ArticlePDF Available

Abstract

A new biofidelity assessment system is being developed and applied to three side impact dummies: the WorldSID-alpha, the ES-2 and the SID-HIII. This system quantifies (1) the ability of a dummy to load a vehicle as a cadaver does, "External Biofidelity," and (2) the ability of a dummy to replicate those cadaver responses that best predict injury potential, "Internal Biofidelity." The ranking system uses cadaver and dummy responses from head drop tests, thorax and shoulder pendulum tests, and whole body sled tests. Each test condition is assigned a weight factor based on the number of human subjects tested to form the biomechanical response corridor and how well the biofidelity tests represent FMVSS 214, side NCAP (SNCAP) and FMVSS 201 Pole crash environments. For each response requirement, the cumulative variance of the dummy response relative to the mean cadaver response (DCV) and the cumulative variance of the mean cadaver response relative to the mean plus one standard deviation (CCV) are calculated. The ratio of DCV/CCV expresses how well the dummy response duplicates the mean cadaver response: a smaller ratio indicating better biofidelity. For each test condition, the square root is taken of each Response Comparison Value (DCV/CCV), and then these values are averaged and multiplied by the appropriate Test Condition Weight. The weighted and averaged comparison values are then summed and divided by the sum of the Test Condition Weights to obtain a rank for each body region. Each dummy obtains an overall rank for External Biofidelity and an overall rank for Internal Biofidelity comprised of an average of the ranks from each body region. Of the three dummies studied, the selected comparison test data indicate that the WorldSID-alpha prototype dummy demonstrated the best overall External Biofidelity although improvement is needed in all of the dummies to better replicate human kinematics. All three dummies estimate potential injury assessment with similar levels of Internal Biofidelity.
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760 Web: www.sae.org
SAE TECHNICAL
PAPER SERIES 2002-22-0024
Development of a New Biofidelity Ranking
System for Anthropomorphic Test Devices
Heather H. Rhule, Matthew R. Maltese, Bruce R. Donnelly and Rolf H. Eppinger
National Highway Traffic Safety Administration
Jill K. Brunner and John H. Bolte IV
Transportation Research Center, Inc.
Reprinted From: Stapp Car Crash Journal, Vol. 46 (November 2002)
(P-383)
46th Stapp Car Crash Conference
Pointe Verdra Beach, Florida
November 11-13, 2002
STAPP CAR CRASH JOURNAL
The journal of the John Paul Stapp Association
Papers presented at the 46th Stapp Car Crash Conference, November 2002
Published annually
The Stapp Association is comprised of an autonomous Stapp Advisory Committee and was founded in honor of John
Paul Stapp, pioneer researcher of human tolerance to crash decelerations. The Association is dedicated to the
dissemination of research findings in impact biomechanics, human tolerance to impact trauma, and crash injury
protection. This is achieved primarily through the annual Stapp Car Crash Journal and associated conference.
The scope of material contained herein includes new data on the biomechanics of injury and human tolerance, new
methods and tools to study the biomechanics of injury based on experimental and analytical studies. Publication of
these papers does not constitute endorsement or agreement with their content by the reviewers, The Stapp
Association, the publisher, or their members or staff. Opinions and data presented in these papers are the sole
responsibility of the authors.
Copyright Information
Except for papers authored solely by one or more U.S. Government employees in the course of their employment, the
materials contained herein are protected by copyright. Single copies of papers may be made for personal use as allowed
by national copyright laws. Any other form of copying or reproduction of material protected by copyright is prohibited
without the express written consent of The Stapp Association, 2901 Baxter Road, Ann Arbor, MI 48109-2150, USA.
Publication Information
The Stapp Car Crash Journal is a publication of the Stapp Association and is published and made available by the
Society of Automotive Engineers, Inc., 400 Commonwealth Drive, Warrendale, PA 15096-0001. Phone: (724) 776-4970,
Fax: (724) 776-1615. All papers contained in this journal are abstracted and indexed in the SAE Global Mobility
Database. A CD-ROM version of the papers included herein is also available from the publisher.
Copyright © 2002 by The Stapp Association
2901 Baxter Road
Ann Arbor, MI 48109-2150
http://www.stapp.org
Published by the Society of Automotive Engineers, Inc.
400 Commonwealth Drive
Warrendale, PA 15096-0001
Phone: (724) 776-4970, Fax: (724) 776-1615
http://www.sae.org
Printed in the United States of America
ISBN 0-7680-1134-5
SAE P-383
__________________________________
ABSTRACT A new biofidelity assessment system is being developed and applied to three side impact dummies: the
WorldSID-α, the ES-2 and the SID-HIII. This system quantifies (1) the ability of a dummy to load a vehicle as a cadaver does,
“External Biofidelity,” and (2) the ability of a dummy to replicate those cadaver responses that best predict injury potential,
“Internal Biofidelity.” The ranking system uses cadaver and dummy responses from head drop tests, thorax and shoulder
pendulum tests, and whole body sled tests. Each test condition is assigned a weight factor based on the number of human
subjects tested to form the biomechanical response corridor and how well the biofidelity tests represent FMVSS 214, side NCAP
(SNCAP) and FMVSS 201 Pole crash environments. For each response requirement, the cumulative variance of the dummy
response relative to the mean cadaver response (DCV) and the cumulative variance of the mean cadaver response relative to the
mean plus one standard deviation (CCV) are calculated. The ratio of DCV/CCV expresses how well the dummy response
duplicates the mean cadaver response: a smaller ratio indicating better biofidelity. For each test condition, the square root is taken
of each Response Comparison Value (DCV/CCV), and then these values are averaged and multiplied by the appropriate Test
Condition Weight. The weighted and averaged comparison values are then summed and divided by the sum of the Test
Condition Weights to obtain a rank for each body region. Each dummy obtains an overall rank for External Biofidelity and an
overall rank for Internal Biofidelity comprised of an average of the ranks from each body region. Of the three dummies studied,
the selected comparison test data indicate that the WorldSID-α prototype dummy demonstrated the best overall External
Biofidelity although improvement is needed in all of the dummies to better replicate human kinematics. All three dummies
estimate potential injury assessment with similar levels of Internal Biofidelity.
KEYWORDSbiofidelity, dummy, DCV/CCV, sled, injury, side impact, ranking
__________________________________
INTRODUCTION
As part of the National Highway Traffic Safety
Administration’s research program to upgrade side
impact crash protection, a new Biofidelity Ranking
System for evaluating anthropometric dummies is
being developed. To demonstrate the utility of the
method, this Biofidelity Ranking System has been
applied to three side impact dummies: the SID-HIII,
ES-2, and prototype WorldSID-α.
The Biofidelity Ranking System described in this
paper evaluates dummies based on two factors; (1)
the ability of a dummy to replicate human loading of
the crash test environment, termed “External
Biofidelity,” and (2) the accuracy with which a
dummy can duplicate the cadaver responses
necessary to predict human injury, termed “Internal
Biofidelity.” External Biofidelity (EB) is calculated
using measures external to the dummy and human
subject and Internal Biofidelity (IB) is calculated
using the dummy and human subject instrumentation
most appropriate for the individual dummy design to
predict injury potential.
The three dummies were subjected to a battery of
side impact tests for which cadaver response data was
available (ISO 1999; Maltese et al. 2002; Bolte et al.
2000; Appendix A). Table 1 shows the matrix of
tests selected for comparison of the dummy responses
to cadaver responses. The test procedures
documented in the above-cited references were
followed for all dummy testing.
Stapp Car Crash Journal, Vol. 46 (November 2002), pp. 477-512
Development of a New Biofidelity Ranking System for
Anthropomorphic Test Devices
Heather H. Rhule, Matthew R. Maltese, Bruce R. Donnelly, and Rolf H. Eppinger
National Highway Traffic Safety Administration
Jill K. Brunner and John H. Bolte IV
Transportation Research Center, Inc.
2002-22-0024
Rhule et al./ Stapp Car Crash Journal 46 (November 2002)
Table 1. Bio Rank Test Matrix and References
Test Condition Test Name Reference
200mm Head Drop Onto Rigid Surface Head Test 1 ISO TR 9790-Head Test 1
7.2g Restrained Sled Neck Test 1 ISO TR 9790-Neck Test 1 & Shoulder Test 3
12.2g Restrained Sled Neck Test 3 ISO TR 9790-Neck Test 3 & Shoulder Test 3
4.5 m/s Rigid Shoulder Pendulum Shoulder Test 1 ISO TR 9790-Shoulder Test 1
4.4 m/s Padded Shoulder Pendulum NHTSA Shoulder Test Bolte et al. 2000
4.3 m/s Rigid Thorax Pendulum Thorax Test 1 ISO TR 9790-Thorax Test 1
8.9 m/s Flat Rigid Wall HS FR Maltese et al. 2002
8.9 m/s Flat Padded Wall HS FP Maltese et al. 2002
6.7 m/s Flat Rigid Wall LS FR Maltese et al. 2002
6.7 m/s Flat Padded Wall LS FP Maltese et al. 2002
6.7 m/s Rigid Abdomen Offset LS RAO Maltese et al. 2002
6.7 m/s Rigid Pelvis Offset LS RPO Maltese et al. 2002
Each test condition is assigned a weight factor based
on 1) the number of human subjects tested to form
the biomechanical response corridor and 2) how well
the biofidelity tests represent FMVSS 214, FMVSS
201 and SNCAP crash environments. For each
response requirement, the cumulative variance of the
dummy response relative to the mean cadaver
response (DCV) and the cumulative variance of the
mean cadaver response relative to the mean plus one
standard deviation (CCV) are calculated. The ratio of
DCV/CCV expresses how well the dummy response
duplicates the mean cadaver response: a smaller ratio
indicating better biofidelity.
