Comparison of Kinematic Behavior and Injury Measures of Male THOR and GHBMC M50-O v6.0 Model in Oblique Far-Side Sled Tests

Abstract

This study was to assess biofidelity of the latest GHBMC 50th percentile male human occupant model (M50-O v6.0) and the Humanetics THOR dummy model v1.8.1 in far-side oblique collisions. The GHBMC M50-O v6.0 human model was reasonably well correlated to the kinematics of PMHS 602 in the two 60 degree oblique far-side sled tests (with 14g and 6.6g pulses) performed by UVA (Forman et al., 2013). The human model estimated injury risks for the body regions of the head, thorax, abdomen, and knee-thigh-hip (KTH) were verified with the post-test PMHS injury outcomes. Additional five simulations for the oblique far-side sled tests were performed with the paired M50-O v6.0 human model and the THOR dummy model, varying with seats (the UVA sled steel seat and a production driver seat), and central console presence. The comparative study results showed that the kinematics of the THOR model was similar to the human model in the lateral motion of upper torso but was quite different in the lower extremities. There were also noticeable differences of the chest deflections outputted from the THOR and the human model. The THOR and the human model predicted similar trends of effects of the seat and center console on the responses of the head, neck, thorax, abdomen, and KTH.

Full text

INJURY BIOMECHANICS RESEARCH
Proceedings of the Forty-Ninth NHTSA Workshop on
Human Subjects for Biomechanical Research
Comparison of Kinematic Behavior and Injury Measures of
Male THOR and GHBMC M50-O v6.0 Model in Oblique Far-
Side Sled Tests
Jay Zhao and Sungwoo Lee
Joyson Safety System Inc., Auburn Hills, Michigan, USA
This paper has not been screened for accuracy nor refereed by any body of scientific peers
and should not be referenced in the open literature.
ABSTRACT
This study was to assess biofidelity of the latest GHBMC 50th percentile male human occupant
model (M50-O v6.0) and the Humanetics THOR dummy model v1.8.1 in far-side oblique collisions.
The GHBMC M50-O v6.0 human model was reasonably well correlated to the kinematics of PMHS
602 in the two 60 degree oblique far-side sled tests (with 14g and 6.6g pulses) performed by UVA
(Forman et al., 2013). The human model estimated injury risks for the body regions of the head,
thorax, abdomen, and knee-thigh-hip (KTH) were verified with the post-test PMHS injury
outcomes. Additional five simulations for the oblique far-side sled tests were performed with the
paired M50-O v6.0 human model and the THOR dummy model, varying with seats (the UVA sled
steel seat and a production driver seat), and central console presence. The comparative study
results showed that the kinematics of the THOR model was similar to the human model in the
lateral motion of upper torso but was quite different in the lower extremities. There were also
noticeable differences of the chest deflections outputted from the THOR and the human model. The
THOR and the human model predicted similar trends of effects of the seat and center console on
the responses of the head, neck, thorax, abdomen, and KTH.
INTRODUCTION
rom the National Automotive Sample System (NASS) data, fatality risk in far-side oblique collisions
was found to be comparable to that in near-side collisions (Gabler et al, 2005, Pintar et al, 2007). Far-
side oblique collisions were the most common impact direction caused serious injuries with more than half
accounted for the head and thorax injuries (Bahous et al, 2015). The abdominal injuries especially to the liver
and spleen also occurred often in far-side collisions (Yoganandan, et al 2000).
F

