂c(τ )
tudes applied successively. The force response is a function of F(t) = G(t − τ ) dτ (3)
−∞ ∂c ∂τ
the magnitude of each applied displacement and of all displace-
ments prior to it. If the limiting case of a series of infinitesimal This convolution integral can be solved for the force response
displacements is considered, the result is the hereditary integral to any applied displacement time history (subject to the limita-
description of the structural response: tions described in Appendix A).
In this study, an exponential form was assigned for the in-
 t
∂c(τ ) stantaneous elastic function:
F(t) = FR (t − τ ) dτ (1)
−∞ ∂τ
Felastic (c) = A · [Exp(B · c) − 1] (4)
where FR (t − τ ) is the relaxation function of the material, c(t) is
where A and B are loading condition-specific model coefficients
the midsternal chest deflection, and F(t) is the force response to
determined from the experimental data. Four exponential terms
the applied displacement. Fung’s QLV theory assumes that the
were used in the reduced relaxation function, G(t), which was
force relaxation function can be divided into a spatial compo-
assigned the form of a sum of exponentials:
nent (i.e., a strain-dependent or, in this case, a chest deflection-
dependent function) and a temporal component (i.e., a time-
dependent function) (Fung, 1981): 
4
G(t) = Gi · Exp(−βi t) + G∞ (5)
i=1

FR (c, t) = G(t) · Felastic (c) (2)

where Felastic (c) is the instantaneous elastic function (i.e., the
force response to an instantaneously applied step deflection)
and G(t) is a normalized, or reduced, relaxation function, which
is a monotonically decreasing function of time. The elastic func-
tion may assume a nonlinear form and, for biological materials
subjected to finite strains, a nonlinear form of the elastic re-
sponse is usually required. Despite this nonlinearity in the force
response, the principle of superposition remains valid due to the
assumption of time linearity, and the resulting hereditary integral
where Gi , G∞ , and βi are model coefficients. The coefficients Gi
and G∞ are normalized so that their sum is unity. A numerical
iteration scheme described in Appendix A was used to determine
independent coefficients for all loading conditions and subjects.
RESULTS
The thoracic response was found to be sensitive to changes in
the loading condition and to be essentially unchanged by muscle
tone. The force-deflection response exhibited slight spatial non-
linearity and exhibited pronounced force relaxation in all tests.
The specific results for the human cadavers and the swine are
presented below.

Porcine Tests

Due to limitations of the laboratory setup for the porcine test-

ing, preconditioning tests and model validation tests were not
possible. As a result, the ramp-hold model development tests
were influenced by the order of testing. As noted in Table V, a
clear change in each subject’s thoracic depth was observed af-
Figure 4 Thoracic model. ter the first ramp-hold test. Following the initial test, the thorax
304 R. KENT ET AL.

depth decreased 0.7 cm (subject 2) and 1.2 cm (subject 1). In

other words, after the first test the subject did not regain its initial
chest shape. This deformation is not believed to be due to tho-
racic hard tissue damage for multiple reasons. First, palpation
before and after all tests indicated no fractures or instability of
the chest wall. Second, these subjects were 4 months old (ap-
proximately 2 months before sexual maturity and approximately
14 months before maximum size) and pediatric ribs typically ex-
hibit large strains prior to fracture. Finally, there was no dramatic
change in the thoracic force-deflection response once the offset
due to the change in chest depth was considered (Figure 5). We
hypothesize that, rather than hard tissue damage, the change in
chest depth is due to damage induced in the superficial muscula-
ture and viscera and in the internal organs and soft tissues as well
as a long-time viscous response similar to that observed in the
human cadaver preconditioning tests (discussed below). The re-
sult of this change in chest depth is that the initial condition was
different between tests. This change is quantified in Figure 5,
which shows the posteriorly measured force as a function of
the position of the string potentiometer. The force-displacement
curves for the first and second tests are clearly offset. These
figures also show, however, that the slope of the force-position
curve is essentially unchanged from the atonic to the tonic con-

Figure 6 Reduced relaxation functions and elastic functions for porcine tests
with and without muscle tone showing interspecimen difference in response but
a small musculature effect.

dition for both subjects, regardless of the order of testing. By

reversing the order of the tonic and atonic tests, it is possible
to evaluate the effect of muscle tone independently of the ef-
fect of test order, and it is clear that the effect of muscle tone
is small. This insensitivity to muscle tension is also apparent in
the QLV model coefficients calculated for each test (Figure 6
and Table VI). While differences in both the elastic function and
the reduced relaxation function can be seen with muscle ten-
sion, the differences are small and are likely within the range of
repeatability for the tests.

