The pressure distributions between volunteer and seat back were also measured using a pressure mat. Simulation of muscle activation for bracing occupant was ...
Experimental and numerical studies of muscular activations of bracing occupant Hyung Yun Choi Sung Jin Sah Bumsoo Lee Hongik University Korea Hyeon Seok Cho Sung Jae Kang Mu Seong Mun Korea Orthopedics & Rehabilitation Engineering Center Korea Inhyeok Lee Hankook ESI Korea Jinhee Lee IPS International Korea Paper No. 05-0139 ABSTRACT Occupants who recognize the approaching crash tend to brace themselves. This reflexive muscular activation can affect the kinematics and kinetics of occupant during the crash event but the mechanisms of potential muscle contraction in car crash event remains poorly understood. A quantitative investigation of muscular activation has been attempted by utilizing dynamometer, sled and EMG devices with human volunteers. The experimental findings have been incorporated into the numerical investigation by utilizing a finite element model of skeletal muscular structure of human body. Eight male subjects were employed and the maximum amount of voluntary isometric muscular contraction for each limb joint at various joint angles was determined using a dynamometer and surface EMG. To mimic the approaching frontal crash and bracing, each volunteer was asked to brace himself when descending in inclined sled system began. During bracing, steering wheel and pedal forces were measured as well as the EMG signals at the volunteer’s shoulder, elbow, wrist, knee and ankle joints. The pressure distributions between volunteer and seat back were also measured using a pressure mat. Simulation of muscle activation for bracing occupant was performed using an optimization process for the joint muscle force calculations. The musculoskeletal model with the optimized muscle parameters was utilized to validate its tensing behavior against the experimental results. The computed axial compressive loads on steering wheel were respectively 144N and
178N for two sled heights which correlates quite well with the average value of test measurements (121.7±46.6N and 151.1±78.9N). The computed reaction forces at pedal and seat back also exhibited quite good agreement with the test measurements. INTRODUCTION The bracing driver in pre-impact situation tends to extend elbow and knee joints, and consequently push the pelvis back into the seat and lean backward as shown in Fig. 1.
Fig. 1 Postures of driver: Before (left) and after (right) bracing The bracing induced by reflexive contractions of joint muscles change the kinematics and kinetics of occupant during the crash. Its effects on injury risk have been also investigated: Begeman et al [1] studied the response of the human musculo-skeletal system to impact acceleration. They employed volunteers and EMG technology to identify the muscular response before, during and after the impact acceleration. It was found that the tone of the lower extremity muscles changed the kinematics of occupant and force
Choi 1
distribution of restraints. However they only focused on the bracing of the lower extremity and also could not quantify the degree of the muscle activations. Klopp et al [2] also studied the effects of the reflexive bracing, a series of computer simulations, pendulum and sled tests with Hybrid III dummies and human cadavers. It was concluded that the effect of muscular preloading was to increase the efficiency of load transmission to the leg and the preloaded legs acted as additional restraints helping the occupant ride down the crash pulse. Gordon et al [3] performed static and dynamic characterizations of human hip, knee and ankle. They computed forces and torques acting on the joints by measuring seat and pedal loads. In this study, the muscular activation of bracing occupants was quantified using a dynamometer, sled system and EMG devices with human volunteers. A deliberate process was taken in the selection of volunteers since the individual divergence in muscular structure between the volunteers might generate large deviations in the bracing test. Therefore, total 8 volunteers having similar body compositions as well as anthropometries were selected. Using the dynamometer, isometric voluntary maximal torques for 5 joints, shoulder, elbow, wrist, knee, and ankle of each volunteer were characterized. EMG signals at the pair of muscles, each representing an extensor and a flexor were also monitored for various joint angles. Assuming the maximal voluntary effort was made, the extension and flexion should have brought the maximum levels of EMG signals from the associated muscle group. The mean rectified EMG signals from the maximally contracted muscle were utilized as a reference value for computing the activation level of corresponding muscle in bracing test with a sled system. To mimic the approaching frontal crash, the inclined sled system driven by gravity was designed and built as shown in Fig. 2. Each volunteer was asked to brace himself when descending began on the slope until the sled stopped by striking an energy-absorbing barrier. During the bracing in the sled, steering wheel and pedal forces were measured from the installed load cells as well as the EMG signals from the volunteer’s joint muscles. The pressure distributions between the back of volunteer and seat were also measured using a pressure mat
Fig. 2 Sled system for bracing test Finite element modeling of skeletal muscular structure and numerical investigation of its activations were performed subsequently to the experimental study. An optimization scheme based on an ergonomic criterion [4] was adopted for the calculation of internal muscular force distributions around joints. The muscle tensing behavior of the model was validated against the test results. EXPERIMENTAL STUDY Selection procedure of volunteers The lean balance, ratio of muscle mass to the body weight and isometric voluntary maximum torques at elbow and knee joints were extra indices in addition to the anthropometric data for selecting volunteers. During the first round of the two-stage selection process, 20 out of 128 volunteers were selected based on BMI (Body Mass Index, kg/m2). The selection criterion of the BMI was 22±1 kg/m2 (Height: 1.75±0.01m, Body mass: 67±1kg). Isometric voluntary maximal elbow and knee joint torques had been measured with those 20 volunteers in the second round and 8 volunteers with responses closest to mean values were then chosen for the final tests. As a consequence, the dispersion in the final group of volunteers, e.g., standard deviation of maximal joint torques had decreased from the first round selections by 41% and 26% for elbow and knee joints, respectively. The average and standard deviation of the final 8 volunteer’s anthropometric data and body compositions are listed Table 1 Table 1 Volunteer data Height Weight BMI RA lean RL lean Age (m) (kg) (kg/m2) balance (%)* balance (%)** Mean 24.2 1.746 67.31 22.09 4.62 13.28 S.D. 1.69 0.84 1.55 0.61 0.31 0.58 : Ratio of right arm muscle mass to total body mass (%) ** : Ratio of right leg muscle mass to total body mass (%) *
Measurement of isometric maximal joint torque and voluntary muscle contraction using dynamometer In order to gauge the maximal voluntary contractions (MVC) of selected muscles, each volunteer was asked to produce the utmost isometric joint torques in a dynamometer (model: BiodexTM System 3 Pro). The EMG activities of a pair of muscles for extension and flexion were simultaneously measured using surface electrodes. The selected joint muscles for EMG measurement are listed in Table 2.