In recent years new automated vehicles (AV) technologies are accelerating. Recent trends in AV
interior seating configurations bring more innovative and versatile design options than the conventional
vehicles. Other than the traditional forward-facing seats, AV seating designs may have seating positions of
oblique-facing, rear-facing, and side facing or the other angle-oriented. The oblique and side-facing seat
positions could become far-side like collision environment in the frontal or side collisions observed often
from the field.
Improvement of current restraint systems for far-side protection motivated us to develop better far-
side oblique sled test methodology with more biofidelic human occupant surrogates for evaluation of the
restraint performance. In their earlier experimental studies, Pintar et al. compared the kinematics and
responses among postmortem human subjects (PMHS), WorldSID, and THOR-NT in the far side sled tests
(Pintar et al., 2007). Forman et al. performed a series of far-side lateral and oblique sled tests with seven
PMHS to investigate effects of various restraints, positioning and collision parameters on the occupant
kinematics and injury outcomes (Forman et al., 2013). The study showed that an oblique impact direction
tended to cause increased head lateral excursion and axial rotation of the torso. Used this test data, Maika et
al. evaluated biofidelity of the earlier version of Global Human Body Model Consortium (GHBMC) 50 th
percentile male occupant model (GHBMC M50-O version 4.4) (Maika et al., 2016).
With the most recent advances in development of more biofidelic human occupant surrogates, new
generation of anthropomorphic test devices of THOR and WorldSid dummies and the FE models have been
used more extensively in various laboratories of the industry. The GHBMC 50th percentile male occupant
model version 6.0 (GHBMC AM50-O v6.0) was also developed. In this study, we were going to evaluate
biofidelity of the GHBMC AM50-O v6.0 human model and the male THOR dummy in oblique far-side
collisions. The objective of this study was to better understand biofidelity and responses of the male THOR
dummy in oblique far-side sled test conditions.
METHODS
This study was conducted in two phases: 1) correlate the latest developed GHBMC M50-O v6.0
human model with the oblique far-side sled tests performed by University of Virginia (UVA) (Forman et al.,
2013); 2) perform parametric comparative analysis on oblique far-side sled test simulations with the paired
GHBMC M50-O v6.0 and the Humanetics male THOR dummy model v1.8.1.
Human Model Correlation
Figure 1 shows the far-side sled test model with the human model (HM). The base sled test model
consisted of the FE models of the base sled, the base seat and the seatbelt built based on the physical
geometries and properties of the UVA far-side sled test fixture (Forman et al., 2013). The D-Ring position
and the shoulder belt routing for the positioned HM were set up per the PMHS test. The base sled test model
was validated from the previous study (Maika et al., 2016) with the GHBMC M50-O v4.4 model. In this far
side sled test model, the GHBMC AM50-O v6.0 model was swapped in.
Figure 1: Far-side sled test environment

GHBMC Occupant Model and PMHS Information. The GHBMC AM50-O v6.0 model represents a
male occupant of 77 kg weight, 175 cm tall and BMI of 25.1. It has the closest body mass and stature to the
PMHS-602 in the test (Table 1).
Table 1. GHBMC Occupant Model and PMHS information
GHBMC OCCUPANT MODEL AND PMHS INFORMATION
Mass [kg] Stature [cm] BMI Age
GHBMC AM50-O 77 175 25.1 -
PMHS-591 (oblique) 86 182 25.9 44
PMHS-602 (oblique) 79 178 24.9 61
PMHS-608 (oblique) 79 172 26.7 56
Initial Position of Occupant Model. The initial seating position and posture of GHBMC M50-O
model v6.0 was set as close as the measurements from the PMHS-602 (Table 2).
Table 2. PMHS/HBM Position Measurements
Case HBM/PMHS H-pt. to Seat
(mm)
D-ring to
Seat
Belt Angle
(deg)
Torso Angle
(deg)
Femur
Angle (deg)
Tibia Angle
(deg)
HM134 PMHS-602 119 525 44 80 14 43
HBM 134 81 525 45 76 13 45
HM135 PMHS-602 118 521 48 81 12 47
HBM 135 81 521 45 76 13 45
HM Correlation Cases. The two UVA 60-degree oblique far-side PMHS tests S0134 and S0135
data were used for the model correlation (Table 3).
Table 3. Test Cases for GHBMC Occupant Model M50-O v6.0 Correlation
Case Test Condition PMHS Test# PMHS #
HM134 Oblique, 6.6g S0123, S0134, S0137 591, 602, 608
HM135 Oblique,14g S0124, S0135 591, 602
HM/THOR Simulation Cases
In the 2nd phase study, the far-side sled test environment model was revised such that the base seat
was replaced with a validated production driver seat model. In addition, a center console model from a
vehicle was added for the simulation cases of HM137 and TR140 in Table 4. The commercial Humanetics
THOR dummy model v1.8.1 was integrated into the simulation cases TR138-140 in Table 4. The THOR
dummy model v1.8.1 representing the current THOR 50M dummy was the latest version further updated and
validated by Humanetics (Maatouki et al., 2018).
HM/THOR Parametric Study Simulation Cases. Table 4 lists the HM/THOR oblique far-side sled
test simulation cases defined for the parametric study.
Table 4. HM/THOR Parametric Study Simulation Cases
Case Crash Pulse, SB, Positioning Occupant Seat Seat Console
HM136
Same as Case HM135
M50-O v6.0 Production Seat No
HM137 M50-O v6.0 Production Seat Yes
TR138 THOR v1.8.1 UVA Sled Seat No
TR139 THOR v1.8.1 Production Seat No
TR140 THOR v1.8.1 Production Seat Yes
Figure 2 shows the HM/THOR oblique far-side sled test simulation models for the parametric study.