Human Cadaver Tests

Ten separate tests were performed on each subject. For each

Figure 5 Porcine force-deflection response with linear regression. Note the
shift due to test order and also that the slope is essentially unchanged between
tests for both subjects, indicating a negligible effect from muscle tetanus.
loading condition, a sinusoidal preconditioning test was per-
formed (Figure 7) followed by a ramp-hold test and then a model
validation test (Table II). The force response was observed to
change during approximately the first three cycles of precon-
ditioning for all loading conditions, but reached a steady-state
response thereafter.
For both subjects, the hub condition generated the lowest
force for a given amount of sternal deflection (see raw data
in Appendix B). The other two loading conditions (distributed,
diagonal belt) generated similar force-deflection responses with
the diagonal belt generating slightly more force at a given
 CHEST BIOMECHANICS 305

Table VI QLV coefficients for porcine tests

Test A (N) B (cm−1 ) G1 G2 G3 G4 G∞ β1 (s−1 ) β2 (s−1 ) β3 (s−1 ) β4 (s−1 )

3 1 507 0.30 0.10 0.09 0.19 0.47 0.15 0.10 2 10 100

3 2 813 0.18 0.12 0.16 0.09 0.35 0.28 0.10 2 25 100
3 3 306 0.50 0.06 0.06 0.25 0.49 0.14 0.10 2 15 100
3 4 690 0.31 0.10 0.06 0.22 0.43 0.19 0.10 2 25 100

displacement for the small female subject and the distributed

load generating slightly more force for the large male sub-
ject. Determination of the elastic function for the QLV model
revealed that the hub loading condition was the least stiff elas-
tically for both subjects: on the order of one-third as stiff as
the other two loading conditions (Figure 8 and Table VII). The
elastic response of the large male was essentially identical for
the distributed load and the diagonal belt load, while the small
female exhibited slightly stiffer response under the diagonal belt
load. The large male, which had substantially more superficial
soft tissue around the rib cage (Table VIII), exhibited more force
relaxation than the small female for all loading conditions. The
reduced relaxation functions for the large male were similar for
all loading conditions, with a G∞ value between 0.104 and 0.143
for all noninjury tests. The small female also showed similar
force relaxation for all noninjury tests, with G∞ ranging from
0.200 to 0.288.
The model validation test results are shown in Appendix B.
For the diagonal belt and distributed loading conditions, the

Figure 7 Preconditioning for the hub loading condition and the small female Figure 8 Elastic functions (i.e., the force response to a step deflection) and
subject (typical). reduced relaxation functions for the cadaver subjects and all loading conditions.
306 R. KENT ET AL.

Table VII Loading condition-specific QLV coefficients for human cadaver subjects

Subject 147 (Female) 147 (Female) 147 (Female) 145 (Male) 145 (Male) 145 (Male)

Loading condition Hub Distributed Diagonal belt Hub Distributed Diagonal belt
Test used for model development 42 45 50 62 57 54
A (N) 472.5 1971.0 2631.0 715.3 5098.0 4741.0
B (cm−1 ) 0.390 0.250 0.250 0.330 0.200 0.200
G1 0.118 0.133 0.076 0.053 0.130 0.065
G2 0.141 0.099 0.124 0.136 0.180 0.079
G3 0.280 0.241 0.262 0.321 0.270 0.337
G4 0.173 0.284 0.337 0.346 0.320 0.388
G∞ 0.288 0.241 0.200 0.143 0.104 0.131
β1 (s−1 ) 0.163 0.280 0.133 0.600 0.450 0.082
β2 (s−1 ) 1.23 1.65 1.39 2.00 2.65 1.42
β3 (s−1 ) 25.3 34.5 32.6 55.9 50.0 43.2
β4 (s−1 ) 155.0 155.0 155.0 155.0 150.0 206.0

