Finite Element Modeling of the Small Side Impact Dummy: SID-IIs
AUTHORS:
Fuchun Zhu, York Huang, Steve Moss
First Technology Safety Systems, Inc.
47460 Galleon Drive
Plymouth, MI 48170
USA
Abstract
This paper presents the development and
validation of the new FTSS finite element model of SID-
IIs dummy for both LS-Dyna and Pamcrash solvers.
The SID-IIs FE model faithfully represents the
current design (level-C) of physical dummy hardware
from head to toe. It includes all the hardware
components, accelerometers, load cells and linear
potentiometers to measure rib deflections. Pro-Engineer
CAD data was used in the model geometry creation. Nine
common materials were used in the model, and their
performance has been proved stable in the other FTSS
dummy models. Totally the SID-IIs model consists of
about 47,000 nodes and over 65,000 elements (Figure 1).
The minimum time step has been controlled as 1.03 µs.
Validation of the model has been done on
various levels, which includes material and component
tests, such as standard SID-IIs head drop test, lateral neck
flexion test, Rib drop test, arm drop test, and pelvis plug
drop test, and full body calibration tests on different areas,
such as shoulder, thorax, abdomen and pelvis. The model
was finally validated against the sled test data. The
criteria for model validation accuracy were controlled at
less than 10% off on component level, and 15% on full
body dummy level. The model showed stable and robust
performance with reasonable accuracy in prediction.
Details about model development and validation will be
presented in this paper.
Introduction
SID-IIs, the small side impact dummy, is the
smallest side impact crash test dummy currently available
in the market, representing the anthropometry of a small
female or a thirteen-year-old child. It is designed
specifically to evaluate the performance of advanced
occupant protection systems, such as side air bags, in
automotive side impact situations. For the past two years
the IIHS, Insurance Institute for Highway Safety, has
been using the SID-IIs dummies to evaluate the
automobile crashworthiness and safety ratings on a wide
range of brands, which makes the SID-IIs dummy the
hottest topic in the automotive safety industry.
The primary objective of this modeling project is
to develop and validate a finite element (FE) model of the
SID-IIs dummy, which can be used in the safety related
automobile industry to accurately predict injury criteria in
various crashworthiness and safety applications. With its
accuracy, robustness, and repeatability the model should
serve a virtual tool for the industry to reduce the lead time
and testing cost in the new vehicle development process.
Figure 1 Assembled SID-IIs Finite Element Model
Modeling Approaches
For the past eight years FTSS has dedicated to
develop a vigorous modeling methodology for all of its
FE dummy models from child to adult dummies. The
development of SID-IIs dummy model is an example of
this proven methodology [1].
The first and crucial step is the accurate
geometry representation of each dummy component,
which can be done relatively easier than before by
employing CAD surface data, such as Pro-Engineer
models. Most of the SID-IIs dummy model components
were directly created based on Pro-E surface and line data
to ensure the faithful geometric representation. These
components include head, neck, ribs, thorax spine box,
arm foam and steel insert, pelvis foam and casting, as well
as upper leg and lower leg foam. There are a few
components which have to be digitized using a 3-D
digitizer to get the surface data, such as shoe, foot, and
some other parts. Engineering drawings are also used to
create some regular shaped components. Hypermesh [2]
is the major tool used in this phase of model geometry
building.
After all the component geometric models have
been built and verified, the second step is to assemble
them together to form sub-assemblies and then the whole
dummy assembly according to specifications required by
the relevant sub-assembly and whole dummy. These
specifications include: locations and orientations of head
CG, neck bracket, pivot joints, H-Point, loadcells and
accelerometers, as well as dimensions of dummy seating
height, and all the other external dimensions.
The third step in dummy modeling is to assign
the suitable material models and properties to individual
dummy components. The same or similar material
modeling concepts have been employed for the SID-IIs
model as for the previously developed FTSS Hybrid III
family dummy models: all the metal parts except for steel
ribs are modeled as rigid bodies such as spine box, pelvis
casting, and limb internal bones; all the deformable soft
parts such as pelvis molded foam, limb flesh foam, head
vinyl skin and neck rubber, are represented by nine
common material models. Revolute joints or spherical
joints are employed to represent each of the dummy
physical joints. In the SID-IIS dummy model the condyle
joint is a special one, which has stiffness characteristics.