Figure 2: HM/THOR obliques far-side sled test simulation models
THOR Dummy Positioning. Table 5 lists the Thor dummy model position measurements. The
measurements of the THOR positioning targets were set as close as those of the HM.
Table 5. Thor Dummy Model Position Measurements
Case HM/PMHS H-pt. to Seat
(mm)
D-ring to
Seat(mm)
Belt Angle
(deg)
Torso Angle
(deg)
Femur Angle
(deg)
Tibia Angle
(deg)
TR138-
140
HM 81 525 45 76 13 45
THOR 122 525 45 67 13 36
Data Analysis and Processing
The data processing and analysis were performed for the three subjects: 1) the PMHS-602 tested in
the UVA sled tests S0134 and S0135; 2) the GHBMC M50-O model v6.0; and 3) the THOR model v1.8.1.
For the PMHS-602, the following test data were processed for the model correlation:
The sensor data of the accelerations of head CG, T1, and pelvis, and the forces of the seatbelts
and seat,
the Vicon measurement data of kinematics targets of the head CG, sternum, T1, T4, T7 and
pelvis, the left and right shoulders, and the left and right knees.
As shown in the Figure 3, the kinematics correlation is processed such that the measured tracking
Vicon targets were initially overlaid with the HM and were synchronized with the HM during the 150msec
time period of the simulations.
Figure 3: The HM kinematics correlation process using the synchronized Vicon targets
time-histories data to trace the HM kinematics (Courtesy: The PMHS test data provided by UVA).
The following outputs and measurements were defined and processed for both HM and THOR
models:
the accelerations of head CG, T1, T4, T12 and pelvis,

the forces and moments of the upper neck, acetabulum, and femur,
the chest and upper abdomen deflections at the measurement locations comparable to the
THOR (four chest deflections and two abdomen deflections), and
the relative displacements (to the seat) of the kinematics targets or Vicon targets of the head,
left and right acromion, T1, T4, T7, pelvis, left and right knees.
The injury risks for the body regions of head, neck, thorax and KTH were calculated separately for
the 50th%ile male HM using the injury risk functions in Table D-2 from our previous study (Zhao, 2019), and
for the THOR dummy using the injury risk functions in Table D-1 (Zhao, 2019).
The injury measures were normalized with the normalization values in Table 6. The normalization
values for the HM and the THOR were defined separately such that each pair of the HM vs. the THOR have
same estimated injury risk for the body region. It is seen from Table 6 that the normalization values for the
HM and the THOR were different since the injury risk functions were not the same.
Table 6. HM/THOR Injury Measure Normalization Values
Injury Measure HM Normalization
Value Risk THOR
Normalization Value Risk
HIC 800 15% 800 15%
BrIC 1 44% 1 44%
NIJ 1 1.1 25.7%
Chest Def-UL 73 50.5% 49.7 50.3%
Chest Def-UR 73 50.5% 49.7 50.3%
Chest Def-LL 73 50.5% 49.7 50.3%
Chest Def-LR 73 50.5% 49.7 50.3%
ABD Def-L 72 50% 50.5 50%
ABD Def-R 72 50% 50.5 50%
Femur Force-L 12 25.6% 8.2 25.7%
Femur Force-R 12 25.6% 8.2 25.7%
RESULTS
Human Model Correlation Results
Kinematics Correlation for Case HM135 with 14g Pulse in 60-Deg . Figure 4 compares snapshots
of the GHBMC M50-O model v6.0 kinematics to the PMHS -602 video from the UVA far-side sled test
S0135 (14g pulse in 60deg) at 0msec, 40msec, 80msec and 120msec. It shows that the kinematics of
GHBMC M50-O v6.0 model was similar to the PMHS.