model predicted the resulting force response within 5% for both
the 0.5 Hz sinusoid and the 3 Hz sinusoid. The hub model was
less predictive, underpredicting the measured force by approxi-
mately 10% for the 0.5 Hz wave and overpredicting the measured
force for the 3 Hz wave.
Both subjects exhibited severe thoracic hard tissue injury
during necropsy. In both cases, however, the rib fracture pat-
tern suggested that the rib fractures occurred primarily during
the injurious test and not before. For the small female subject
(147), the rib fractures were concentrated in the area of diagonal
belt loading (Table IX) and for the large male subject (145) the
fractures were proximate to the hub edge. Palpation between
tests revealed only one fracture for the small female prior to the
injurious test and no fractures for the large male prior to the in-
jurious test. Comparison of the injurious test response with the
response from the noninjurious test with the same loading con-
dition supports the assertion that the thoracic structure was not
compromised substantially prior to the final test. The presence
of the single rib fracture prior to the final test also verifies that,
as intended, these tests approached injury and that the structural
models are applicable to that level.
DISCUSSION
The structural characteristics of the thorax have been stud-
ied since the 1960s. Due to the low rates of seat belt use in the
1960s and 1970s, early studies focused on thoracic loading expe-
rienced by an unbelted subject. Force-deflection characteristics
of the thorax were developed using blunt objects representing a
steering wheel hub or an instrument panel, and loading rates were
representative of those experienced by unrestrained occupants in
severe collisions (e.g., Kroell et al., 1971, 1974; Mertz & Gadd,
Table VIII Summary of superficial soft tissue depth
Distance from outer skin surface to rib (level of 4th rib)
Subject 5 cm Left of sternum Midsternum 5 cm Right of sternum
145 (Large male) 1.1 cm 0.6 cm 1.3 cm
147 (Small female) 0.7 cm 0.4 cm 0.7 cm
1971; Nahum et al., 1970). Force-deflection corridors were gen-
erated based on these data, which formed the primary thoracic
validation standard for the Hybrid III dummy (Schneider et al.,
1989) and for numerous finite element and other thoracic mod-
els. As seat belt use rates increased, researchers recognized the
need to evaluate thoracic response under localized belt loading
and at lower loading rates. The primary source of thoracic re-
sponse data for diagonal belt loading is the test series presented
by L’Abbe et al. (1982) and Cesari and Bouquet (1990, 1994),
which involved a series of human cadavers positioned supine
on a table and subjected to seat belt-like loading on the anterior
thorax via a pendulum and cable system.
For the purposes of dummy and computer model validation,
however, neither the Kroell et al. (1971, 1974) tests nor the
Cesari and Bouquet (1990, 1994) tests are ideal. First, substan-
tially different test methodologies were used in the two data
sets and tests of multiple loading conditions on the same sub-
ject were not performed. Second, the force-deflection corridors
developed in the past were obtained at discrete impactor energy
levels and are therefore not unique descriptors of the elastic,
viscous, and inertial characteristics of the thorax. In other words
if only discrete impactor energy levels are considered in the
assessment of a thoracic model, it is possible that the elastic,
viscous, and inertial characteristics of the model could all be
incorrect but offsetting in such a manner that the overall force-
deflection response of the model falls within even multiple cor-
ridors. A third limitation of the available blunt hub data is that
they consist largely of impact tests in which only the anterior
force was measured. When the thorax is subjected to an impact
and the force is measured at the impacting surface, large iner-
tial forces are measured prior to significant deformation of the
thorax. As a result, the characteristic thoracic force-deflection
corridor is dominated by the inertial response early in the cor-
ridor and it is not possible to separate the elastic and viscous
characteristics. This characteristic of the data limits its useful-
ness in many contemporary applications. In the case of seat belt
loading, for example, there is never an impact between the belt
and the thorax and the thoracic response has a smaller inertial
component. A more robust evaluation of thoracic models can
be performed if the elastic, viscous, and inertial contributions
 CHEST BIOMECHANICS
Table IX Human cadaver thoracic necropsy results
Loading condition
Subject for injury test Injury summary
145 (Large male) Hub 6 Fractures on 6 ribs (3L, 4L, 2R, 3R, 4R, 5R), no soft tissue injury
147 (Small female) Diagonal belt 14 Fractures on 11 ribs (3L, 4L, 5L (2), 6L (2), 7L, 8L, 9L, 2R, 3R, 5R, 6R (2)), no soft tissue injury
Note: In several of the later tests, both subjects exhibited subcutaneous emphysema during pulmonary pressurization, which strongly suggests that a pneumo-
thorax was caused during a previous test. The presence of a pneumothorax could not be verified at autopsy, however, and it is unlikely that a pneumothorax by
itself would affect the thoracic structural response.
to the applied force can be determined uniquely and separated.
The models presented here provide a means to define the elas-
tic and viscous parameters and therefore provide data for a
more robust assessment of physical and computation thoracic
models.
The findings of this study should therefore be considered
in the design of future test dummies (ATDs), in the evaluation
of thoracic finite element models and in the interpretation of
the measurements obtained using current ATDs. The two most
important results of this study are: first, that full muscle tetanus
had no measurable effect on the structural response of the thorax;
and, second, that the structural response was strongly dependent
