ISO/TR 12350:2013

TECHNICAL ISO/TR
REPORT 12350
Second edition
2013-10-01
Road vehicles — Injury risk curves for
the evaluation of occupant protection
in side impact tests
Véhicules routiers — Courbes de risques de blessures pour l’évaluation
de la protection des occupants en choc latéral
Reference number
ISO/TR 12350:2013(E)
©
ISO 2013

---------------------- Page: 1 ----------------------
ISO/TR 12350:2013(E)

COPYRIGHT PROTECTED DOCUMENT
© ISO 2013
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized otherwise in any form
or by any means, electronic or mechanical, including photocopying, or posting on the internet or an intranet, without prior
written permission. Permission can be requested from either ISO at the address below or ISO’s member body in the country of
the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail copyright@iso.org
Web www.iso.org
Published in Switzerland
ii © ISO 2013 – All rights reserved

---------------------- Page: 2 ----------------------
ISO/TR 12350:2013(E)

Contents Page
Foreword .iv
1 Scope . 1
2 Methodology . 1
2.1 Selection of PMHS sample to be used for the construction of the injury risk curves . 1
2.2 Dummy data . 3
2.3 Age adjustment . 3
2.4 Statistical analysis . 4
3 Injury risk curves for the WorldSID 50th . 7
4 Related electronic documents .20
Annex A (informative) PMHS head test data .21
Annex B (informative) PMHS shoulder test data (shoulder impactor tests) .29
Annex C (informative) PMHS thorax test data (thorax impactor tests) .36
Annex D (informative) PMHS abdomen test data (abdomen impactor tests) .42
Annex E (informative) PMHS pelvis test data (pelvis impactor tests).44
Annex F (informative) PMHS sled test data .54
Annex G (informative) Assessment of the quality of the sled test results.66
Annex H (informative) WorldSID results .69
Annex I (informative) Data scaling .82
Annex J (informative) Steps to build injury risk curves dedicated to the WorldSID 50th .93
Bibliography .107
© ISO 2013 – All rights reserved iii

---------------------- Page: 3 ----------------------
ISO/TR 12350:2013(E)

Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out
through ISO technical committees. Each member body interested in a subject for which a technical
committee has been established has the right to be represented on that committee. International
organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of
electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www.iso.org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation on the meaning of ISO specific terms and expressions related to conformity
assessment, as well as information about ISO’s adherence to the WTO principles in the Technical Barriers
to Trade (TBT) see the following URL: Foreword - Supplementary information
The committee responsible for this document is ISO/TC 22, Road vehicles, Subcommittee SC 12, Passive
safety crash protection systems.
This second edition cancels and replaces the first edition (ISO/TR 12350:2004), which has been
technically revised.
iv © ISO 2013 – All rights reserved