The system allows both a comparison of dummy
response to cadaver response and a comparison
among two or more dummies. An “absolute” ranking
scale does not exist at this time and the resulting
dummy ranks are relative; however, the ranks do
provide a sense of the “number of standard deviations
away” from the mean human response. This
Biofidelity Ranking System is a work in progress and
the results of its application are dependent upon the
tests selected for the evaluation and the weight factor
assigned to the tests.
METHODS
Figure 1 shows a flowchart to graphically summarize
the Biofidelity Ranking System. Biofidelity tests
were selected such that each pertinent body region
was impacted. The Biofidelity Ranking System
presented here includes the tests listed in Table 1;
however, additional tests may be incorporated as the
system evolves. Five component tests were included
(ISO 1999) along with a padded shoulder pendulum
impact (Bolte et al. 2000) and six whole-body sled
test configurations (Maltese et al. 2002). This set of
tests was felt to be comprehensive and cost effective.
Limited resources made it impractical to include any
more tests at this time.
Figure 1. Flowchart summarizing the Biofidelity Ranking
System.
Dummies Humans
Biofidelit
y
Tests
Dummy Impact
Biofidelity Rank
Dummy Injury
Criteria Fidelity
Ran
k
Data
Processing
Test Condition
Weight
Response
Measurement
Comparison Value
Standard
Biomechanical
Corridors Subject
Score
Test
Severity
Score
Rhule et al./ Stapp Car Crash Journal 46 (November 2002)
External Biofidelity
External Biofidelity ranks include both an overall
score and individual scores for the head/neck,
shoulder, thorax, abdomen and pelvis. The scores for
each body region give additional information as to
how the overall rank was achieved for a particular
dummy. In other words, the ranks of the body
regions can identify strong and weak areas of a
dummy’s biofidelity.
External Biofidelity ranks are based on a comparison
of response measurements made externally to the
dummy and human subject. For example, only the
pendulum force measurements are used for pendulum
tests. For NHTSA sled tests, only the impact load
wall force measurements are used. For Neck Tests 1
and 3, four displacements and one angle are
measured via video motion analysis. These
measurements are relevant to External Biofidelity
because they reflect the dummy’s ability to replicate
human loading of a vehicle in a crash. Table 2 shows
the measurements used in each test condition which
comprise the External Biofidelity ranks for each body
region. Because the human body behaves as an
interdependent system, the response of one body
region will influence the response of neighboring
body regions. Thus, the body regions are given equal
importance and the overall External Biofidelity rank
is the average of the EB ranks of each of the body
regions.
Internal Biofidelity
An Internal Biofidelity rank is calculated for each
body region that has an associated potential injury
criterion. For NHTSA purposes, the head, thorax,
abdomen and pelvis were given IB ranks. The
overall dummy Internal Biofidelity is the average of
the IB ranks of each of the body regions.
As the three dummies do not have identical
measurement capabilities, their injury prediction
capabilities are different. For example, the SID-HIII
is an acceleration-based dummy and the ES-2 and
WorldSID-α dummies have the capability of
Table 2. Response Measurements and Test Conditions Used for Each External Biofidelity Calculation
Test Conditions
N
eck
Test
1
Neck
Test
3
Shoulder
Test 1
NHTSA
Shoulder
Test
Thorax
Test 1
HS
FR
HS
FP
LS
FR
LS
FP
LS
RAO
LS
RPO
Head/Neck External Biofidelity
Peak Horizontal Displacement of
Head cg Relative to T-1 X
Peak Vertical Displacement of
Head cg Relative to T-1 X
Peak Horizontal Displacement of
Head cg Relative to Sled X
Peak Flexion Angle X X
Shoulder External Biofidelity
Pendulum Force X X
Peak Horizontal Displacement of
T-1 Relative to Sled X
Thorax External Biofidelity
Pendulum Force X
Thorax Plate Force X X X X X X
Abdomen External Biofidelity
Abdominal Plate Force X X X X X X
Pelvis External Biofidelity
Pelvis Plate Force X X X X X X
Rhule et al./ Stapp Car Crash Journal 46 (November 2002)
measuring thorax deflections. Thus, the SID-HIII
Internal Biofidelity rank of the thorax (IBthorax)
consists of the lower spine lateral acceleration and
struck side upper and lower rib lateral accelerations
used in currently regulated Thoracic Trauma Index
(TTI) (49 CFR 572.214) calculations compared to
those response corridors of the human subjects.
Similarly, the IBthorax ranks of the ES-2 and
WorldSID-α dummies consist of the upper spine
resultant acceleration and upper and lower thorax
deflections used in potential injury criteria that
combine acceleration and deflection compared to
those response corridors of the human subjects. In
addition, the SID-HIII dummy has no abdominal
instrumentation and the ES-2 dummy has the
capability of measuring abdominal loads, however no
correlation between dummy abdomen loads and that
of cadavers exists. Thus, the SID-HIII and ES-2
dummies do not have Internal Biofidelity ranks for
the abdomen. As the WorldSID-α dummy has the
capability of measuring abdominal deflections, its
IBabdomen rank consists of those channels associated
with an injury criterion of deflection compared to
those response corridors of the human subjects.
Table 3 shows the dummy and cadaver measurements
and test conditions that comprise the Internal
Biofidelity ranks for the body regions of each
dummy.
Bio Rank Characterization
The calculation of the External Biofidelity and
Internal Biofidelity ranks follow identical procedures
and involve two factors: response measurement
comparisons and Test Condition Weights. Each rank
is calculated using Equation 1.
=
=
=
=m
j
j
m
j
n
k
kj
j
V
n
R
V
B
1
1
1
,
*
(1)
where
R = Response Measurement Comparison Value
(DCV/CCV)
V = test condition weight
B= biofidelity rank
j = test condition
k = response measurement
m = number of test conditions
n = number of response measurements per test
condition
Each response measurement of each test condition is
compared to its corresponding biomechanical
corridor to obtain a Response Measurement
Comparison Value, R. The square root of R is then
taken. (If multiple tests of the same test condition are
performed on a dummy, the multiple R values are
averaged for a single response requirement.) The R
values for all of the measurements of a single test are
then averaged and multiplied by the corresponding
Test Condition Weight, V. Finally, all of the
weighted average test condition response values are
summed and divided by the sum of the Test
Condition Weights to obtain a biofidelity rank, B.
Response Measurement Comparison Values. The
essence of the biofidelity rank lies in the comparison
of each dummy response to its corresponding mean
human subject response. In Equation 1, R is the ratio
of the cumulative variance between the dummy
response and mean human response (DCV) over the
cumulative variance between the mean human
response and the mean plus one standard deviation
(CCV) (Morgan et al. 1986). It is important to
include the CCV portion of the ratio because if the
variability in the human response is large, it is not
reasonable to demand strict adherence of the
dummy’s response to the mean cadaver response.
Figure 2 illustrates the definitions of Dummy
Cumulative Variance (DCV) and Cadaver
Cumulative Variance (CCV) for any given response
measurement. The lower the DCV/CCV ratio, the
closer the dummy response is to that of the mean
cadaver. The ratio values are more easily interpreted
in terms of multiples of standard deviations rather
than variances. Thus, once R has been determined
for a given measurement, its square root is calculated
in order to establish a “cumulative standard
deviation” rather than cumulative variance. For those
response requirements that consist of only a peak
measurement, the square of the difference between
the dummy and mean cadaver peaks was used as the
dummy variance, rather than dummy cumulative
variance. Likewise, the square of the difference
between the cadaver upper corridor peak and the
mean cadaver peak response was used as the cadaver
variance, rather than cadaver cumulative variance.
Rhule et al./ Stapp Car Crash Journal 46 (November 2002)
Table 3. Response Measurements and Test Conditions used for each Internal Biofidelity Rank Calculation
Test Conditions
Dummies Response Measurements
Head
Test 1 HS FR HS FP LS FR LS FP LS RAO LS RPO
Head Internal Biofidelity
ES-2 X
WSID X
SID-HIII
Peak Head Resultant Acceleration
X
Thorax Internal Biofidelity
T-1 Resultant Acceleration X X X X X X
Upper Thoracic Lateral Deflection X X X X NC X
ES-2
Lower Thoracic Lateral Deflection X X X X X X
T-1 Resultant Acceleration X X X X * X
Upper Thoracic Lateral Deflection X X X X * NC X
WSID
Lower Thoracic Lateral Deflection X X X X * X
T-12 Lateral Acceleration X X X X X X
Struck-side Upper Rib Acceleration X X X X X X
SID-HIII
Struck-side Lower Rib Acceleration
X X X X X X
Abdomen Internal Biofidelity
ES-2 N/A
WSID Mid-Abdominal Deflection NC DL X X * X
SID-HIII N/A
Pelvis Internal Biofidelity
ES-2 X X X X X X
WSID X X X X * X
SID-HIII
Pelvis Lateral Acceleration
X X X X X X
* - WorldSID dummy not tested in LS RAO condition
DL - Data Loss
NC - No corridor
N/A - Not applicable
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Standard Corridors
For each response measurement required in the
biofidelity tests prescribed here, there exists a
biomechanical performance corridor developed from
human subject responses. Because the biofidelity
tests were conducted by various researchers, the
definition of the various response corridors is not
consistent. For example, some corridors are defined
as the mean human response plus or minus one
standard deviation from the mean. Other corridors
are defined to envelop the entire data set. In order for
the cumulative variance calculations to be consistent,
a “standard” corridor definition was established for
the Bio Rank System. The standard corridor, for
purposes of calculating the R values, was defined to
be the mean human response plus or minus one
standard deviation from the mean. For those
performance corridors whose definition did not
match the standard or was unknown, a standard
corridor was established.