on the loading condition.
The influence of thoracic musculature has been studied in the
past using human volunteers. Lobdell et al. (as cited in King &
Mertz, 1973) discussed a series of seven tests involving male vol-
unteers subjected to quasidynamic hub loading on the anterior
thorax in both a “relaxed” and a “tensed” state. A 338% increase
in linear elastic thoracic stiffness (from 70 N/cm to 236 N/cm)
was observed when the volunteers maximally tensed the muscles
of their shoulders, thorax, arms, back, and neck. Stalnaker et al.
(as cited in King & Mertz, 1973) corroborated this result using
two human volunteers. They found a 300% increase in thoracic
stiffness (from 403 N/cm to 1,140 N/cm) when the volunteers
were in a “tensed” state. In both of these test series, however,
loading could not be performed to potentially injurious levels
of chest deflection (approximately 11% of chest depth in the
Lobdell et al. study and 8% in the Stalnaker et al. study). As
a result, it was not known if this dramatic increase in stiffness
would remain at larger deflections when the rib cage, rather than
the musculature itself, is the structure primarily responsible for
the measured stiffness. The current study addressed this issue by
performing tests to potentially injurious levels on postmortem
swine with and without muscle tetanus. These tests were per-
formed to deflection levels of approximately 15% (subject 2)
and 25% (subject 1) of the initial chest depth. At this level of
deflection, the rib cage appears to be primarily responsible for
the elastic response of the thorax, particularly in the swine with
its less-pronounced anterior musculature. It is logical that the
response at low levels of chest deflection would be dominated
by the effect of musculature, but that this effect would become
less pronounced as the deflection level increases and the rib cage
becomes loaded more heavily.
The second finding of this study, that the loading condition
strongly influences the thoracic response, is supported by nu-
307
MAIS
3
4
merous studies (e.g., Bierman et al., 1946; Fayon et al., 1975;
Patrick et al., 1965) and is intuitively correct. We believe this
is the first study, however, to quantify the change in response
for the same subject loaded by different conditions to levels ap-
proaching rib fracture threshold. Our finding that the hub load
generates a lower effective stiffness than the distributed load or
the diagonal belt load was expected due simply to the differ-
ence in the loaded area. The hub engages only approximately
180 cm2 , while the diagonal belt engages 250 cm2 and the dis-
tributed load engages 730 cm2 . The area of load application
does not completely explain the observed differences in effec-
tive stiffness, however. For both subjects, the diagonal belt and
distributed conditions generated similar values of elastic stiff-
ness despite a large difference in loaded area. It is apparent that
the specific anatomical structures that are engaged play at least
as large a role as the loaded area. The diagonal belt engages
the short and thick upper ribs as well as the clavicle. These stiff
structures were not engaged by either of the other two loading
conditions.
The viscous force relaxation, on the other hand, was not as
dependent on the loading condition, with long-time relaxation
(G∞ ) reaching approximately 25% of the instantaneous maxi-
mum force for the small female and approximately 10% for the
large male, regardless of the loading condition. The force was,
however, observed to relax more in the injurious test than in the
noninjurious tests for both subjects.
It has been established that the Hybrid III thorax is stiffer
than a human’s (e.g., Cesari & Bouquet, 1991, 1994) and the
current study provides a partial explanation for this. Based on
the human volunteer-based studies of muscle tensing discussed
above, the cadaver-based force-deflection corridors used in the
design of the Hybrid III dummy and the THOR dummy (and
often used as finite element model validation corridors) were
adjusted by approximately 650 N (Kroell as cited in Backaitis,
1994). In other words, due to the observed increase in stiffness
when the muscles were tensed in those low-deflection tests, the
force values for the corridor were increased to simulate a tensed
driver or passenger in a collision. The porcine tests presented
here cast some doubt on the validity of this adjustment. The
obvious result of this adjustment is that the chest deflection
measured by a Hybrid III in a sled test is not the same de-
flection that would be experienced by a human. As a result,
injury risk functions used in the interpretation of Hybrid III
chest deflection measures must be developed explicitly for that
dummy (e.g., Kent, Bolton et al., 2001) and the determination of
308 R. KENT ET AL.
Hybrid III deflection thresholds based on cadaver thresholds is
not straightforward.
Limitations of the human cadaver tests presented here include
the use of a constrained back condition, which may result in
different response than a thorax loaded only by its inertia as it is
in most frontal car crashes. One possible effect of the posterior
boundary is an increase in elastic stiffness due to constraint of
the costovertebral articulations, though to our knowledge this
increase has never been shown and is likely small compared to
interspecimen variability and intertest variability.
Due to interperson variability, it is not possible to quantify the
effect of load distribution without performing multiple tests on
a single subject. The issue of response changes due to repeated
tests was addressed in three ways in this study. First, test order
was varied between subjects so that the effect of test order could
be separated from the effect of load distribution. Second, the
thorax was preconditioned prior to each ramp-hold test. Finally,
a single loading condition was repeated for each subject. As
shown in Figures B1 and B2, the force-deflection curve for the