---------------------- Page: 4 ----------------------
TECHNICAL REPORT ISO/TR 12350:2013(E)
Road vehicles — Injury risk curves for the evaluation of
occupant protection in side impact tests
1 Scope
This Technical Report provides injury risk curves to assess occupant protection in side impact tests. The
curves are given for the WorldSID 50th, a mid-size adult male side impact dummy. Injury risk curves for
other side impact dummies could be added as soon as the necessary material is available and processed
as described in this Technical Report. These dummies are used during tests carried out according to
ISO 10997 or which are under investigation by regulatory bodies and consumer testing organizations.
2 Methodology
2.1 Selection of PMHS sample to be used for the construction of the injury risk curves
An in-depth review of the postmortem human subjects (PMHS) tests available in the literature and in
the NHTSA database (http://www-nrd.nhtsa.dot.gov/database/aspx/biodb/querytesttable.aspx) was
performed. The listed tests were analysed in order to determine if they could be accurately repeated
with dummies and included in the construction of injury risk curves.
This clause summarizes the series of tests that were conducted by body region and type of loading.
Reasons for including or excluding each particular test series are detailed. The PMHS characteristics
are provided in the form of related electronic documents available through the ISO website and detailed
in Clause 4. The detailed descriptions of the PMHS configurations allowing the reproduction of the test
with a dummy are presented in Annex A to Annex F, as well as the reasons for inclusion or exclusion.
[18]
The rigid and padded head impactor tests conducted by Calspan were included and are detailed in
[51]
Annex A. The head impactor tests of the Highway Safety Research Institute (HSRI) were excluded
because the impact speeds were not known. The head impactor tests of the University of Michigan
Transportation Research Institute (UMTRI) (NHTSA database) were excluded because the impactor
characteristics were not known.
[22]
The whole body drop tests with head impact conducted by Wayne State University (WSU) , those
conducted by the Association Peugeot-Renault (APR) without helmet, and the head drop tests conducted
[56]
by Medical College of Wisconsin (MCW) were included and are detailed in Annex A. The whole body
drop tests with head impact conducted by APR with helmet were excluded because the helmet properties
were unknown.
[2] [14] [15] [17] [26] [30]
The shoulder impactor tests performed by APR , INRETS , and WSU were included.
The shoulder impactor tests conducted by Ohio State University (OSU) on a rigid bench were also
[3] [4]
included . These configurations are detailed in Annex B. The oblique shoulder impactor tests
performed by OSU on a 1996 Ford Taurus seat were excluded because the characteristics of the seat
were unknown.
[43] [44] [45]
All, but one, of the thorax impactor tests conducted by HSRI  were included. The single-impact
[54] [55] [34] [35] [49]
WSU thorax impactor test was also included. The UMTRI and OSU thorax impactor
tests were included when the level of load was deemed to be below the threshold of rib fracture (700
N), such that the fractures could be attributed to the final high-speed impact. These test configurations
are detailed in Annex C. The 76T038 HSRI test was excluded because the data were questionable. The
HSRI tests 77T079 and 77T080 were excluded because it does not seem realistic to have 18 rib fractures
[16]
for 2 165 N of impact force. All the WSU and INRETS multi-impact tests, as well as some UMTRI
tests (83E085, 83E086, 83E106, 83E107, 83E108) and OSU tests (0505OTH25L01, 0505LTH25R01,
0506OTH25R01, 0506LTH25L01, 0601LTH25L01, 0601OTH25R01), were excluded because it was not
possible to determine which impact caused each injury.
© ISO 2013 – All rights reserved 1

---------------------- Page: 5 ----------------------
ISO/TR 12350:2013(E)