All the other joints are 1-G friction joints. Loadcells have
been simplified as BEAM (in LS-DYNA) or SPRING (in
Pamcrash) elements to measure all six possible channels
of load, namely three moment outputs and three force
output. Accelerometers are defined as ELEMENT
_SEATBELT_ACCELEROMETER (in LS_DYNA) or
through local coordinate node output blocks THLOC (in
Pamcrash). Rigid BEAM or BAR elements are
extensively used to represent local coordinate axis of
loadcells, accelerometers and joints for better
visualization. Special positioning tree structures have
also been created for the following Pre-Processors:
Hypermesh, Easicrash [3] and Primer [4]. Users can
easily position the whole dummy or just limbs by using
any one of the above mentioned Pre-processors.
The next step of the dummy modeling is to first
validate the sub-assembly models against test data, then
the whole dummy assembly. This is the stage when the
fine tuning of the material model/properties has to be
performed to get better correlations between dummy
model predictions and test data. The following section
will describe the details of testing and validations.
Test Designs
In addition to the standard certification tests,
special tests for FE modeling had to be carefully designed
to collect data in terms of material properties and
component performances. Based on these collected test
data the dummy component models are to be validated.
These tests were designed to be as simple as possible to
avoid introducing complication to the model validation
process.
• Material tests
There are three types of commonly used soft
materials in a physical dummy: Ensolite foam, molded
rubber, and molded vinyl. Material properties have to be
extracted for all these materials.
A set of tests with three different drop speeds
were conducted to test out the soft Ensolite foam. Based
on the sample size, drop head force and drop head
displacement, the engineering stress/strain curve was
constructed, which would be used in the FE model as the
Ensolite foam was defined as low density foam material.
The visco-elastic behavior of the rubber and
vinyl had to be determined through dynamic frequency
sweep tests, stress relaxation tests and shear tests. From
all these test data the long term and short term shear
modulus can be determined.
Since SID-IIs dummy shares many parts with the
Hybrid III 5th percentile (h3-05) female dummy such as
head, neck, upper leg, lower leg, foot, shoe and ensolite
foam pad, the material testing results for above mentioned
three materials of h3-05 dummy can be used for SID-IIs
dummy model. These shared components can be directly
imported into the SID-IIs model from the previously
developed and validated h3-05 dummy model. Of course
the validation still has to be done to fulfill the specific
requirements of the SID-IIs dummy.
• Head Drop Test
The Figure 2 shows the SID-IIs head drop test
setup. This is a standard certification test [5].
Figure 2 SID-IIs Head Drop Test setup
The head is released from a height of 7.87 inches
(200 mm) to impact onto a rigid flat surface with a speed
of 2 m/s. The major criterion of this test is that the head
CG resultant acceleration falls between 115 and 145 g’s.
Please note that there is an angle of 35 degrees between
the horizontal plane and the head sagittal plane.
• Neck Calibration Test
The lateral neck pendulum impact test is also a
standard calibration test. The Figure 3 shows the test
fixtures and test setup.
Figure 3 SID-IIs Neck Calibration Test Setup
In this test a head form has to be used instead of
the real SID-IIs dummy head. The head form is designed
to have the same moment-inertia properties as the real
dummy head. In this way the instrumentation can be
easily mounted onto the head form. Since the dummy
neck is not symmetric about the impact middle plane, this
lateral impact test will observe the neck twisting mode
instead of pure bending mode. The test performance
specifications include impact velocity, timing of
pendulum pulse, maximum D-plane rotation, neck
moment, and others. For more details please refer to the
SID-IIs dummy user manual [5].
The following four tests are specially designed
for the FE model development purpose. More details
regarding these tests can be found in the SID-IIs FE
model component report [6].
• Arm Drop Impact Test
The simplified SID-IIs upper arm is constructed
of a steel weldment tube molded with foam and vinyl
layers. The arm will often get caught between vehicle
door trim and dummy upper torso in a side impact test.
So it is important to validate the arm foam properties.