repeated loading condition was essentially unchanged for both
subjects.
A second limitation of this study is the use of a porcine model.
The swine’s thoracic structure differs from the human’s in sev-
eral important ways. Most importantly, due to its body position
during locomotion, the swine’s musculature is more concen-
trated on the dorsal aspect of the thorax than the human’s. The
swine thorax also has a different shape than the human, being
narrower with a greater depth, and the swine’s ribs are less sloped
than the human’s. These characteristics may result in stiffer re-
sponse for a porcine chest compared to its human counterpart
(Viano & Warner, 1976) and may decrease the effect of muscle
tensing. The swine is used often as a reasonable surrogate for a
human, however, and its thoracic structural response has been
used as a model for the human’s in the past (e.g., Viano & Warner,
1976; Viano et al., 1977). Future research should evaluate the
effect of muscle tensing when the porcine thorax is loaded from
directions other than frontal. For example, loading the porcine
thorax posteriorly, where the musculature is concentrated, may
better simulate anterior loading on a human.
Another limitation of this study is that it was not possible with
the current test setup to quantify the degree of muscle tensing
attained in the various muscle groups. While the muscle tension
could be verified both visually and palpably, it is difficult to
define the degree of muscle tone and it can be described only
in quantitative terms. On initiation of the action potential, the
subject visibly tensed including adduction of the extremities
due to contraction of the pectoralis muscles and the abdominal
muscles. If the degree of adduction of these extremities is used
as a marker of the degree of muscle tone, then the muscle tensing
could not be observed to change during the hold portion of the
test. Following the test, strong muscle contractions could still
be obtained by activating the stimulus and the muscle could be
felt to flex strongly up to several minutes after the completion
of testing. It is possible, however, that the diaphragm was not
tensed using the external electrodes. Tensing of this muscular
Figure 9 Comparison of model-predicted force-deflection response and hu-
man volunteer force-deflection response for quasistatic diagonal belt loading.
structure may play a role in stiffening the thoracic response since
a tensed diaphragm may be less prone to displace inferiorly when
the pressurized lungs are loaded by the hub. Future research
should include attempts to quantify the degree of muscle tensing
obtained and to ensure that the diaphragm is being stimulated.
Another limitation of this study is the development of struc-
tural models specifically for one loading condition (belt, dis-
tributed, hub). These loading conditions resulted in large vari-
ation in model coefficients, and it is inappropriate to use the
coefficients developed for one loading condition to predict the
response at a different loading condition. In other words, unlike
a material model, the structural models developed here have
not been generalized to accommodate changes in loading con-
dition. As discussed above, it is insufficient to simply generalize
the models using the area of loading because the inhomogene-
ity of the thorax results in the particular region of loading (e.g.,
the shoulder) being more important than the area of loading in
terms of dictating the structural response. Future research should
investigate methods for generalizing these structural models to
account for this inhomoegeneity and to account for the area of
loading.
Two human cadaver subjects and two porcine subjects have
been presented here. While the human cadavers represented an-
thropometric extremes, one important finding of this study is
that the relationship between subject size and thoracic stiffness
is not straightforward due to the geometric complexity of the
thorax. As a result, it is difficult to determine a priori which
subjects will bound the response corridors and the only way to
develop a reasonable range of responses is to perform numerous
tests. Plans for future research include additional subjects tested
under identical conditions. The cadavers used in these tests were
both relatively young at death (54 years and 63 years), were not
obese, and the cause of death was unlikely to cause significant
bone degeneration so their responses are assumed to be reason-
able representations of a large number of living adults. In fact,
the model developed for the diagonal belt condition correlated
well with the quasistatic responses measured by Fayon et al.
 CHEST BIOMECHANICS
(1975) using 13 human volunteers (Figure 9, a 10-s loading time
is assumed for the quasistatic human volunteer tests). There is,
however, wide variation among humans. The contribution of the
research presented here is the relative change in effective stiff-
ness that accompanies a change in loading condition. Additional
tests are necessary in order to quantify the range of responses
that may exist.
CONCLUSIONS
The primary conclusions of this study are: first, that even full
tetanus of the thoracic musculature had a small effect on the
structural response of the thorax (elastic or viscous) when de-
flection levels approaching injury were considered; and, second,
that the thoracic response is highly sensitive to changes in the
loading condition, due both to the changing area of load distri-
bution and the changing anatomical structures that are loaded.
Loading from a diagonal belt results in a substantially stiffer
response than loading from a hub, and this increase has been
quantified for cadavers representing anthropometric extremes.
The data presented here are intended to provide robust finite el-
ement model validation standards as well as guidelines for the
design and assessment of ATDs. While the strain distribution