Figure 4: Snapshots of the GHBMC M50-O model v6.0 kinematics compared to the PMHS -602
test video from the UVA sled test S0135 (14g pulse in 60deg) at 0msec, 40msec, 80 msec, and 120msec
(Courtesy: The PMHS test data provided by UVA).
Figure 5 shows snapshots of the GHBMC M50-O model v6.0 kinematics from the UVA oblique far-
side sled test S0135 (14g pulse in 60deg) traced with the synchronized In-vivo PMHS -602 Vicon targets at
0msec, 40msec, 80msec and 120msec. The movement of the GHBMC M50-O v6.0 model body targets were
close to the PMHS except for the left shoulder.
Figure 5: Snapshots of the GHBMC M50-O model v6.0 kinematics overlaid with the synchronized
In-vivo PMHS -602 Vicon targets at 0msec, 40msec, 80msec and 120msec in the UVA oblique far-side sled
test S0135 (14g pulse in 60deg). (Courtesy: The PMHS test data provided by UVA)
Figure 6 shows the time-histories of multiple kinematics targets displacements of the GHBMC
M50-O model v6.0 compared to the Vicon data of PMHS -602 from the UVA oblique far-side sled test
S0135 (14 pulse in 60deg).

Figure 6: The time-histories of the Vicon targets displacements of the PMHS -602 (Yellow curve)
in the UVA oblique far-side sled test S0135 (14g pulse in 60deg) correlated with the GHBMC M50-O model
v6.0 (Red curve) outputs. (Courtesy: The PMHS test data provided by UVA)
Kinematics for Correlation Case HM134 with 6.6g Pulse in 60-Deg . Figure 7 compares snapshots
of the GHBMC M50-O model v6.0 kinematics to the PMHS -602 video from the UVA far-side sled test
S0134 (6.6g pulse in 60deg) at 0msec, 40msec, 80msec and 120msec. The kinematics of GHBMC M50-O
v6.0 model was also similar to the PMHS in this test case.
Figure 7: Snapshots of the GHBMC M50-O model v6.0 kinematics compared to the PMHS -602
test video from the UVA sled test S0134 (6.6g pulse in 60deg) at 0msec, 40msec, 80 msec, and 120msec
(Courtesy: The PMHS test data provided by UVA).

Figure 8 shows snapshots of the GHBMC M50-O model v6.0 kinematics overlaid with the
synchronized In-vivo PMHS -602 Vicon targets at 0msec, 40msec, 80msec and 120msec from the UVA
oblique far-side sled test S0135 (14g pulse in 60deg). The movement of the GHBMC M50-O v6.0 model
body targets were also close to the PMHS except for the left shoulder in this test case.
Figure 8: Snapshots of the GHBMC M50-O model v6.0 kinematics overlaid with the synchronized
In-vivo PMHS -602 Vicon targets at 0msec, 40msec, 80msec and 120msec in the UVA oblique far-side sled
test S0134 (6.6g pulse in 60deg). (Courtesy: The PMHS test data provided by UVA)
Figure 9 shows the time-histories of the kinematics target displacements of the GHBMC M50-O
model v6.0 compared to the Vicon data of PMHS -602 in the UVA oblique far-side sled test S0134 (6.6g
pulse in 60deg).

Figure 9: The time-histories of the kinematics target displacements of the GHBMC M50-O model
v6.0 (Red curve) compared to the Vicon data of PMHS -602 (Yellow curve) from the UVA oblique far-side
sled test S0134 (6.6g pulse in 60deg). (Courtesy: The PMHS test data provided by UVA)
HM/THOR Comparative Study Results
Kinematics Comparison. Figure 10 shows snapshots of the GHBMC M50-O model v6.0 kinematics
at 10msec, 60msec, 100msec, and 135msec in the simulated sled test case TM135 (Table 3) in comparison to
the paired THOR dummy in the case TR138 (Table 4), where the blue is the Thor dummy and the yellow is
the human model (HM). It is seen that the Thor dummy and the HM had similar kinematics initially during
the time period up to 60msec. After 60msec, the HM had more lateral swing movement of the head neck and
lower legs than the Thor dummy. At 135msec, the HM lower legs raised higher and swing to the right much
more than the THOR dummy.
Figure 10: Snapshots of the kinematics of GHBMC M50-O model v6.0 in the simulated oblique
far-side sled test case TM135 (Table 3) compare to the paired THOR dummy in the case TR138 (Table 4).
Blue: the Thor dummy; Yellow: the HM.
Response Comparison. Figure 11 compares the normalized injury measures of the GHBMC M50-O
model v6.0 (case TM135) to the THOR dummy (case TR138). It shows that the normalized injury measures
of the THOR and HM were similar except for the chest deflections, of which the HM were higher than the
THOR dummy.
10 msec
60 msec
100 msec
135 msec