[54] [55]
Only one of the abdomen impactor tests performed by WSU (WSU063-34) was included because
all the other subjects were impacted more than once in the abdomen and/or thorax. All the OSU abdomen
impactor tests but two (93VRTAB08, 93VRTAB09) were included. These two tests were excluded because
the abdomen deflection exceeded the target level of 16 % of the chest breadth. The test configurations
are detailed in Annex D.
[52]
The Laboratory of Accidentology and Biomechanics (LAB) abdomen impactor tests were excluded
because a measurement system was positioned at the level of the liver and could have influenced the
abdominal injuries.
Most of the reviewed pelvis impactor tests were multi-impact tests. The pelvis impactor tests performed
[54] [55] [33] [36] [11] [12] [5] [6]
by WSU , UMTRI , ONSER , and INRETS were included when an increase in
impactor speed was accompanied by an increase in energy for a given PMHS, as this was assumed to
be an indication of no injury. The configurations are detailed in Annex E. Two UMTRI tests (83E087,
83E109) were conducted with an APR pad that is no longer available. The ONSER multi-impact tests C3,
C4, D3, E2, F2, F3, H4, H5, I6, J2, J3, N7, S3, S4, X1, X2, Y2, Z1, and Z2 were excluded because there was a
possible weakening of the pelvis bone.
[2] [53]
APR conducted lateral drop tests with PMHS . A review of the films failed to confirm the position
of the subjects’ lower extremities and whether or not an impact surface was provided to catch the lower
extremities. The test films revealed that some tests were conducted with the subjects’ head, some
were conducted without the head, and for the others, the film coverage did not reveal if the head was
attached or removed. Some subjects were observed to rotate during the free fall. For these reasons, and
because the APR padding cannot be reproduced, all of the whole body drop tests were excluded from the
construction of injury risk curves for the thorax, abdomen, and pelvis.
[25] [29] [39] [40] [28] [27] [7] [8] [9] [10] [23]
Some sled tests performed by Heidelberg , MCW/OSU  , and WSU
were included and are detailed in Annex F. Several checks were done to select the PMHS to be included
in the construction of the injury risk curves. The checks are detailed in Annex G.
— The position of the PMHS at the time of impact was first checked. The Heidelberg tests (H82014,
H82018, H82019, H82015) and MCW/OSU tests (SC126, SC105, SC131) were then excluded.
— The consistency between the thorax-pelvis transmissibility and the contact times of the thorax
and pelvis plates were also checked. The Heidelberg tests (H82014, H82018, H82019, H82015) and
MCW/OSU (SC126, SC105) were then excluded.
— The total momentum was checked. Tests for which the total momentum differed from other tests
with the same impact wall configuration were excluded (MCW/OSU SC131).
— The absence of shoulder interaction with the wall was checked in the MCW/OSU configuration. Sled
tests with PMHS seating height under 826 mm and shoulder interaction with the wall observed
on the film were excluded from the shoulder, thorax, and abdomen injury risk curves (MCW/OSU
SC137, SC138, SC119, 94LSI32P04, LSI32R08, SC30A102).
— PMHS characteristics were checked. The PMHS having sternotomy wires were excluded because the
PMHS response and injuries were questionable (MCW/OSU SC122, SC132, LSI32P11, SC103, SC112,
SC30A103, SC20A101).
— PMHS injuries were checked. PMHS from the MCW/OSU SC114 test with a right hemithorax, which
could have resulted from secondary impact, was excluded.
— Some of the checks required the analysis of the wall plates loads. Some tests were excluded because the
impact wall was not instrumented with load cells or because the data were questionable and then the
[31]
checks could not be done. This was the case for the HSRI sled tests , the first test series conducted by
WSU (NHTSA database), some Heidelberg tests (H82009, H80011, H80013, H80014, H80017, H80024,
H81002, H81004, H81006, H81016, H81022, H81025, H81027, H82002, H82020, H80018, H80020,
H80021, H80023, H81011, H81012, H81015, H81021, H83008, H83016, H83021, H83030, H83031), as
well as some MCW/OSU sled tests (98LSI32R17, SC106, SC127, 96LSI32R07, SC123).
2 © ISO 2013 – All rights reserved

---------------------- Page: 6 ----------------------
ISO/TR 12350:2013(E)