Head Test 1
According to ISO TR 9790, seven cadavers were
dropped from various heights ranging between 139
and 188 mm. The head was assumed to behave as a
linear spring-mass system, where the peak
acceleration of the mass is directly proportional to the
impact velocity. The biofidelity corridor reported in
ISO TR 9790 was drawn to envelop the entire data
set as shown in Figure 3. Since it was necessary to
define a standard corridor, this same linear
assumption was followed, and the cadaver peak head
resultant accelerations were interpolated to the 2 m/s
impact velocity, which is the impact velocity used for
dummy testing. The cadaver responses at the 2m/s
impact velocity were averaged and the standard
corridor was determined from the mean +/- one
standard deviation, as shown in Figure 3. Appendix
B shows the original cadaver responses and
corresponding drop heights, with the interpolated
data and new standard corridor information. It is
important to note that the dummy peak resultant
acceleration was calculated at the head cg, while that
of the cadaver was measured at the non-impacted
side of the head.
Neck Test 1
ISO TR 9790 reported biofidelity response corridors
as the mean +/- one standard deviation for peak
horizontal and peak vertical displacements of the
head cg with respect to T-1 and peak head flexion
angle. However, there was no explanation for how
the corridor for peak horizontal displacement of T-1
relative to the sled was defined. Thus, this corridor
was also assumed to be defined as the mean +/- one
standard deviation.
Neck Test 3
The ISO TR 9790 response corridors for peak
horizontal displacement of the head cg with respect
to the sled and peak head flexion angle were derived
using the results of one cadaver. It was assumed that
these corridors were equivalent to a one-standard
deviation spread for the purposes of having consistent
corridor definitions.
=
=n
t
tCVCCV
0
2
)(
CV(t)
Time
Mean Cadaver
Response
Mean + 1 Standard
Deviation
Force, Acceleration or Deflection
=
=n
t
tCVCCV
0
2
)(
CV(t)
Time
Mean Cadaver
Response
Mean + 1 Standard
Deviation
Force, Acceleration or Deflection
2
0
)(tDVDCV
n
t
=
=
Time
Mean Cadaver
Response
Dummy Response
Force, Acceleration or Deflection
DV(t)
2
0
)(tDVDCV
n
t
=
=
Time
Mean Cadaver
Response
Dummy Response
Force, Acceleration or Deflection
DV(t)
Figure 3. Head Test 1 cadaver results with standard an
d
ISO TR 9790 response corridors.
Figure 2. Dummy Cumulative Variance (DCV, top) an
d
Cadaver Cumulative Variance (CCV, bottom).
0
50
100
150
200
0 0.5 1 1.5 2 2.5
Impact Velocity (m/s)
ISO TR 9790 'Standard'
Peak Head Resultant Acceleration (G)
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Shoulder Test 1
The force versus time history corridor was developed
from three cadaver responses, one of which was
impacted at 15 degrees forward of lateral rather than
a straight lateral impact. ISO TR 9790 does not
provide an explanation of the development of the
corridor. Thus, based on the plot of force versus time
data and the corridor, it was assumed that the corridor
was defined as the mean +/- two standard deviations
as approximately 95% of the data appeared to be in
the corridor. As the electronic data was not available
for analysis, in order to establish the standard
corridor, the mean was determined from the average
of the upper and lower corridors and one standard
deviation was determined from the mean.
Thorax Test 1
ISO TR 9790 states that the force versus time
response corridor was drawn around the normalized
cadaver curves and then shifted upward 700 N to
account for muscle tone. Thus, based on the plot of
force versus time data and the corridor, it was
assumed that the corridor was defined as the mean +/-
two standard deviations as approximately 95% of the
data appeared to be in the corridor. As the electronic
data was not available for analysis, in order to
establish the standard corridor, the mean was
determined from the average of the upper and lower
corridors and one standard deviation was determined
from the mean.
Data Processing
In order to make a fair comparison of the dummy
data to the cadaver data, similar filters had to be
applied to both sets of data. In the cases where
electronic cadaver data was not available, the dummy
data was filtered at the same filter class as that
applied to the referenced cadaver data. The External
Biofidelity channels were all filtered at SAE Class
1000, with two exceptions: the pendulum force from
the NHTSA Shoulder Test was filtered at SAE Class
180 and the pendulum force of Thorax Test 1 was
filtered using a 100 Hz FIR filter. The FIR filter was
applied to the dummy data in order to compare to the
cadaver data from ISO TR 9790. For the data that
was determined via video motion analysis, high-
speed digital cameras with a recording rate of 1000
frames per second were used. As only peak
measurements were determined via motion analysis,
filter information is not applicable.
As injury criteria call for specific filters to be used,
the Internal Biofidelity channels were filtered
accordingly, wherever possible. Since the cadaver
data from Head Test 1 was not available for filtering,
it was necessary to filter the dummy head
acceleration data at SAE Class 1000. This allowed a
direct comparison between the cadaver responses of
ISO TR 9790 and the dummy responses. The
Thoracic Trauma Index, referenced in FMVSS 214,
calls for the 100 Hz FIR filter to be used on the upper
and lower rib lateral accelerations and lower spine
lateral acceleration. For a combination of upper
spine resultant acceleration and upper and lower
chest deflections, SAE J211 filters were used.
Abdominal deflections were filtered at SAE Class
600. FMVSS 214 specifies that a 100 Hz FIR filter
be used on lateral pelvis acceleration.
All data, except for the T-12, rib, and pelvis
accelerations, were sub-sampled to 3200 Hz,
fulfilling the Nyquist criterion of a sampling rate of at
least two times the maximum frequency content of
the data. The lower spine, rib and pelvis
accelerations were sub-sampled to 1600 Hz as part of
the FMVSS 214 requirement for injury criteria
calculations. For all NHTSA tests, the force and
acceleration data was limited to include all data from
the onset of the event up to 10% of the peak of the
cadaver mean, occurring after the peak. Chest
deflection data from NHTSA sled tests was limited to
include data up to 20% of the peak of the cadaver
mean, occurring after the peak, as the cadaver chest
band data did not always return to 10% of the peak.
Data from Shoulder Test 1 and Thorax Test 1 were
limited to include all data up to the specified length
of the ISO TR 9790 corridor (In cases where the ISO
corridor upper and lower boundaries were of different
lengths, the length of the longer boundary was cut to
that of the shorter boundary). All other tests only
required peak measurements and the amount of data
included is not applicable.
For all NHTSA sled tests, “time-zero,” or the onset of
the event, was defined to be the last zero crossing
before the maximum value for a selected channel.
The time-zero established for the selected channel
was then applied to the other data channels collected
during that particular test. For flat wall tests, the
selected channel for defining time-zero was the
thorax load plate force of the impact wall. For offset
tests, the selected channel was the offset load plate
force. For pendulum tests, the onset of the event was
indicated by contact with the pendulum, where an
event channel signified contact. For Head Test 1 and
Neck Tests 1 and 3, time-zero did not need to be
defined, as the response requirements only called for
peak measurements.
Test Condition Weights. The Test Condition Weights,
V, rate the importance of each test condition of the
Biofidelity Ranking System, with higher V values
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
being of greater importance. The Test Condition
Weights are based on two components: (1) the
number of human subjects from which the pertinent
biomechanical corridors were developed, termed
“Subject” score, and (2) the biofidelity test relevance,
termed “Test Relevance” score. The Test Condition
Weights are determined from Equation 2, where the
Test Relevance score is given twice the importance
of the Subject score. Both components of V were
normalized to a maximum of ten points so that the
Test Condition Weights would be on a scale from 1
to 10.
)(*67.0)(*33.0 ncetestrelevasubjectV += (2)
Subject Score
The Subject score for each test condition is
equivalent to the number of human subjects from
which the biomechanical corridor was developed, up
to a maximum of ten. For a given test condition,
response corridors exist for several channels. As not
all of the human subjects were always included in the
development of each corridor, the maximum number
of subjects utilized for creating the corridors was
used in determining the Subject score. The more
human subjects used in a test, the higher the Subject
score. Table 4 lists the Subject score for each test
condition.
Table 4. Subject Scores for Each Biofidelity Test
Test Name Subject Score
Head Test 1 7
Neck Test 1 9
Neck Test 3 1
Shoulder Test 1 3
NHTSA Shoulder Test 6
Thorax Test 1 7
HSFR 7
HSFP 8
LSFR 5
LSFP 5
LSRAO 3
LSRPO 3
Test Relevance Score
It is important that a dummy is biofidelic in test
conditions that replicate the intended use
environment. The NHTSA intends to utilize side
impact dummies in research and regulatory crash
testing that is represented by FMVSS 214, SNCAP
and FMVSS 201 Pole testing. The Test Relevance
score indicates how well each biofidelity test
represents FMVSS 214, SNCAP and FMVSS 201
Pole crash tests. The biofidelity tests whose dummy
responses are within the range of dummy responses
in these crash tests receive higher Test Relevance
scores. In other words, more weight is given to the
biofidelity tests whose dummy responses are similar
to those of the crash tests.