Figure 11: Comparison of the normalized injury measures between the GHBMC M50-O model
v6.0 (case TM135) and the THOR dummy (case TR138).
Figure 12 compares the normalized injury measures in groups among the HM cases (TM135,
TM136, TM137) and the THOR cases (TR138, TR139, TR140). It is seen that the trends of the injury
measures varying with the seat and central console were same across the two groups, although the HM had
the higher normalized chest deflections than the THOR.
Figure 12: The normalized injury measures in groups among the GHBMC M50-O model v6.0
cases (TM135, TM136, TM137) and the THOR dummy cases (TR138, TR139, TR140).
Injury Risks Comparison. Table 7 compares the injury risks of the body regions for the three HM
simulation cases (TM135, TM136, TM 137) estimated with the GHBMC M50-O model v6.0 and the three
paired THOR cases (TR138, TR139, TR140) estimated with the THOR model v1.8.1.
Table 7. HM/THOR Injury Risks for Parametric Study Simulation Cases
Body
Region Measure Risk
TM135-
UVASeat
HM Injury
Risk
TM136-
GSeat HM
Injury Risk
TM137-
GSeatCC
HM Injury
Risk
TR138-
UVASeat
THOR Injury
Risk
TR139-
GSeat
THOR
Injury Risk
TR140-
GSeatCC
THOR
Injury Risk
Head HIC AIS3+ 0% 0% 0% 0% 0% 0%
BrIC AIS 4+ 0% 2% 2% 0% 0% 0%
Neck Nt AIS3+ 0.0% 0.0% 0.0% 0.1% 0.0% 0.0%
Nc AIS3+ 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%
Thorax Chest
Def AIS3+ 82.8% 77.5% 74.7% 67.5% 58.1% 63.1%
KTH Femur
Force AIS 2+ 0.7% 0.7% 0.7% 0% 0% 0%

Discussion
From Table 7, both the HM and the THOR models predicted high injury risks for the Thorax region
and very low injury risks for the other body regions. To verify these results, the injury outcomes of the tested
PMHS 602 were checked. Table 8 summarized the post-test Autopsy examination reported results for the
PMHS 602. The model predicted high chest injury risk was in coincidence with the post-test observation. For
the reported lumbar spine injury, capability of the HM model prediction was not established, which will be
the next step for the enhancement.
Table 8. AIS Summary for PMHS 602
Body Region Injury
AIS Code 2005
(1998 where different)
Thorax
Fractures of 5 Left ribs (L4, L5, L6, L7, L9) and 4 right ribs
(R2, R3, R5, R6)
(without flail)
450203.3 (450230.3)
Thorax Sternum fracture* (note that this fracture involved one of the
sternum instrument mount holes, thus may be artifactual) 450804.2
Lumbar Spine Transverse process fracture, L2 left side 650620.2
Figure 10 shows a big difference in the kinematics of the lower extremities between the HM and the
THOR in the oblique far-side sled test. This could indicate poor biofidelity of the THOR pelvis and lower leg
body regions responding to the lateral inertia loads from the oblique far-side collisions. Another factor
affecting the Thor lower legs lateral movement could be the friction forces of the foot to the foot plate.
Figure 13 shows comparison of the contact forces of the left and right foot to the foot plate between the HM
and the THOR models from the simulated far-side sled test S0135 (14g pulse in 60deg). Although same
friction coefficient was defined for the foot to foot-plate contacts for both cases, the contact forces of the
HBM and the THOR were different. Compared to the HM, the THOR dummy had the longer duration in the
right foot contact force and higher peak force to the left foot, which was mainly due to insufficient lift up
motion of the THOR legs as compared to the observed from PMHS or the HM in the test.
Figure 13: Comparison of the contact forces of the left and right foot to the foot plate between the
HM (blue curve) and the THOR (red cure) from the simulated far-side sled test S0135 (14g pulse in 60deg).
In this study, we did not evaluate the latest WorldSID dummy behaviors in the far-side sled test
conditions. We planned to perform same evaluation study as for the THOR in the next step.
CONCLUSIONS
The latest GHBMC 50th percentile male occupant model version v6.0 demonstrated reasonably well
biofidelic kinematic responses compared to the PMHS 602 in the two 60 degree oblique far-side sled tests
(with 14g and 6.6g pulses) performed by UVA (Forman et al., 2013). However, the shoulder target
movements were seen different from the PMHS, which should be further investigated.