within the thorax (information necessary for a true validation of
a thoracic finite element model) was not measured in these tests,
a finite element model that can match both the elastic and the
viscous responses of the subjects tested here, and match them
for all the loading conditions considered here, will have a more
robust validation than previously has been possible.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the support of Nissan Motor Company.
The findings expressed herein are the authors’ and do not necessarily represent
the views of the funding agency. The staff and students of the UVA Center
for Applied Biomechanics assisted in the design and manufacture of the test
apparatus and in the performance of the tests. In particular, the contributions
of Martin Davis, David Lessley, Mark McCardell, Jason Kerrigan, Jim Bolton,
and Dr. Edward Ullman are recognized. The contributions of Dr. Greg Hall are
also acknowledged. The UVA Department of Emergency Medicine and the UVA
Center for Comparative Medicine supported the porcine testing and oversaw the
work.
REFERENCES
Backaitis, S. (Ed.). (1994) Biomechanics of Impact Injury and Injury Tolerances
of the Thorax-Shoulder Complex, Publication PT-45, Society of Automotive
Engineers, Warrendale, PA.
Bierman, H. R., Wilder, R. M., Hellems, H. K. (1946) The Physiological Effects
of Compressive Forces on the Torso, Report No. 8, Naval Medical Research
Institute Project X-630, Bethesda, MD.
Cesari, D., Bouquet, R. (1990) Behavior of Human Surrogates Thorax under Belt
Loading, SAE Paper 902310, Society of Automotive Engineers, Warrendale,
PA.
Cesari, D., Bouquet, R. (1994) Comparison of Hybrid III and Human Cadaver
Thoracic Deformations, Paper 942209, Proc. 38th Stapp Car Crash Confer-
ence, pp. 65–76, Society of Automotive Engineers, Warrendale, PA.
309
Eppinger, R., Sun, E., Bandak, F., Haffner, M., Khaewpong, N., Maltese, M.,
Kuppa, S., Nguyen, T., Takhounts, E., Tannous, R., Zhang, A., Saul, R. (1999)
Development of Improved Injury Criteria for the Assessment of Advanced
Automotive Restraint Systems: II., National Highway Traffic Safety Admin-
istration, U.S. Department of Transportation, Washington, DC.
Fayon, A., Tarriere, C., Walfisch, G., Got, C., Patel, A. (1975) Thorax of 3-
Point Belt Wearers during a Crash (Experiments with Cadavers), SAE Paper
751148, Society of Automotive Engineers, Warrendale, PA.
Flugge, W. (1967) Viscoelasticity, Blaisdell, Waltham, MA.
Fung, Y. C. (1981) Biomechanics: Mechanical Properties of Living Tissues,
Springer-Verlag, New York.
Funk, J. R., Crandall, J. R., Tourret, L. J., MacMahon, C. B., Bass, C. R.,
Khaewpong, N., Eppinger, R. H. (2001) The Effect of Active Muscle Tension
on the Axial Injury Tolerance of the Human Foot/Ankle Complex, Proc. 17th
International Technical Conference on the Enhanced Safety of Vehicles (ESV),
Paper 237, Amsterdam, The Netherlands.
Funk, J., Martin, P., Crandall, J. (2001) A Method for Approximating High-Rate
Viscoelastic Properties from Low-Rate Tests, Proc. Annual BMES.
Huxley, A. F. (1957) Muscle Structure and Theories of Contraction, Prog. Bio-
phys. Biophys. Sci., Vol. 7, pp. 255–318.
Kent, R., Bolton, J., Crandall, J., Prasad, P., Nusholtz, G., Mertz, H., Kallieris,
D. (2001) Restrained Hybrid III Dummy-Based Criteria for Thoracic Hard
Tissue Injury Prediction, Proc. 2001 Conference of the International Research
Council the Biomechanics of Impact (IRCOBI), Isle of Man.
Kent, R., Crandall, J., Bolton, J., Prasad, P., Nusholtz, G., Mertz, H. (2001) The
Influence of Superficial Soft Tissues and Restraint Condition on Thoracic
Skeletal Injury Prediction, Stapp Car Crash J., Vol. 45, pp. 183–203.
King, W., Mertz, H. (Eds.). (1973) Human Impact Response Measurement and
Simulation, Plenum Press, New York.
Kroell, C., Schneider, D., Nahum, A. (1971) Impact Tolerance and Response of
the Human Thorax, SAE Paper 710851, Society of Automotive Engineers,
Warrendale, PA.
Kroell, C., Schneider, D., Nahum, A. (1974) Impact Tolerance and Response of
the Human Thorax: II., SAE Paper 741187, Society of Automotive Engineers,
Warrendale, PA.
L’Abbe, R., Dainty, D., Newman, J. (1982) An Experimental Analysis of
Thoracic Deflection Response to Belt Loading. Proceedings of the 7th
International Research Council on the Biomechanics of Impact (IRCOBI)
Conference, Bron, France, pp. 184–194.
Marieb, E. (1992) Human Anatomy and Physiology, Benjamin/Cummings,
Redwood City, CA.
Mertz, H., Gadd, C., (1971) Thoracic Tolerance to Whole-Body Deceleration,
SAE Paper 710852, Society of Automotive Engineers, Warrendale, PA.
Nahum, A, Gadd, C., Schneider, D., Kroell, D. (1970) Deflections of the Human
Thorax under Sternal Impact, SAE Paper 700400, Society of Automotive
Engineers, Warrendale, PA.
Nahum, A. M., Schneider, D. C., Kroell, C. K. (1975) Cadaver Skeletal Re-
sponse to Blunt Thoracic Impact, Proc. 19th Stapp Car Crash Conf., pp. 259–
293.
Ore, L. S. (1992) Design Requirements and Specifications: Dummy Lower Ex-
tremity Development Task, Event Report, National Highway Traffic Safety
Administration, U.S. Department of Transportation, Washington, DC.
Patrick, L. M., Kroell, C. K., Mertz, H. J. (1965) Forces on the Human Body in
Simulated Crashes, Proc. 9th Stapp Car Crash Conf., pp. 237–260.
Petit, P., Portier, L., Trosseille, X. (1998) Rigid Body Model of the Hybrid III
Dummy Lower Limb Including Muscle Tension under Car Crash Conditions,
Proc. International Conference on the Biomechanics of Impact (IRCOBI),
pp. 173–188.
Schneider, L., King, A., Beebe, M. (1989) Design Requirements and Specifica-
tions: Thorax-Abdomen Development Task, Interim Report, Trauma Assess-
ment Device Development Program, Report No. DOT HS 807 511, National
Highway Traffic Safety Administration, U.S. Department of Transportation,
Washington, DC.
Verriest, J., Chapon, A. (1985) Validity of Thoracic Injury Criteria Based on
the Number of Rib Fractures, SAE Paper 856027, Society of Automotive
Engineers, Warrendale, PA.
310 R. KENT ET AL.