In this test the arm was fixed to a rigid flat
surface with two pieces of double-sided tape. A drop
head weighing 4.56 kg was released from different
heights to impact onto the arm vinyl-foam surface at
different velocities. The test data were then examined to
determine if the maximum compression had been reached
before the foam material bottomed out. A stress-strain
curve was then generated from the test data of 3.8 m/s
speed.
The Figure 4 shows the arm drop test setup.
Figure 4 SID-IIs Arm Drop Test Setup
• Thorax Rib Test
This test is designed to determine the stiffness
characteristics of the rib sub-assembly with damping
material. The rib deflection and drop head acceleration
were recorded major parameters. Two series of drop test
were performed on a sub-assembly containing the spine
box, ribs, damping material strips, rib stiffeners and rib
bibs. The first series was to test the single rib (middle
thorax rib) with two impact speeds of 3 m/s and 4 m/s.
The drop head weight for this series tests was 3.3 kg. The
second series was to test the three thorax ribs together
also with the same two speeds. The drop head weight for
the second series tests was 8.3 kg. The Figure 5 shows
the test setup for the three rib drop test.
Figure 5 SID-IIs Three-Rib Drop Test Setup
• Lumbar Spine Pendulum Impact test
As pointed out above the lumbar spine pendulum
impact test is also a unique test designed only for the SID-
IIs FE model development since the lumbar spine
performance calibration test is not required by the SID-IIs
dummy specifications.
Figure 6 SID-IIs Lumbar Spine Flexion Test Setup
Standard neck pendulum test fixtures were used
for the lumbar spine flexion pendulum test. The SID-IIs
lumbar spine was connected to the pendulum through a
loadcell on one end, and connected to a mass structure on
the other end. As usual two impact speeds were excised
at 3.45 m/s and 5.15 m/s. Loadcell force, moment,
pendulum acceleration, and lumbar spine rotation were
recorded major parameters. The Figure 6 shows the test
setup for the lumbar spine flexion test.
• Pelvis Plug Drop Test
Pelvis plug is made of low density foam. It is
inserted in the pelvis acetabulum hole and transmits load
to the acetabulum loadcell upon a side impact. Since it is
subject to permanent deformation after the impact, the
pelvis plug has to be replaced after each test.
This drop tower test was to collect data of force-
compression relationship of the pelvis foam plug. A
series of drop tests were conducted at different speeds to
reach the maximum engineering strain of about 85%
before the pelvis plug bottomed out. Drop head
acceleration was the only recorded parameter. The test
data at 9 m/s were use to construct the stress-strain curve,
which is implemented in the model. The Figure 7 shows
the setup of pelvis plug drop test.
Figure 7 SID-IIs Pelvis Plug Drop Test Setup
The following four whole dummy pendulum
tests are standard certification tests required by SID-IIs
dummy specifications. FTSS test database was used in
the model validation tests for all these four standard
certifications. The following summarizes the four test
setup requirements. For more details about these four
tests please refer to the SID-IIs dummy user manual [5].
• SID-IIs Shoulder Pendulum Impact Test
The SID-IIs dummy with pants but without
jacket sits on top of two sheets of 2 mm thick Teflon
supported by a flat rigid surface. The dummy arm has to
be positioned 90 degrees relative to the vertical plane.
The left side impact pendulum weighing 13.97 kg is to
impact at the center point of the shoulder plug at a speed
of 4.5 m/s. Major criteria are the shoulder rib deflection
and probe force calculated from the product of pendulum
mass and pendulum acceleration. The Figure 8 shows the
test setup.
Figure 8 SID-IIs Shoulder Impact Test Setup
• SID-IIs Thorax Pendulum Impact Test
The same procedure of dummy preparation and
seating as for the shoulder impact test has to be observed
in the thorax impact test except that the arm has to be
removed and thorax ensolite pad has to be installed. The
same impact pendulum is used, and the differences
between these two tests are the impact location and
impact speed. For the thorax impact test the impact point
is on the center of the middle thorax rib, and the impact
speed is 4.3 m/s. More performance specifications are
required, which include spine accelerations and rib
accelerations. The Figure 9 shows the thorax impact test
setup.
Figure 9 SID-IIs Thorax Impact Test Setup
• SID-IIs Abdomen Pendulum Impact Test
For the abdomen impact test the impact location
is lowered to the center point between two abdomen ribs.