— Finally, sled tests for which the impact wall padding or the airbag was no longer available were
excluded (Heidelberg tests H82008, H82021, H82022, H83008, H83016, H83021, H83030, H83031,
H83011, H83020, H84008, H83010, H83012, HSRI tests 76T029, 76T034, 76T039, 76T042, MCW/OSU
nd
tests SAC 101, SAC 103, SAC 104, SAC 105, WSU 2 test series SIC-09, SIC-10, SIC-11, SIC-12, SIC-13,
SIC-14, SIC-15, SIC-16, SIC-17).
[1]
— The severity of PMHS injuries were coded according to the Abbreviated Injury Scale 2005 . Table 1
summarizes the body regions and injury severity levels for which PMHS data are available to
construct injury risk curves. There were no AIS ≥ 3 shoulder injuries from the PMHS tests. Therefore,
injury risk curves for the shoulder can only be constructed for the AIS ≥ 2 level of injury. For the
thorax, abdomen, and pelvis, injury risk curves were constructed at the AIS ≥ 3 injury level and
either the AIS ≥ 2 or AIS ≥ 4 level if the PMHS injury/no injury results were better balanced at these
AIS levels. Note that all rib fractures are coded in thoracic skeletal AIS, including those that resulted
from abdominal impacts.
Table 1 — Body regions and AIS levels for which injury risk curves are constructed
Body region AIS levels used in the injury risk curve construction
Head AIS ≥ 3
Shoulder AIS ≥ 2
Thorax (skeletal) AIS ≥ 3 and AIS ≥ 4
Thorax (soft tissue) AIS ≥ 2 and AIS ≥ 3
Abdomen AIS ≥ 2 and AIS ≥ 3
Pelvis AIS ≥ 2 and AIS ≥ 3
2.2 Dummy data
Once the PMHS sample to build the injury risk curves is selected, the dummy results reproducing these
PMHS test configurations are collected.
The injury risk curves are proposed in this Technical Report for a 50th percentile male dummy. Only
WorldSID 50th percentile results are presented in the current version of this Technical Report. It is
intended to add injury risk curves for the WorldSID 5th percentile adult female dummy to a future
edition of the ISO/TR 12350. There are no plans to add injury risk curves for the ES-2 or ES-2re, and it is
not appropriate to use the WorldSID injury risk curves with measures from either ES-2 or ES-2re.
The dummy test results reproducing the PMHS test configurations selected for the injury risk curve
construction are presented in Annex H. The build level of the production version used was not provided
with the results. It is to be noted that there was no head result available. Moreover, the shoulder deflection
was only available for the impactor test and not for the sled test configurations.
[50]
The test results presented in Annex H are filtered data (according to Reference and according to
filters indicated in Annex H) that have not been scaled and should not be used directly to construct
dummy-specific injury risk curves.
The PMHS used in the biomechanical tests described in Annex A to Annex F were generally not mid-size
adult males. Ideally, the test condition for the dummy tests should be scaled such that the test poses
an equally severe impact as the individual PMHS test. However, many of the dummy tests used in this
Technical Report were conducted at the same velocity and the same impactor mass as the PMHS tests.
It is therefore necessary to scale the results of the dummy tests before they are paired with the PMHS
injuries. The dummy data from impactor tests, drop tests, and sled tests were scaled using the formulae
included in Annex I. The scaled dummy data are included in Annex A to Annex F.
2.3 Age adjustment
The injury risk curves are provided with age adjustment. It is out of the scope of this Technical Report to
recommend an age to be used. The injury risk curves can be built for any age using the formulae included
© ISO 2013 – All rights reserved 3

---------------------- Page: 7 ----------------------
ISO/TR 12350:2013(E)

in Table 5. However, the quality index cannot be computed from these formulae. As it was not possible
to include the quality index for all ages, only two ages were considered. As indicated in Petitjean et al.
(2009), the injury risk curves were constructed for a dummy representing a 45-year-old male, as this
age has been used previously to represent the average age of an adult male in the field data. The injury
risk curves were also constructed for median age of the PMHS included in the samples available for the
construction of the WorldSID 50th injury risk curves (67 years old). This latest age was used because it
provides the values with the higher confidence because the PMHS data are mostly around that age.
2.4 Statistical analysis
Guidelines for the construction of the injury risk curves were agreed on within ISO/TC 22/SC 12
(Resolution 2, N851).
The guidelines include several steps.
2.4.1 Step 1: Collect the relevant data
The first step is to collect the relevant data, including injuries and injury criterion.
According to the methodology developed in this Technical Report, relevant data corresponded to the
paired PMHS injuries and scaled dummy measurements from tests performed in similar configurations.
2.4.2 Step 2: Assign the censoring status (left, right, interval censored, exact)
Once the biomechanical data were available, the censoring status was assigned (left, right, interval
censored, exact). After this step, the dataset included one column with the injury criteria values
associated with the censoring status indicated in a second column.
[37]
2.4.3 Step 3: Build the injury risk curve with the Consistent Threshold Estimate (CTE) and
check for dual injury mechanism
The step function was visually investigated in order to detect potential change in slope corresponding
to different injury mechanisms.
2.4.4 Step 4:
— If there was an evidence of dual injury mechanism, the sample was separated into samples with
single injury mechanism and Step 1 was performed.
— If there was no evidence of dual injury mechanism, the injury risk curve was built with the survival
analysis according to the following steps.
2.4.5 Step 5: Estimate the parameters of the Weibull, log-normal, and log-logistic distributions
with the survival analysis method
2.4.6 Step 6: Identify overly influential observations using the dfbetas statistics
The overly influential observations were identified using the dfbetas statistics.
2.4.7 Step 7: Check the distribution assumption graphically using a qq-plot or the CTE method
2.4.8 Step 8: Choose the distribution with the best fit, based on the Akaike information cri-
terion (AIC)
The distribution with the lowest AIC among the three distributions was selected.
4 © ISO 2013 – All rights reserved