Viano, D. C. (1978) Thoracic Injury Potential, Proc. 3rd International Meeting unknown, time-varying, value, (t):
on Simulation and Reconstruction of Impacts in Collisions, International
Research Council on the Biomechanics of Impact (IRCOBI), Bron, France, F(c0 , tmin ) F(c0 , tmeas )
pp. 142–156. Felastic (c0 ) ≈ ≈
Viano, D., Kroell, C., Warner, C. (1977) Comparative Thoracic Impact Response G(tmin ) G(tmeas )
(A3)
of Living and Sacrificed Porcine Siblings, SAE Paper 770930, Society of
≈ (t) · F(c0 , tmeas )
Automotive Engineers, Warrendale, PA.
Viano, D., Warner, C. (1976) Thoracic Impact Response of Live Porcine Sub-
jects, SAE Paper 760823, Society of Automotive Engineers, Warrendale, So, the scaling factor, (t), is the ratio of the reduced relaxation
PA. functions:

G(tmin )
(t) = (A4)
G(tmeas )
APPENDIX A—DESCRIPTION OF MODEL
DEVELOPMENT and Fmeas (c0 ) · F(c0 , tmeas ) is the measured force response to c0
applied over the finite rise time tmeas . Using the approximation
The loading rate used in these tests was too low to allow for of Eq. (A3), (t) · Fmeas (c0 ) serves as the elastic function in the
the assumption of a step input in chest deflection. This obviously hereditary QLV integral of Eq. (3). An iterative approach may
has implications for the model development. In a viscoelas- be used to solve for (t) using the entire time history of a set
tic description of the response, the effect of a lower loading of experimentally derived data. The measured force is scaled
rate is that shorter time behavior, which is significant at higher by (t) at each point in time and the relaxation coefficients are
loading rates, may have decayed by the time the maximum dis- found that best fit the scaled data over the entire time history.
placement has occurred at the lower loading rate. A model fit- The scaling function is partitioned into ramp (load) and hold
ting strategy has been developed to estimate these shorter-time (relax) components:
constants.