The wood impact pendulum head has to be used with a
smaller diameter. The total weight of the pendulum is
still kept at 13.97 kg. The impact speed is 4.5 m/s.
Similar performance specifications as for the thorax
impact test are required.
• SID-IIs Pelvis Pendulum Impact Test
The last standard whole dummy certification test
is the pelvis impact test. The same pendulum as for the
shoulder test is used here. The impact speed is 6.7 m/s,
and the impact location is at the center point of the pelvis
plug. Only two parameters are required as for the
performance criteria: the impact force and pelvis y-axis
acceleration.
• SID-IIs Rigid Seat Sled Test
In addition to above mentioned tests, one last test
is to test the whole dummy in the sled situation. The test
was conducted at VRTC at impact speed of 6.7 m/s (15
mph). The test severity is very high since no restraint
system has been applied to the dummy. High rib
deflections are observed.
Various data channels were recorded during the
sled test. The SID-IIs dummy model was excised to
correlate most important injury criteria.
Component Validations
This section describes the component model
validation results such as head, neck, arm, thorax ribs,
lumbar spine, and pelvis plug.
• Head
The SID-IIs head comprises a vinyl skin, which
fits over a cast aluminum skull. The rear of the skull is
detachable to allow access to the inside of the head. A tri-
axial accelerometer is mounted at the center of mass of
the head.
The skull geometry and the geometry of the
external surface of the skin were obtained by scanning
physical components. The location of the accelerometer
set was taken from drawings. Material properties were
derived from FTSS Free Motion Head Form model with
modification to fit the calibration test.
The skull was modeled as rigid shell elements.
The vinyl skin was modeled using two layers of visco-
elastic solid elements with a single integration point.
There is mesh connectivity between the skin and the skull.
The inertia properties of the skull were calculated from a
geometrically accurate solid model of the skull. Lumped
masses were added to the shell skull model until the mass
and inertia properties were correct (Figure 10).
Figure 10 SID-IIs Head Model (half cut)
The head assembly was then tested
under the standard certification test condition: 2
m/s drop speed with an angle of 35 degrees
relative to the impact surface. Figure 11 shows
the orientation of the model at the beginning of
the event.
Figure 11 SID-IIs Head Drop Test
Figure 12 shows the correlation of acceleration
to the test data. The acceleration magnitude is shown
here: solid green line stands for test result, and dotted red
line represents the model prediction. The acceleration
components are measured in the head’s local coordinate
system, not in the global coordinate system.
Figure 12 SID-IIs Head Drop Test Correlation
It is obvious that the model predicted the peak
acceleration quite well in both magnitude and timing. But
the loading and uploading duration is shorter in the model
prediction. The discrepancy was caused by the limitation
of VISCO_ELASTIC material model used to define head
skin. Now FTSS is working with a more promising
material model: OGDEN_RUBBER in both LS-DYNA
and Pamcrash to overcome this discrepancy. An updated
version of SID-IIs head model will be available in the
near future.
• Neck
The SID-IIs neck, the same as H3_05 neck is a
butyl rubber piece and aluminum segmented assembly,
with an internal steel cable to limit neck stretching and
increase durability. Each rubber disc has a partial-depth
horizontal slit on the anterior side. This results in a less
stiff response in neck extension (head bends back) than in
flexion (head bends forward). Each slit has a different
depth.
The butyl rubber material properties were
modified slightly based on the original H3_05 neck
properties, which were derived from the standard H3_05
neck calibration test. When incorporated in the SID-IIs
model, these properties gave accurate results in lateral
pendulum impact tests.
The aluminum discs were modeled as rigid solid
elements. The rubber discs were modeled using visco-
elastic solid elements with 8-point full integration. The
steel cable was modeled as elastic beam elements with
contact defined between the cable nodes and segments on
the aluminum discs. To represent the diameter of the
physical cable, the diameter of holes in the aluminum
discs was reduced by adding rigid solid elements. The
completed model is shown in Figure 13.
The validation of the neck model was performed
using the standard SID-IIs neck calibration test (Figure
14). The tests were performed at the standard velocity of
5.61m/s. The deformation of the neck during the lateral
test is shown in Figure 15. Figure 16 and Figure 17 show
the correlation of loadcell moment and shear forc