In order to establish a basis for the magnitude of
responses observed in these crash tests, peak force,
acceleration and deflection measurements of the SID-
HIII and ES-2 dummies were compared among
biofidelity tests and crash tests. For the SID-HIII, all
measurement data from the 2001 NHTSA Crash
database were utilized. The ES-2 dummy had been
used in SNCAP and Pole test environments for
research purposes, and those data were used in order
to study chest deflection and abdomen force
responses. For each biofidelity test, pertinent peak
responses were plotted for both the SID-HIII and ES-
2 dummies. In addition, the minimum-to-maximum
ranges of the same dummy responses from the crash
tests were plotted for each of the SID-HIII and ES-2
dummy responses. Each dummy’s crash test
response ranges were combined to form the dummy’s
“vehicle test data range.” The dummy responses
measured in each biofidelity test were compared to
the vehicle test data range to ascertain the Test
Relevance score.
Table 5 shows a summary of how the Test Relevance
scores were achieved, including the dummy
responses and criteria used for awarding points. For
each test condition, the Test Relevance score is the
number of points awarded divided by the total
number of points possible. Figures 4-9 show the
comparison of biofidelity and crash test responses.
The time history of the resultant acceleration of the
head cg was used as a tool for scoring the relevance
of Head Test 1 because this measure is used in
calculating the Head Injury Criterion for SNCAP and
Pole tests (49 CFR 571.201). Figure 4 shows the
comparison of responses from crash tests and Head
Test 1.
Rhule et al./ Stapp Car Crash Journal 46 (November 2002)
Table 5. Test Relevance Score Summary
Dummy Points
Test Condition Response Channels SID-HIII ES-2 Criteria Possible Awarded Figure
Head Test 1
Head Resultant
Acceleration & Time
Duration
X X
Point awarded if:
-head acceleration and time
duration are equal to or less
than maximum values measured
during vehicle crash test
OR
-head acceleration is < 20G
2 2 4
T-1 Lateral Acceleration X
Point awarded if response is
equal to or below maximum
value measured during vehicle
crash test
Upper Neck Fy X X
Upper Neck Fz X X
Lower Neck Fy X
Lower Neck Fz X
Shoulder Fx X
Shoulder Fy X
Upper Neck Mx X X
Upper Neck Mz X X
Lower Neck Mx X
Lower Neck Mz X
Point awarded if response is
within +/- range of vehicle test
data
Neck Test 1
Head Resultant
Acceleration & Time
Duration
X X Same as Head Test 1
17 16 4, 5, 6
Neck Test 3 Same as Neck Test 1 17 16 4, 5, 6
Upper Rib Acceleration X
Lower Rib Acceleration X
Lower Spine Acceleration X
Upper Rib Deflection X
Shoulder Test 1
Lower Rib Deflection X
Point awarded if response was
equal to or below maximum
value measured during vehicle
crash tests
5 5 7
NHTSA
Shoulder Tes
t
Same as Shoulder Test 1 5 5 7
Thorax Test 1 Same as Shoulder Test 1 5 5 7
Upper Rib Acceleration X X
Lower Rib Acceleration X X
Lower Spine Acceleration X X
Pelvis Acceleration X X
Abdomen Load X
Upper Rib Deflection X
HS FR
Lower Rib Deflection X
Point awarded if response is
equal to or below the maximum
value measured in vehicle tests
11 2 8
HS FP Same as HS FR 11 10 8
LS FR Same as HS FR 11 8 8
LS FP Same as HS FR 11 11 9
LS RAO Same as HS FR 11 7 9
LS RPO Same as HS FR 11 9 9
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
0
100
200
300
400
500
600
700
800
900
0 20 40 60 80 100 120 140 16 0 180 200
Duration (msec)
Head CG Resultant Accelerati on (g
)
SID-HIII Neck Test 1 SID-HIII Neck Test 3 SID-HIII Vehicle Tests
ES-2 Neck Test 1 ES-2 Neck Test 3 ES-2 Vehicle Tests
ES-2 Head Test 1 SID-HIII Head Test 1
Figure 4. Head resultant acceleration and duration for Head
Test 1 and Neck Tests 1 and 3 compared to vehicle crash tests.
-250
-200
-150
-100
-50
0
50
100
150
200
250
Moment (Nm)
N
eck Test 1
N
eck Test 3 Vehicle Test Data Rang e
Upper
N
eck Mx Upper
N
eck Mz Lowe r
N
eck Mx Lower
N
eck Mz
SID-HIII
SID-HIII
ES-2 ES-2
ES-2
ES-2
Figure 6. Neck Tests 1 and 3 moment comparison to vehicle
crash tests.
0
25
50
75
100
125
150
175
200
225
250
Acceleration (g)
0
10
20
30
40
50
60
70
Load x 10-2 (N) and Deflec tion (mm)
Low Speed Flat Rigid High Speed Flat Padded
High Speed Flat Rigid Vehicle Test Data Range
SID- HIII SID -HIII
SID- HIII
SID- HIII
ES-2
ES-2
ES-2
ES-2 ES-2
ES-2 ES-2
Pelvis
Acceleration
Lower R ib
Acceleration
Lower R ib
Deflection
Lower S pine
Acceleration
Upper Rib
Acceleration
Upper Rib
Deflection
Abdomen
Load
Figure 8. NHTSA HS FR, HS FP and LS FR sled tes
t
comparison to vehicle crash tests.
0
25
50
75
100
125
150
175
200
Acceleration (g
)
0
10
20
30
40
50
60
Load x 10-2 (N) and Deflection (mm
)
Low Speed Flat Padded Low Speed Rigid Abdomen Offset
Low Speed Rigid Pelvis Offset Vehicle Test Dat a Range
Pelvis
Acceleration
Lower R ib
Acceleration
Lower R ib
Deflection
Lower S pine
Acceleration
Upper Rib
Acceleration
Upper Rib
Deflection
Abdomen
Load
SID-HIII SID-HIII SID-HIII
SID-HIII
ES-2
ES-2
ES-2
ES-2
ES-2 ES-2
ES-2
Figure 9. NHTSA LS FP, LS RAO, and LS RPO sled test
comparison to vehicle crash tests.
0
25
50
75
100
125
150
175
200
Acceleration (g)
0
10
20
30
40
50
60
Deflection (mm)
Shoulder Test 1 Vehicle Test Dat a Range
NHTSA Shoulder Test Thorax Test 1
Lower Rib
Acceleration
Lower Rib
Deflectio n
Lower Sp ine
Acceleration
Upper Rib
Acceleration
Upper Rib
Deflection
ES-2 ES-2 ES-2
ES-2
ES-2
Figure 7. Pendulum test response comparison to vehicle cras
h
tests.
-50
-25
0
25
50
75
100
Acceleration (g)
-30
-20
-10
0
10
20
30
40
50
Load x 10-2 (N)
Neck Test 1 Neck Test 3 Ve hicle T est Data Range
Upper
Neck Fy
T-1
Y-Acce l
Upper
Neck Fz
Lower
Neck Fy
Lower
Neck Fz
Shldr
Fx
Shldr
Fy
ES-2
ES-2
ES-2
ES-2 ES-2
ES-2
ES-2
SID-HIII
SID-HIII
Figure 5. Neck Tests 1 and 3 acceleration and loa
d
comparison to vehicle crash tests.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
As shown in Table 5, two points were possible for the
Head Test 1 Test Relevance score: one from the SID-
HIII response and one from the ES-2 response of
Head Test 1. One point was awarded if (a) the peak
resultant head acceleration was below 20g or (b) the
peak acceleration and the duration of the impact were
both less than or equal to the peak acceleration and
duration of the crash response. Duration of impact
was defined as the time between the initial 10% of
the peak head acceleration and the final 10% of the
peak head acceleration. Both dummies’ acceleration
responses were less than the peak acceleration and
duration of the crash response; therefore, Head Test 1
received 2/2 (or 1) for its Test Relevance score.
Some of the response requirements included in Neck
Tests 1 and 3 are measured via video analysis and are
not available from the crash tests. These
measurements could not be used as a comparison
with the crash data for assigning Test Relevance
scores. Those measurements from biofidelity and
crash tests that can be compared are shown in Table
5. Figures 4, 5 and 6 show the comparison of
responses from crash tests, Neck Test 1 and Neck
Test 3.
The pendulum biofidelity test requirements consist of
pendulum force and either shoulder deflection for
shoulder impacts or upper spine acceleration for
thorax impacts. In order to compare with vehicle test
data, additional channels were collected during
pendulum tests with the ES-2 (due to time
constraints, the additional channels were not
collected on the SID-HIII). The four required
channels in the current FMVSS 214 were collected,
along with upper and lower rib deflections. Since
pelvis accelerations were not affected in the shoulder
and thorax pendulum tests, it was not used in
determining the Test Relevance score. Figure 7
shows the comparison of acceleration and deflection
responses for the crash tests, Shoulder Test 1, Thorax
Test 1, and the NHTSA Shoulder Test. The four
required channels in the current FMVSS 214 were
recorded in both the SID-HIII and ES-2 dummies,
along with rib deflections and abdomen forces in the
ES-2 for the NHTSA sled tests. Figures 8 and 9
show the comparison of responses from crash and
NHTSA sled tests.
The HSFP and LSFP test conditions received high
Test Relevance scores, indicating good representation
of vehicle crash test conditions. The HSFR test
condition received a very low Test Relevance score,
indicating poor representation of vehicle crash test
conditions.
Table 6 summarizes the final Test Condition Weights
for all test conditions, using the information from
Tables 4 and 5, Figures 4-9 and Equation 2.