Consider the definition of a QLV force response, F, to an  G(t · α)/G(t),


applied step displacement, c0 H(t) (where H(t) is the Heaviside  0 < t < tmeas (load)
step function): (t) = (A5)

G(tmeas · α)/G(tmeas )


tmeas ≤ (relax)
F(c0 , t) = G(t) · Felastic (c0 ) (A1)
where the factor α represents a loading rate greater than that
where G(0) = 1. Determining the true value of Felastic (c0 ) is not obtained experimentally. Funk, Martin, and Crandall (2001)
possible experimentally because a step displacement cannot be showed that, for one application (biaxial stretch of skin), a
applied over an infinitesimally small time. The elastic response 10-fold increase in strain rate (α = 0.1) can be adequately mod-
must be determined experimentally using a displacement applied eled using the above technique.
over a small, but finite, time interval. Thus, the instantaneous This iterative solution technique requires a numerical descrip-
force is approximated by tion of the model. Representing the reduced relaxation function
as a parallel assortment of Maxwell elements, a convolution
algorithm can be used to determine the force at time t j+1 given
F(c0 , tmin )
Felastic (c0 ) ≈ (A2) a past history of deformation to time t j and an incremental
G(tmin ) displacement:

where tmin is a theoretically attainable experimental rise time F(c, t + dt)

for the applied ramp displacement. If tmin represents the highest

loading rate environment in which the material model will be t+dt
∂Felastic
applied, then only longer-time behavior (i.e., force at t > tmin ) = (Gi Exp[−βi (t + dt − τ )] + G∞ ) dτ
0 ∂τ
is of concern and Eq. (A2) is an adequate approximation of the

instantaneous elastic response. Unfortunately, the limitations of t
∂Felastic
the loading apparatus used in the thorax characterization tests = Gi Exp(−βi dt) Exp[−βi (t − τ )] dτ
0 ∂τ
were such that the experimental rise time, tmeas , was on the order
of a time range of interest. The experimentally measured forces  
t
∂Felastic t+dt
may therefore underestimate the force at higher loading rates + G∞ dτ + Gi Exp[−βi (t + dt − τ )]
during the early force response (i.e., force at t ≤ tmeas ) because 0 ∂τ t

the force may have relaxed during the rise time tmeas . In order 
to approximate the elastic force response associated with higher ∂Felastic t+dt
∂Felastic
× dτ + G∞ dτ
loading rates, the experimental force must be scaled by some ∂τ t ∂τ
 CHEST BIOMECHANICS 311

= Exp(−βi dt) · Si (c, t) + S∞ (c, t) and

 elastic F∞ (c, t + dt) = F∞ (c, t) + G∞ · {[Felastic @(t + dt)]
Gi ∂F
+ (1 − Exp[−βi dt]) + G∞ dt ;
βi ∂t − [Felastic @(t)]} (A9)

i = 1, 2 , . . . , n (A6) The relaxation constants can then be determined by curve

fitting the numerical convolution over the entire force-time his-
tory. The strategy for determining the model coefficients was as
A first-order, forward difference approximation may be applied follows:
to the time derivative of the elastic force function and the updated
force response may be approximated by 1. The elastic model of Eq. (4) was fit to the elastic portion
of the posterior force–central chest deflection curve (i.e., the

n portion of the curve prior to the hold portion) using a non-
F(c, t + dt) = Fi (c, t + dt) + F∞ (c, t + dt); linear regression scheme that minimized the mean squared
i=1 error (MSE) between the function and the measured values.
2. The reduced relaxation function coefficients were determined
i = 1, 2 , . . . , n (A7) by minimizing the MSE between the calculated force-time-
deflection response and the entire measured force-time-
where the components of the force are deflection history. Note that (t) is a function of the reduced
relaxation function coefficients.
3. The elastic portion of the force-deflection curve was multi-
Fi (c, t + dt) plied by (t) to approximate the elastic response that would
be obtained if the test had been performed at approximately

Gi [1 − Exp(−βt)] an order of magnitude greater rate.
= Fi (c, t)Exp(−βt) + 4. The elastic model of Eq. (4) was fit to these scaled data using
βi
nonlinear regression.
 5. The model coefficients found using the scaled elastic data and
[Felastic @(t + dt)] − [Felastic @(t)] the reduced relaxation function coefficients found in step 2
× (A8)
dt are the final QLV model parameters.
312 R. KENT ET AL.

APPENDIX B—TEST DATA AND MODEL VALIDATION

Figure B1 Temporal data and force-deflection cross-plot for three different Figure B2 Temporal data and force-deflection cross-plot for three different
loading conditions for the large male subject (145). Note that duration of hold loading conditions for the small female subject (147). Note that duration of hold
portion was 77 seconds; only the first 0.5 seconds are shown in the top two plots. portion was 77 seconds; only the first 0.5 seconds are shown in the top two plots.
 CHEST BIOMECHANICS 313

Figure B3 Model validation results for all loading conditions and both subjects.
314 R. KENT ET AL.

APPENDIX C—PHOTOGRAPHS OF TEST SETUP

Figure C2 Photographs of small female test conditions.

Figure C1 Photographs of large male test conditions.

Copyright of Traffic Injury Prevention is the property of Taylor & Francis Ltd and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.