---------------------- Page: 8 ----------------------
ISO/TR 12350:2013(E)

2.4.9 Step 9: Check the validity of the predictions against existing results (such as accidentol-
ogy outcome), if available
2.4.10 Step 10:
— Step 10.1: The 95 % confidence intervals of the injury risk curve were calculated with the normal
approximation of the error.
— Step 10.2: The relative sample size of the confidence interval was defined as the width of the 95 %
confidence interval at a given risk relative to the value of the stimulus at this same risk. They were
calculated at 5 %, 25 %, and 50 % risk.
2.4.11 Step 11: Provide the injury risk curve associated with the quality index based on the rela-
tive sample size of the 95 % confidence interval
A scale of quality indexes based on the relative sample size was defined with four categories (“good” from
0 to 0,5, “fair” from 0,5 to 1, “marginal” from 1 to 1,5, “unacceptable” over 1,5). The injury risk curves
associated with the quality indexes were provided. The scale was determined using biomechanical
samples in order to distribute injury risk curves in the four categories. Illustrations for one example of
each class of quality index are provided in Table 2.
Table 2 — Illustrations of the width of the 95 % confidence interval for an injury risk curve for
each of the quality indexes
Quality index Width of the confidence Example (solid line: injury risk curve, dotted line: 95 %
interval at that particular confidence intervals)
risk divided by the crite-
rion value at that risk
1
Good 0-0,5
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0 200 400 600 8001 000
width of the confidence interval at that particular risk divided
by the criterion value at that risk = 0,25
© ISO 2013 – All rights reserved 5
Risk (%)

---------------------- Page: 9 ----------------------
ISO/TR 12350:2013(E)

Table 2 (continued)
Quality index Width of the confidence Example (solid line: injury risk curve, dotted line: 95 %
interval at that particular confidence intervals)
risk divided by the crite-
rion value at that risk
1
Fair 0,5-1
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0 200 400 600 8001 000
width of the confidence interval at that particular risk divided
by the criterion value at that risk = 0,75
1
Marginal 1-1,5
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0 200 400 600 8001 000
width of the confidence interval at that particular risk divided
by the criterion value at that risk = 1,25
1
Unacceptable >1,5
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0 200 400 600 8001 000
width of the confidence interval at that particular risk divided
by the criterion value at that risk = 1,75
6 © ISO 2013 – All rights reserved
Risk (%) Risk (%)
Risk (%)

---------------------- Page: 10 ----------------------
ISO/TR 12350:2013(E)