Table 6. Test Condition Weight Summary
Test Name
Test
Relevance
Score
Normalized
Test
Relevance
Score
Weighted
Normalized
Test
Relevance
Score
Subject
Score
Weighted
Subject
Score
Test
Condition
Weight
Head Test 1 1.00 10 6.70 7 2.31 9
Neck Test 1 0.94 9 6.03 9 2.97 9
Neck Test 3 0.94 9 6.03 1 0.33 6
Shoulder Test 1 1.00 10 6.70 3 0.99 8
NHTSA Shoulder Test 1.00 10 6.70 6 1.98 9
Thorax Test 1 1.00 10 6.70 7 2.31 9
HS FR 0.18 2 1.34 7 2.31 4
HS FP 0.91 9 6.03 8 2.64 9
LS FR 0.73 7 4.69 5 1.65 6
LS FP 1.00 10 6.70 5 1.65 8
LS RAO 0.64 6 4.02 3 0.99 5
LS RPO 0.82 8 5.36 3 0.99 6
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
METHODOLOGY APPLIED TO SIDE IMPACT
DUMMIES
Although a Biofidelity Rank of B<1.0 would seem to
be a desirable goal, the Biofidelity ranks have not
been shown to signify an absolute level of
biofidelity. The ranks are best interpreted as an
indication of relative biofidelity among various
dummies for the tests selected and the Test Condition
Weights assigned. As the Bio Rank System is still
under development, these preliminary results for the
three side impact dummies are presented to
demonstrate the utility of the methodology.
Rankings
If only the measurements made externally to the
dummy are considered, the External Biofidelity ranks
of the dummies are as shown in Table 7. The
WorldSID-α has the best overall External Biofidelity
rank. The SID HIII has the best rank for the
head/neck and, in fact, it is the best rank in the table.
ES-2 has the best ranks for the shoulder and pelvis.
WorldSID-α has the best rank for the thorax and the
abdomen ranks are fairly comparable between the
ES-2 and WorldSID-α.
Table 7. Preliminary External Biofidelity Ranks
External
Biofidelity SID-HIII ES-2 WorldSID
Overall Rank 3.8 2.7 2.5
Head/Neck Ran
k
1.0 3.7 2.1
Shoulder Rank 5.1 1.4 2.1
Thorax Rank 6.1 3.2 2.1
Abdomen Rank 3.0 2.5 2.2
Pelvis Rank 3.8 2.7 3.8
If the channels that are used to predict injury for each
dummy are considered, the dummies achieve Internal
Biofidelity ranks as shown in Table 8. As the SID-
HIII does not have abdominal instrumentation and no
cadaver data exists for abdomen force, which the ES-
2 can measure, these dummies do not have IBabdomen
ranks. In order to make a valid comparison among
the dummies, the overall IB ranks were averaged
without the IBabdomen ranks. The overall IB rank of
the WorldSID-α is shown for completeness. Without
the IBabdomen ranks included, all three dummies have
overall Internal Biofidelity ranks below 2.0. The
WorldSID-α has the best Internal Biofidelity ranks
for the head and thorax, while the ES-2 has the best
rank for the pelvis. In all cases, the difference in
ranks among the three dummies is quite small.
Table 8. Preliminary Internal Biofidelity Ranks
Internal
Biofidelity SID-HIII ES-2 WorldSID
Overall Rank
with abdomen n/a n/a 1.6
Overall Rank
without abdomen 1.9 1.6 1.5
Head Rank 1.1 1.0 0.8
Thorax Rank 2.2 1.7 1.4
Abdomen Rank n/a n/a 1.8
Pelvis Rank 2.5 2.1 2.4
Bio Rank Validation
In order to truly validate the Biofidelity Ranking
System a single human cadaver should be compared
to response corridors developed from a normally
distributed population of human cadavers for each
test condition. Because a single cadaver could never
be subjected to multiple tests and because the
existing population of cadaver test subjects is
relatively small, this is not possible. In order to
provide a sense of what the Bio Rank values mean a
validation exercise was performed using one test
condition and the set of cadaver responses that made
up its associated response corridors.
The High Speed Padded Flat Wall sled test condition
was selected because it received the highest Test
Condition Weight of the six NHTSA sled tests, with
seven human subjects making up the response
corridors for the impact wall force plates, and 90% of
the dummy responses in biofidelity tests within the
range of crash tests. Six “validation” response
corridors were established from the normalized
responses of the seven cadavers. The techniques
developed by Maltese (Maltese et al. 2002) were used
in creating these demonstration corridors. Cadaver
number one was included in all corridors as the
“master” response upon which the corridors were
calculated. The corridors were created by averaging
all combinations using cadaver number one and five
of the other subjects; i.e., 1 through 7 without number
2, 1 through 7 without number 3, and so forth. This
was done for the thorax, the abdomen and the pelvis
force plate responses. The cadaver that was not
included in the creation of the validation corridor was
then used as the single subject to be ranked; i.e.,
number 2, then number 3, etc. Response Comparison
Values were obtained, and then square roots taken,
for each of the six cadavers, 2 through 7, for the
thorax, the abdomen and the pelvis force plate
responses. Thus, eighteen validation corridors were
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
created and eighteen Rvalidation values were obtained
(see corridor plots and Rvalidation values in Appendix
C and in Table 9). From these Rvalidation values,
External Biofidelity ranks were determined for the
six separate individuals and are shown in column five
of Table 9.
Since External Biofidelity Ranks for cadavers
compared to other cadavers have been calculated, all
presumably from a normally distributed population, it
is expected that the ranks should be B<1.0 for 68% of
the trials: approximately four out of six of the
cadavers. Since none of the cadavers ranked below
Table 9. External Biofidelity Validation Ranks
R Value
Individual
Thorax Abdomen Pelvis
External
Biofidelity
Rank
Cadaver 2 1.1 1.5 1.0 1.2
Cadaver 3 0.8 0.8 1.5 1.0
Cadaver 4 2.2 1.5 1.6 1.7
Cadaver 5 1.0 0.7 1.2 1.0
Cadaver 6 0.8 2.1 2.1 1.7
Cadaver 7 1.1 1.2 2.4 1.5
1.0, the results of the validation test indicate that
either the cadaver population is not normally
distributed, or that seven subjects was not a large
enough sample to demonstrate the effects of a normal
distribution. However, these results do provide a
sense of what the B values mean. A lower External
Biofidelity rank indicates a more biofidelic dummy
than does a high number and a dummy with an
External Biofidelity rank of B< 2.0 can be considered
to respond as much like the cadaver corridor as
would another human subject.
External Biofidelity
Applying this methodology to the three side impact
dummies indicates that 1) the WorldSID-α prototype
dummy exhibits better overall External Biofidelity
over the current designs of the European and U.S.
side impact dummies, 2) the overall External
Biofidelity of all of the dummies has room for
improvement, and 3) except for isolated body
components, none of these dummies can be expected
to load a vehicle like a human cadaver in a crash test.
If each body region is considered independently, it is
apparent that each dummy has its strengths and
weaknesses. For example, if the head/neck External
Biofidelity is considered, the SID-HIII ranks better
than the WorldSID-α, which ranks better than the
ES-2 as shown in Table 7. The head/neck rank is
determined from five measurements of Neck Tests 1
and 3. Figures 10 and 11 show the dummy results
and biofidelity response corridors plotted.
Looking at the shoulder External Biofidelity, the ES-
2 ranks better than the other dummies. The shoulder
ranks are determined from the displacement of T-1
relative to the sled in Neck Test 1 (Figure 10) and
pendulum forces of Shoulder Test 1 (Figure 12) and
the NHTSA Shoulder Test (Figure 13). No data was
available for the WorldSID-α in Shoulder Test 1, so
its shoulder External Biofidelity rank was only based
on two measurements. Figure 10 shows that the SID-
HIII had much more horizontal displacement
between T-1 and the sled than the corridor in Neck
130
64
162
9494
230
139
46
131
114
127
98
88
124
63
67 74
72
44
59
53
40
0
50
100
150
200
250
0
10
20
30
40
50
60
70
Lower bound U pper bound ES-2 SID-H3 WSID
Peak Horiz. Displ.
of Head cg
relative to T-1
Peak Vert. Displ.
of Head cg
relative to T-1
Peak Head
Flexio n Angle
Peak Horiz. Displ.
of T-1
relative to Sled
Displacement (mm)
Flexion Angle (deg)
shoulder
response
Figure 10. Neck Test 1 dummy responses with
standard corridors.
185
226
173
198
221
208
75
50
63
53
62
55
150
160
170
180
190
200
210
220
230
40
45
50
55
60
65
70
75
80
Lower bound Upper bound ES-2 SID-H3 WSID
Peak Horizontal Disp lacemen
t
of Head CG Relative to Sled
Peak Head Flexion Angle
Flexion Angle (deg)
Displacement (mm)
Figure 11. Neck Test 3 dummy responses with
standard corridors.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Test 1. Figures 12 and 13 show that the SID-HIII
pendulum force response rises much slower than that
of the corridor and the other dummies. These
differences in response may be attributed to its half-
arm composition of foam.
According to the thorax External Biofidelity ranks,
the WorldSID-α EBthorax is much improved over that
of the ES-2 and SID-HIII. These ranks are based on
the pendulum force of Thorax Test 1 (Figure 14) and
the thorax plate force of the six NHTSA sled tests
(Figures D1(a)-D1(f) in Appendix D). However, the
WorldSID-α dummy was not subjected to the
abdomen offset sled test, so its rank is based on six
responses instead of seven. The thorax pendulum test
procedure calls for the dummy’s arm to be positioned
out of the way so as to impact the thorax. The half-
arm of the SID-HIII cannot be raised, so the test was
run with the jacket and arm in place. The effect of
the soft foam arm can be seen in Figure 14. Figures
D1(a)-D1(f) show the WorldSID-α thorax load plate
force responses to be much closer to the biofidelity
response corridors than the ES-2 or SID-HIII.