The severities of the injury risks for the body regions of the head, thorax, abdomen, and knee-thigh-
hip (KTH) predicted with the human model were in consistent with the post-test PMHS injury observations.
The Humanetics mid-sized male THOR dummy v1.8.1 has similar kinematics to the HM in the
upper body but is quite different in the lower extremities from the simulated oblique far-side sled test. Such
kinematics difference could indicate poor biofidelity of the THOR pelvis and lower leg body regions
responding to the lateral inertia loads from the oblique far-side collisions.
The THOR and the human model predicted similar trends of effects of the seat and center console
on responses of the head, neck, thorax, abdomen, and KTH. However, the HM model had larger chest
deflections than the THOR dummy model in the simulated 60 degree oblique far-side sled test with 14g
pulse.
ACKNOWLEDGEMENTS
Thanks for Jason Kerrigan, Richard Kent, Jason Forman at University of Virginia who provided the
PMHS data for the human model case correlation. And also, thanks are given to Scott Gayzik’s team at Wake
Forest University School of Medicine who provided the GHBMC human body model for this study.
REFERENCES
FORMAN, J., Lopez-Valdes, F., et al. (2013). Occupant kinematics and shoulder belt retention in far-side
lateral and oblique collisions: a parametric study. Stapp Car Crash Journal, 57: pp.343–85
GABLER, H. C., Fitzharris, M., Scully, J., Fildes, B. N. (2005). Far side impact injury risk for belted
occupants in Australia and the United States. Proceedings of the 19th International Technical Conference
on the Enhanced Safety of Vehicles, Washington D.C., 05-0420.
PINTAR, F. A., Yoganandan, N., et al. (2007). Comparison of PMHS, WorldSID, and THOR-NT responses
in simulated far side impact. Stapp Car Crash Journal, 51: pp,313–60.
BAHOUS, G., Murakhovskiy, D., Digges, K., Rist, H., Wiik, R. (2015). Opportunities for reducing far-side
casualties. Proceedings of the 23rd International Technical Conference on the Enhanced Safety of
Vehicles, Gothenburg, Sweden, 15-0444.
YOGANANDAN, N., Pintar, F. A., Gennarelli, T. A., Maltese, M. R. (2000). Patterns of abdominal injuries
in frontal and side impacts. Annual Proceedings of the Association for the Advancement of Automotive
Medicine, 44: pp.17–36.
KATAGIRI, M., Zhao, J., Kerrigan, J., Kent, R., Forman, J.. (2016). Comparison of Whole-Body Kinematic
Behaviour of the GHBMC Occupant Model to PMHS in Far-Side Sled Tests. IRCOBI Conference 2016,
IRC-16-88.
MAATOUKI I., Fu, S., Zhou, Z. (2018). Latest FE Model Development of THOR-50 M Crash Test Dummy.
15th International Ls-Dyna Users Conference.
ZHAO, J., Jakkamsetti, P. K., Katagiri, M., Lee, S. (2019). New Passenger Restraints with Adaptivity to
Occupant Size, Seating Positions and Crash Scenarios Through Paired ATD-HM Study. Proceedings of
the 26th International Technical Conference on the Enhanced Safety of Vehicles, Eindhoven,
Netherlands, 19-0322.

All Authors’ full name, address, and e-mail
1. Jay Zhao, Joyson Safety Systems, 2500 Innovation Drive, Auburn Hills, MI 48326, 248-
377-6019, jay.zhao@joysonsafety.com
2. Sungwoo Lee, Joyson Safety Systems, 2500 Innovation Drive, Auburn Hills, MI 48326,
248-364-6029, sungwoo.lee@joysonsafety.com