2.4.12 Step 12: Recommend one curve per body region, injury type, and injury level
— Step 12.1: If several injury risk curves could be compared with the AIC and if the difference of the
AIC was greater than 2, then the curve with the lowest AIC was recommended over the others.
— Step 12.2: If an injury risk curve had an “unacceptable” quality index, it should not be recommended.
— Step 12.3: If several injury risk curves were still available for a given injury type and level, engineering
judgment was used to recommend one curve over another.
The recommended injury thresholds should be provided with their associated quality indexes.
3 Injury risk curves for the WorldSID 50th
The injury risk curves were constructed by correlating the dummy responses to the PMHS injuries in
the same test configurations.
The injury risk curves were built with the following steps:
Step 1: Paired PMHS injuries and dummy measurements were collected, after having selected the PMHS
sample and checking the dummy data.
Step 2: The censoring status was assigned to each pair depending on if it was left, right, interval
[42]
censored, or exact. The WorldSID 50th injury risk curves were computed with the R software . With
the R software, the right, exact, left, and interval censored data are coded 0, 1, 2, and 3 respectively.
[37]
Step 3: The injury risk curve should be built with the Consistent Threshold Estimate (CTE) in order
to check for dual injury mechanism.
Step 4:
— If there was an evidence of dual injury mechanism, the sample was separated into samples with
single injury mechanism and Step 1 was performed.
— If there was no evidence of dual injury mechanism, the injury risk curve was built with the survival
analysis according to the following steps.
However, the injury risk curves here included the age as a covariable so it would be necessary to separate
the sample into different classes of age before building the CTE injury risk curves. The resulting sub-
sample was too small to build reliable injury risk curves so dual injury mechanisms were not checked
for the WorldSID 50th.
Step 5: The parameters of the Weibull, log-normal, and log-logistic distributions with the survival
analysis method were estimated (see Table J.1).
Step 6: The overly influential observations were identified with the dfbetas test (see Table J.2 to
Table J.7). These observations were checked for any specificity. As there was no evidence of difference
between these observations and the others included in the sample, these observations were kept in the
construction of the injury risk curve.
Step 7: The distribution assumption should be checked graphically using a qq-plot or the CTE method.
However, the injury risk curves here included the age as a covariable so it would be necessary to separate the
sample into age classes before building the CTE injury risk curves. The resulting sub-sample was too small
to build reliable injury risk curves so distribution assumption was not checked in this Technical Report.
Step 8: The distribution with the best fit, based on the Akaike information criterion (AIC), was chosen
(see Table J.1 and Table 3).
Step 9: Check the validity of the predictions against existing results (such as accidentology), if available.
In the case of the WorldSID 50th, there was no prediction to be validated.
© ISO 2013 – All rights reserved 7

---------------------- Page: 11 ----------------------
ISO/TR 12350:2013(E)

Step 10: The 95 % confidence intervals were calculated, as well as the relative sample size of the
confidence interval (width of the confidence intervals at 5 %, 25 %, and 50 % relative to the value of the
stimulus at 5 %, 25 %, and 50 % of risk, respectively) (see Table J.8).
Step 11: The injury risk curves were provided with their associated quality indexes based on the relative
sample size of the confidence interval (see Table 4).
Step 12: Recommend one curve per body region, injury type, and injury level.
— Step 12.1: If several injury risk curves could be compared with the AIC and if the difference of the
AIC was greater than 2, then the curve with the lowest AIC was recommended over the others.
The samples that could be compared were those with the same PMHS sample and the same level and
type of injury. The AIC were then compared between:
— the skeletal risk AIS3+ as a function of the maximum thoracic rib deflection and viscous criterion;
— the skeletal risk AIS4+ as a function of the maximum thoracic rib deflection and viscous criterion;
— the abdomen risk AIS2+ as a function of the maximum abdomen rib deflection and viscous criterion,
as well as of the lower spine Y acceleration 3 ms;
— the abdomen risk AIS3+ as a function of the maximum abdomen rib deflection and viscous criterion.
There was no comparison possible for the shoulder and pelvis injury risk curves.
Table 3 — AIC values for the WorldSID 50th injury risk curves
Injury risk WorldSID measurement AIC
Skeletal thoracic Maximum thoracic rib deflection (measured by 1D IR-TRACC) (mm) 24,883 7
AIS3+
Maximum thoracic rib VC (measured by 1D IR-TRACC) (m/s) 29,691 1
Skeletal thoracic Maximum thoracic rib deflection (
...