The EBabdomen ranks are based on the responses of the
abdomen plate force of the six NHSTA sled tests,
with the exception of the WorldSID-α as it was not
tested in the abdomen offset condition (Figures
D2(a)-D2(f) in Appendix D). To complete the
comparison, the WorldSID-α dummy ought to be
subjected to this condition before final analysis of the
abdomen External Biofidelity is made. Figure 15
indicates the effects of the SID-HIII arm in loading
the abdomen plate. This is the second test that called
for the dummy’s arm to be positioned out of the way
of the thorax, which the SID-HIII is unable to
accommodate. Accordingly, this affects its
biofidelity rankings.
-500
0
500
1000
1500
2000
2500
3000
3500
4000
0 10203040506070
Time (msec)
ES-2 SID-HIII WorldSID Corri dor Cadaver mean
Pendulum Force (N)
Figure 13. NHTSA Shoulder Test dummy responses an
d
standard corridor.
Figure 12. Shoulder Test 1 dummy responses and standard
corridor.
-1000
0
1000
2000
3000
4000
0 5 10 15 20 25 30 35 40
Time (msec)
Pendulum Force (N)
ES-2 SID-HII I 'Standard' corr idor Cadaver mean
0
1000
2000
3000
4000
5000
0 5 10 15 20 25 30 35 40
Time (msec)
ES-2 SID-HIII WorldSID
'Standard' corridor Cadaver mean
Pendulum Force (N)
Figure 14. Thorax Test 1 dummy responses and standar
d
corridor.
Figure 15. External Biofidelity response for abdomen plate
force in LS RAO test.
0
5000
10000
15000
20000
25000
0 1020304050
Time (mSec)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
Low Speed Rigid Abdomen Offset
SID-HIII R1/2 = 5.86
ES-2 R1/2 = 5.23
WorldSID R1/2 = n/a
Abdomen Plate Force (N)
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
The pelvis External Biofidelity ranks for the
dummies were 2.7 and greater, indicating that
improvement is needed in this body region for all
three dummies. These ranks are based on the
responses of the pelvis plate force of the six NHSTA
sled tests, with the exception of the WorldSID-α, as it
was not subjected to the abdomen offset condition
(Figures D3(a)-D3(f) in Appendix D). Figure 16 is a
typical plot of the flat wall sled test responses for the
three dummies showing a delay in the force response
of the pelvis plate, as compared to the cadaver
corridors (see Figure D3 in Appendix D). This may
be an indication of the difference in anthropometry
between the dummies and the cadavers. These
dummies have broad shoulders and narrow hips,
based on anthropometry data of mid-sized males.
The cadaver subjects used in developing the corridors
did not necessarily match the shoulder and hip
breadth upon which the dummies were designed.
The broader the shoulders and the narrower the hips,
the longer the time delay will be between first contact
with the wall and when the pelvis begins to impart
load to the wall.
Internal Biofidelity
Applying this methodology to the three side impact
dummies indicates that their Internal Biofidelity is
adequate. The IB ranks in Table 8 indicate the injury
prediction capability of each body region (Response
Measurement Comparison Values are shown in
Appendix E and plots of the corresponding channels
are shown in Appendices F-H). Table 8 includes the
overall ranks with, and without, the abdomen ranks
because the abdomen of the SID-HIII and the ES-2
cannot be evaluated. All of the overall IB ranks are
less than 2.0. This indicates that all three dummies
appear to be quite biofidelic when using the injury
criteria channels most appropriate for the individual
dummy design. Because the dummies have been
designed to produce a particular response, the
Internal Biofidelity ranks are generally much smaller
than the External Biofidelity ranks.
The IB ranks for the head were 1.1, 1.0 and 0.8,
respectively, for the SID-HIII, ES-2 and WorldSID-
α. The IBhead ranks are determined from the peak
resultant head acceleration and duration during Head
Test 1. Additional testing of both cadaver and
dummy heads in various configurations would allow
for an improved evaluation of head biofidelity.
The IB ranks for the thorax are all quite low at 2.2,
1.7 and 1.4. It should be noted that the SID-HIII
IBthorax rank is based on the three thoracic
acceleration channels used in TTI. The ES-2 and the
WorldSID-α IBthorax ranks are based on two thoracic
deflections and an acceleration. Because acceleration
is more sensitive to small variations in impact
conditions, it is not surprising that an acceleration-
based IB value would be somewhat larger. As noted
previously a value of B=2.0 is approximately as
biofidelic as another cadaver.
All three dummies achieved IBpelvis ranks greater than
2.0, at 2.5, 2.1 and 2.4. This is likely due to the
pelvis response delay discussed previously (see
Appendix H).
DISCUSSION
This system to rank dummy biofidelity is a new
method under development for analyzing dummy
responses from many test exposures. The Bio Rank
System introduces External Biofidelity as an equal
partner to Internal Biofidelity in the analysis of the
ability of a dummy to replicate human dynamics.
Both parts of the Biofidelity Ranking System are
necessary to achieve a complete understanding of
how a dummy responds relative to a human. Without
the External Biofidelity ranks, one might be led to
believe that all three dummies perform with similar
levels of adequate biofidelity. Without the Internal
Biofidelity ranks, the internal responses of the
dummies, and their ability to predict injury potential,
would be ignored. A dummy needs to load a vehicle
as a human does in a crash and to duplicate the
responses necessary for assessing injury potential.
The Bio Rank System also identifies the body regions
of dummies that need improvement to more closely
replicate human dynamics, which will aid in the
development of more biofidelic dummies.
Figure 16. External Biofidelity response for pelvis plate
force in HS RF test.
0
10000
20000
30000
40000
50000
0 5 10 15 20 25 30
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HI II
ES-2
WorldSID
High Speed Rigid Flat Wall
SID-HII I R1/2 = 4.73*
ES-2 R1/2 = 3.35
World SID R1/2 = 3.7 7
* data cli pped
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
The Bio Rank System is valuable as a dummy
biofidelity assessment tool and as a dummy
comparison tool for the purpose of developing
dummies for injury mitigation research. The body
region ranks that are generated as part of the External
Biofidelity and the Internal Biofidelity ranking
process are especially useful to assess the strong and
weak points of a dummy design. As all body regions
are weighted equally, the interactive effect that each
region has on the whole dummy response is
recognized and allows the dummy designer and user
to study the body region responses and make an
informed judgment of overall dummy biofidelity.
The Bio Rank System proposes a method of ranking
dummy response; however, there is a great deal of
work to be done in the future. A primary goal would
be to fill in the gaps where there is missing data, to
allow the comparison of all dummies on an equal
basis. The Test Condition Weights may need some
adjustment in the way the Test Relevance scores are
assigned. Currently, points are awarded to test
conditions for achieving results at or “below the
maximum” of crash tests. The below the maximum
criterion was selected because regulated dummies are
used in all environments up to the maximum severity
of FMVSS 214, FMVSS 201 and SNCAP crash tests.
Perhaps the points ought to be awarded based on the
number of standard deviations away from the mean
response of the crash tests so that the highest Test
Relevance scores (and Test Condition Weights) are
awarded to those biofidelity tests that most closely
represent the crash test environment. The Test
Condition Weights may also need adjustment in the
way that the Subject scores are assigned. The post-
test human subject condition was not taken into
consideration, and the Subject score for test
conditions whose response corridors were created
from severely damaged subjects ought to be
penalized.
Additional work might include correlating the injury
criteria to the Internal Biofidelity rank calculation.
For example, if a combination of acceleration and
deflection, rather than acceleration only, were a better
indicator of injury in the thorax, a correlation factor
in the calculation would show the ES-2 and
WorldSID-α to be better predictors of injury over the
SID-HIII since they have the capacity to measure
thoracic deflection. Since the Internal Biofidelity
rank is dependent on the injury criterion being
considered, the rank could also be weighted to reflect
the potential for serious injury to different body
regions, e.g., thoracic injury over abdominal injury in
a side impact. The ranks could also be separated into
ranks for shape, phase and magnitude of the signal to
identify reasons to support the current body region
ranks. For example, the pelvis acceleration would
have a poor phase rank to indicate the time delay
discussed previously. Finally, it is recognized that
other valuable biomechanical response corridors exist
for side impact that are not included in the Bio Rank
System at this time. Since the results of any
biofidelity ranking system are dependent upon the
tests selected for evaluation and the weight factors
assigned to the tests, it is critical that the set of tests
included in the system is appropriate. As the Bio
Rank System is developed further, the set of tests
included in the system will be evaluated for
appropriateness and modified as needed.
The Bio Rank System only ranks those aspects of
dummy response that can be compared to the selected
cadaver test results. If a dummy has a non-human
characteristic, such as an artificial load path, the Bio
Rank System would not identify it as a lack of
biofidelity. Other significant dummy problems such
as non-repeatability or poor durability are not
included in the Bio Rank. A particular dummy may
not have been designed to have biofidelity in a
particular region or may not record any meaningful
data from a region. In these cases, it is uncertain
what a dummy rank for that body region means. The
SID-HIII abdomen is an example of the former case
and the ES-2 internal abdomen force is an example of
the latter case. Finally, a dummy may produce a
reliable response, such as peak acceleration, that is a
valuable predictor of injury, while it is not
particularly biofidelic based on the selected tests. In
summary, the choice of a dummy for crash research
and regulation purposes depends on many factors in
addition to External Biofidelity and Internal
Biofidelity, all of which need to be considered.
The International Standards Organization (ISO)
Committee developed a biofidelity rating system for
side impact dummies in the ISO Technical Report
9790 (ISO TR 9790) (ISO 1999). The ISO TR 9790
methodology formed the basis of the system
presented here; however, there are significant
differences between the two approaches. In this
approach the difference between a mean cadaver time
history and a dummy time history is quantified using
a cumulative variance calculation; the dummy
biofidelity is divided into external and internal
response; the weighting schemes for response
measurement, test condition and body region are
defined differently; and the tests selected for
comparison are quite different. Although this
Biofidelity Ranking System evolved from the ISO
methodology, a direct comparison between the two
approaches would be interesting but difficult.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
CONCLUSIONS
A new system for ranking dummy biofidelity has
been developed and is being offered for consideration
to the safety community. While there is future work
required (such as filling in data needs and adjusting
the weights assigned to the test conditions), the
proposed methodology appears to provide useful
insights in the evaluation of side impact dummies.
ACKNOWLEDGMENTS
The authors thank the WorldSID Organization for
allowing NHTSA to use their only existing prototype
dummy in biofidelity testing. The Transportation
Research Center Inc. (TRC), Medical College of
Wisconsin, and the Ohio State University were
instrumental in performing tests. Patrick Brown was
invaluable in data processing and analysis and his
contribution to this work is very much appreciated.
REFERENCES
Bolte, J.H., Hines, M.H., McFadden, J.D., and Saul,
R.A. (2000) Shoulder Response Characteristics and
Injury Due to Lateral Glenohumeral Joint Impacts.
Proc. 44th Stapp Car Crash Conference, pp. 261-
280. Society of Automotive Engineers,
Warrendale, PA.
Hodgson, V.R. and Thomas, L.M. (1975) Head
Impact Response. Vehicle Research Institute
Report – VRI 7.2. Society of Automotive
Engineers.
International Standards Organization. (1999)
Technical Report 9790: Road Vehicles –
Anthropomorphic Side Impact Dummy – Lateral
Impact Response Requirements to Assess the
Biofidelity of the Dummy.
Maltese, M.R., Eppinger, R.H., Rhule, H.H.,
Donnelly, B., Pintar, F.A. and Yoganandan, N.
(2002) Response Corridors of Human Surrogates
In Lateral Impacts. Stapp Car Crash Journal 46.
Morgan, R.M., Marcus, J.H., and Eppinger, R.H.
(1986) Side Impact – The Biofidelity of NHTSA’s
Proposed ATD and Efficacy of TTI. Proc. 30th
Stapp Car Crash Conference, pp. 27-40. Society
of Automotive Engineers, Warrendale, PA.
SAE. (1995) Instrumentation for Impact Test. SAE
Handbook Volume 4 – On-Highway Vehicles &
Off-Highway Machinery. SAE J211/1, Rev. Mar
1995. Society of Automotive Engineers,
Warrendale, PA
SAE. (1995) Instrumentation for Impact Test. SAE
Handbook Volume 4 – On-Highway Vehicles &
Off-Highway Machinery. SAE J211/1, Rev. Mar
1995. Society of Automotive Engineers,
Warrendale, PA.
U. S. Code of Federal Regulations. 49 CFR 571.201.
Standard No. 201, Occupant Protection in Interior
Impact. Office of the Federal Register, National
Archives and Records Administration. October 1,
2001.
U. S. Code of Federal Regulations. 49 CFR 571.214.
Standard No. 214, Side Impact Protection. Office
of the Federal Register, National Archives and
Records Administration. October 1, 2001.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
APPENDIX A: NHTSA SHOULDER TEST
DATA AND CORRIDOR
Padded shoulder pendulum tests were conducted on
six cadavers. The setup was identical for each test
and all data were normalized to a 50th percentile male
sized subject (Bolte et al., 2000). Biomechanical
corridors were created by calculating the mean of the
six normalized plots and adding and subtracting one
standard deviation from the mean. None of the
subjects used in the corridor development were
injured during the pendulum impact.
The event “time zero” was indicated by shoulder
contact with the pendulum, where an event channel
signified contact. Non-normalized data from the six
cadaver tests are shown in Figure A1, and the six
normalized cadaver curves and calculated corridors
are displayed in Figure A2. Results from three of the
subjects were published previously (Bolte et al.,
2000), while three tests were performed subsequent
to the publication of the aforementioned paper.
Figure A1. NHTSA Shoulder Test Non-normalized
cadaver data.
Figure A2. NHTSA Shoulder Test normalized cadaver
data and calculated corridor.
NHTSA Should er Impacts - Non-normalize d Data
-500
0
500
1000
1500
2000
2500
3000
3500
0 20 40 60 80 100 120
Time (msec)
Force (Newtons)
Subject 1
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
NHTSA Shoulder Impact Normalized Data and Corridors
-500
0
500
1000
1500
2000
2500
3000
3500
4000
0 20 40 60 80 100 120
Time (msec)
Force (Newtons)
Subject 1
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
Corridor
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
APPENDIX B: HEAD TEST 1
Table B1. Head Test 1 cadaver responses and standard corridor
Cadaver
ID
Impact
Velocity
(m/s)
Equivalent Free
Fall Drop
Height (mm)
Peak Resultant Acceleration at
non-impacted side of head (g)
Slope = peak
acceleration /
impact velocity
Peak Resultant Acceleration at
non-impacted side of head at 2
m/s (g)
2864 1.92 188 107 55.7 111
2953 1.74 154 108 62.1 124
3030 1.92 188 135 70.3 141
3042 1.92 188 118 61.5 123
3083 1.92 188 96 50.0 100
3116 1.65 139 121 73.3 147
3184 1.74 154 101 58.0 116
mean 123
std. dev. 16
STANDARD CORRIDOR = 123 +/- 16 = 107 – 139
0
20
40
60
80
100
120
140
160
Standard corridor Cadaver mean ES-2 SID-HIII WSID
ES-2 SID-HIII WSID
Peak Head Resultant Acceleration (g)
N
ote: Test procedure requires peak head resultant acceleration at the non-impacted side of the head.
Dummy measurements were made at the cg of the head.
Figure B1. Head Test 1 dummy responses with standard corridor.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
APPENDIX C: VALIDATION
Figure C1. Thorax plate force validation responses
Figure C1. Thorax plate force validation plots.
-0.02 00. 02 0.04 0. 06 0.08
-2000
0
2000
4000
6000
8000
10000
(a.) Thorax Force-Cadaver #2
Time (sec onds)
Force (Newt ons)
R1/2 = 1.0551
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
(b.) Thorax Force-Cadaver #3
Time (sec onds)
Force (Newt ons)
R1/2 = 0.82869
-0.02 00. 02 0.04 0. 06 0.08
-2000
0
2000
4000
6000
8000
10000
(c.) Thorax Force-Cadaver #4
Time (sec onds)
Force (Newtons)
R1/2 = 2.1573
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
(d.) Thorax Force-Cadaver #5
Time (sec onds)
Force (Newtons)
R1/2 = 0.96409
-0.02 00. 02 0.04 0. 06 0.08
-2000
0
2000
4000
6000
8000
10000
(e.) Thorax Forc e-Cadaver #6
Time (sec onds)
Force (Newtons )
R1/2 = 0.80822
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
(f.) Thorax Force-Cadaver #7
Time (sec onds)
Force (Newtons )
R1/2 = 1.0909
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
-0.02 00.02 0.04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(a.) Abdomen Force-Cadaver #2
Time (sec onds)
Force (Newt ons)
R1/2 = 1.4772
-0.02 00.02 0. 04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(b.) Abdom en Force-Cadaver #3
Time (sec onds)
Force (Newt ons)
R1/2 = 0.77679
-0.02 00.02 0.04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(c.) Abdomen Force-Cadaver #4
Time (sec onds)
Force (Newtons)
R1/2 = 1.5297
-0.02 00.02 0. 04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(d.) Abdom en Force-Cadaver #5
Time (sec onds)
Force (Newtons)
R1/2 = 0.73403
-0.02 00.02 0.04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(e.) Abdomen Force-Cadaver #6
Time (sec onds)
Force (Newtons )
R1/2 = 2.1615
-0.02 00.02 0. 04 0.06 0. 08
-1000
0
1000
2000
3000
4000
5000
(f.) Abdomen Forc e-Cadaver #7
Time (sec onds)
Force (Newtons )
R1/2 = 1.1662
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Figure C2. Abdomen plate force validation plots.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
12000
(a.) Pelvis Forc e-Cadaver #2
Time (sec onds)
Force (Newtons)
R1/2 = 0.97685
-0.02 00.02 0. 04 0.06 0. 08
-2000
0
2000
4000
6000
8000
10000
12000
(b.) Pelvis Force-Cadaver #3
Time (seconds)
Force (Newtons)
R1/2 = 1.5339
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
12000
(c.) Pelvis Force-Cadaver #4
Time (sec onds)
Force (Newt ons)
R1/2 = 1.5503
-0.02 00.02 0. 04 0.06 0. 08
-2000
0
2000
4000
6000
8000
10000
12000
(d.) Pelvis Force-Cadaver #5
Time (seconds)
Force (Newt ons)
R1/2 = 1.2411
-0.02 00.02 0.04 0.06 0.08
-2000
0
2000
4000
6000
8000
10000
12000
(e.) Pelvis Forc e-Cadaver #6
Time (sec onds)
Force (Newtons)
R1/2 = 2.0903
-0.02 00.02 0. 04 0.06 0. 08
-2000
0
2000
4000
6000
8000
10000
12000
(f.) P elvis Forc e-Cadaver #7
Time (seconds)
Force (Newtons)
R1/2 = 2.3526
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Cadaver
Mean
Upper
Lower
Figure C3. Pelvis plate force validation plots.
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
APPENDIX D: EXTERNAL BIOFIDELITY
Table D1. SID-HIII External Biofidelity Response Measurement Comparison Values
SID-HIII External Biofidelity Shoulder
Test 1
NHTSA
Shoulder
Test
Thorax
Test 1
Neck
Test 1
Neck
Test 3 HSFR HSFP LSFR LSFP LSRAO LSRPO
V(s) – Test Setup Weights (1-10) 8 9 9 9 6 4 9 6 8 5 6
Pendulum Force 2.644 3.370 2.450
Thorax Plate Force 5.846 8.124 7.177 6.500 5.225 8.081
Abdominal Plate Force 2.400 3.374 0.927 1.660 5.864 4.390
Pelvis Plate Force 4.728 7.393 2.355 2.464 2.310 2.307
Peak Horizontal. Displacement of
T-1 relative to sled 9.000
Peak Horizontal. Displacement of
Head cg relative to T-1 0.436
Peak Vertical Displacement of
Head cg relative to T-1 2.332
Peak Horizontal. Displacement of
Head cg relative to sled 0.755
Peak Flexion Angle 0.436 0.849
Head/Neck Biofidelity Rank 1.0
Shoulder Biofidelity Rank 5.1
Thorax Biofidelity Rank 6.2
Abdomen Biofidelity Rank 3.0
Pelvis Biofidelity Rank 3.8
Overall Biofidelity Rank 3.8
Table D2. ES-2 External Biofidelity Response Measurement Comparison Values
ES-2 External Biofidelity Shoulder
Test 1
NHTSA
Shoulder
Test
Thorax
Test 1
Neck
Test 1
Neck
Test 3 HSFR HSFP LSFR LSFP LSRAO LSRPO
V
(
s
)
– Test Setu
p
Wei
g
hts
(
1-10
)
8 9 9 9 6 4 9 6 8 5 6
Pendulum Force 1.870 1.055 1.136
Thorax Plate Force 4.714 3.643 3.556 4.393 2.661 3.207
Abdominal Plate Force 1.506 3.027 0.575 1.506 5.316 3.491
Pelvis Plate Force 3.353 4.800 1.701 2.572 1.949 0.905
Peak Horizontal. Displacement of
T-1 relative to sled
1.470
Peak Horizontal. Displacement of
Head cg relative to T-1
3.250
Peak Vertical Displacement of
Head cg relative to T-1
10.067
Peak Horizontal. Displacement of
Head cg relative to sled
1.584
Peak Flexion Angle 1.000 2.846
Head/Neck Biofidelit
y
Ran
k
3.7
Shoulder Biofidelity Rank 1.4
Thorax Biofidelity Rank 3.3
Abdomen Biofidelity Rank 2.5
Pelvis Biofidelity Rank 2.7
Overall Biofidelit
y
Ran
k
2.7
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Table D3. WorldSID External Biofidelity Response Measurement Comparison Values
WorldSID External Biofidelity Shoulder
Test 1
NHTSA
Shoulder
Test
Thorax
Test 1
Neck
Test 1
Neck
Test 3 HSFR HSFP LSFR LSFP LSRAO LSRPO
V
(
s
)
– Test Setu
p
Wei
g
hts
(
1-10
)
8 9996496 8 56
Pendulum Force DL 1.931 1.292
Thorax Plate Force
2.675 2.825 0.775 2.498 * 2.831
Abdominal Plate Force
1.229 3.190 0.635 1.984 * 3.335
Pelvis Plate Force
3.772 5.966 3.217 3.154 * 2.033
Peak Horizontal. Displacement of
T-1 relative to sled
2.176
Peak Horizontal. Displacement of
Head cg relative to T-1
3.312
Peak Vertical Displacement of
Head cg relative to T-1
3.100
Peak Horizontal. Displacement of
Head cg relative to sled
0.230
Peak Flexion Angle 1.533 2.231
Head/Neck Biofidelit
y
Ran
k
2.1
Shoulder Biofidelity Rank 2.1
Thorax Biofidelity Rank 2.1
Abdomen Biofidelity Rank 2.2
Pelvis Biofidelity Rank 3.8
Overall Biofidelit
y
Ran
k
2.5
DL data loss
* WorldSID dummy not tested in LSRAO condition
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Figure D1. External Biofidelity thorax plate force responses.
-2000
0
2000
4000
6000
8000
10000
0 102030405060
Time (mSec)
Thorax Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
Low Speed Rigid Abdomen Offset
SID-HIII R1/2 = 5.23
ES-2 R1/2 = 2.66
WorldSID R1/2 = n/a
0
5000
10000
15000
20000
25000
30000
0 1020304050
Time (mSec)
Thorax Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
High Speed Rigid Flat Wal l
SID-HIII R1/2 = 5.85
ES-2 R1/2 = 4.71
WorldSID R1/2 = 2.67
-2000
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0 102030405060
Time (mSec)
Thorax Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HI II
ES2
Wor l dS I D
High Speed Padded Flat Wal l
SID-HIII R1/2 = 8.12
ES-2 R1/ 2 = 3.64
WorldSID R1/2 = 2.83
0
2000
4000
6000
8000
10000
12000
14000
16000
0 102030405060
Time (mSec)
Thorax Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid Flat Wall
SID-HIII R1/2 = 7.18
ES-2 R1/2 = 3.56
WorldSID R1/2 = 0.78
-1000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
0 102030405060
Time ( mSec)
Thorax Plate Force
(
N
)
CORRIDOR
CADAVER MEAN
SID- HI II
ES2
Wor l d SI D
Low Speed Padded Fl at Wall
SID-HIII R1/2 = 6.50
ES-2 R1/2 = 4.39
WorldSID R1/2 = 2.50
(a) (b)
(c) (d)
(e) (f)
0
2000
4000
6000
8000
10000
12000
0 1020304050607080
Time (mSec)
Thorax Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid Pelvis Off set
SID-HIII R1/2 = 8.08
ES-2 R1/2 = 3.21
WorldSID R1/2 = 2.83
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Figure D2. External Biofidelity abdomen plate force responses.
-500
0
500
1000
1500
2000
2500
3000
0 5 10 15 20 25 30 35
Time (mSec)
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid Pelvis Offset
SID-HIII R1/2 = 4.39
ES-2 R1/2 = 3.49
World SID R1/2 = 3.34
-2000
0
2000
4000
6000
8000
10000
12000
0 1020304050
Time (mSec )
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
High Speed Rigid Flat W all
SID-HIII R1/2 = 2.40
ES-2 R1/2 = 1.51
WorldSID R1/2 = 1.23
-1000
0
1000
2000
3000
4000
5000
0 102030405060
Time (mSec)
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
High Speed Padded Flat Wall
SID-HIII R1/2= 3.37
ES-2 R1/2 = 3.03
WorldSID R1/2 = 3.19
-1000
0
1000
2000
3000
4000
5000
0 1020304050
Time (mSec)
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid Flat Wall
SID-HIII R1/2 = 0.93
ES-2 R1/2 = 0.57
WorldSID R1/2 = 0.63
-500
0
500
1000
1500
2000
2500
3000
0 10 2030405060
Time (mSec)
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
WorldSID
ES2
Low Speed Padded Flat Wa ll
SID-HIII R1/2 = 1.66
ES-2 R1/2 = 1.51
WorldSID R1/2 = 1.98
0
5000
10000
15000
20000
25000
0 1020304050
Time (mSec)
Abdomen Plate Force (N)
CORRIDOR
CADAVER MEA N
SID-HIII
ES2
Low Speed Rigid Abdomen Offset
SID-HI II R1/2 = 5.86
ES-2 R1/2 = 5.23
WorldS ID R1/2 = n/a
(
a
)
(
b
)
(c) (d)
(e) (f)
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
Figure D3. External Biofidelity pelvis plate force responses.
0
10000
20000
30000
40000
50000
0 5 10 15 20 25 30
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES-2
WorldSID
High Speed Rigid Flat Wa ll
SID-HIII R1/2 = 4.73*
ES-2 R1/2 = 3.35
WorldSID R1/2 = 3.77
* data clipped
-2000
0
2000
4000
6000
8000
10000
12000
14000
16000
0 102030405060
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
LOWER CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
SID-HIII R1/2 = 7.39
ES-2 R1/2 = 4.80
WorldSID R1/2 = 5.97
High Speed Padded Flat Wall
-2000
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0 1020304050
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid F lat Wall
SID-HIII R1/2 = 2.36
ES-2 R1/2 = 1.70
WorldSID R1/2 = 3.22
-2000
0
2000
4000
6000
8000
10000
12000
0 102030405060
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Padded Flat Wall
SID-HIII R1/ 2 = 2.46
ES-2 R1/2 = 2.57
WorldSID R1/2 = 3.15
-1000
0
1000
2000
3000
4000
5000
6000
7000
8000
0 1020304050
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
Low Speed Rigid Abdomen Offset
SID-HIII R1/2 = 2.31
ES-2 R1/2 = 1.95
WorldSID R1/2 = n/a
-5000
0
5000
10000
15000
20000
25000
0 102030405060
Time (mSec)
Pelvis Plate Force (N)
CORRIDOR
CADAVER MEAN
SID-HIII
ES2
WorldSID
Low Speed Rigid Pelvis Offset
SID-HIII R1/2 = 2.31*
ES-2 R1/2 = 0.90
WorldSID R1/2 = 2.03
* data clipped
(a) (b)
(c) (d)
(e) (f)
Rhule et al. / Stapp Car Crash Journal 46 (November 2002)
APPENDIX E: INTERNAL BIOFIDELITY
Table E1. Response Measurement Comparison Values for Internal Biofidelity Responses
Test Condition
Dummy Response Measurement
Head
Test 1 HSFR HSFP LSFR LSFP LSRAO LSRPO IB
RANK
Test Condition Weights (1-10) 9 4 9 6 8 5 <