Choi 2
Table 2 Joint muscles for EMG activity monitoring in dynamometer test Muscle Extensor Joint Shoulder Posterior deltoid Elbow Medial triceps Wrist Extensor capri radialis Knee Rectus femoris Ankle Soleus* * : plantaflexor, **: dorsiflexor
90o
45o
Flexor
wrist joint Anterior deltoid Biceps brachii Flexor capri radialis Biceps femoris Tibialis anterior**
The dynamometer test setup with a volunteer is shown in Fig. 3. The measured maximal joint torques with various joint angles for five joints in upper and lower limbs (shoulder, elbow, wrist, knee, ankle) during isometric muscle contractions are shown in Fig. 4.
1
20o
knee joint
Fig. 3 Measurement of maximal voluntary joint torque in a dynamometer Torque (nm)
shoulder joint 30o
90o
150
ankle joint
100
50
0
45
90o
o
-50 -20
-10
0
10
20
30
30o
elbow joint
20o
Joint angle (degree)
( : extension(+), : flexion(-), : S.D.) Fig. 4 Maximal isometric voluntary joint torques from dynamometer test During the dynamometer test, EMG activities of representative pairs of joint muscles in Table 2 were monitored. Fig. 5 shows a typical raw data set of the elbow joint composed of torque and EMG signals
Choi 3
obtained from dynamometer and surface electrodes, respectively. 5000
60
Biceps Triceps Joint torque
25Hz ~ 250Hz band pass filter
30
0
0 0
5
10
15
-2500
20
25
Torque (Nm)
EMG (microV)
2500
Raw EMG
-30
Full wave rectify -5000
time (s)
-60
Fig. 5 Typical raw EMG signals and joint torque from dynamometer test (Elbow joint at 75o angle) Processing the raw EMG signal, rectifying, filtering (low pass filter: 250 Hz, high pass filter: 25Hz), and smoothing (LP filtering), a MR EMG (Mean Rectified EMG) signal was obtained as shown in Fig. 6. A RMS (Root Mean Square) value was then computed from the MR EMG signal, which represents an intensity of the EMG signal and an index of muscle activation level at the maximal voluntary contraction (MVC). There were considerable divergences in RMS values between volunteers in spite of their similar lean balances. This might be due to the different amount of subcutaneous fat tissue between volunteers and variability in electrode positioning relative to active muscle fibers.
Linear envelope
Mean Rectified EMG
F ig. 6 EMG signal processing Measurement of activation level of bracing muscles in sled system Fig. 7 shows a volunteer’s configuration in sled test. Two different sled heights, 0.9m and 1.0m were tried twice each and the measurements were very repeatable.
(a) Initial joint angles in sled test
(b) Volunteer in descending slope (Left: before bracing, Right: after bracing) Fig. 7 Configuration of volunteer in sled test
Choi 4
Shoulder 40
S.D. mean
15.9
%MVC
Typical profiles of reaction forces at steering wheel and pedal are presented in Fig. 8 with mean rectified EMG signals monitored at the muscles of the elbow and ankle joints. In general, the reaction force developed 0.3-0.5s after the onset of EMG activity, which is similar to the timing observed in an earlier volunteer test [1].
14.4
20 8.6
7.1
25.8
23.0 11.3
10.7 0
Extensor (Posterior Deltoid)
Flexor (Anterior Deltoid)
Extensor (Posterior Deltoid)
0.9 m
Flexor (Anterior Deltoid) 1.0 m
Sled height
50
120 Biceps Triceps
Elbow
Steering wheel
S.D. mean
60
80 30
20
%MVC
16.9
Force (N)
MR EMG (micro V)
40
40
10.3 30 41.6
35.3 0
Extensor (Medial Triceps)
10 0 3
3.1
9.1
10.6
Flexor (Biceps Brachia)
Extensor (Medial Triceps)
0.9 m
0 2
2.3
Flexor (Biceps Brachia) 1.0 m
Sled height
4 Time (s)
(a) Upper limb
Wrist
S.D. mean
70
300 %MVC
30 Tibialis anterior
200
10
16.5
14.9 31.5
27.8
Pedal
16.3
15.4
0
Force (N)
MR EMG (micro V)
Soleus 20
35.1 29.8 35
Extensor (Extensor Capri Radialis)
Flexor (Flexor Capri Radialis)
Extensor (Extensor Capri Radialis)
0.9 m
100
Flexor (Flexor Capri Radialis) 1.0 m
Sled height
0
0 2
3
Knee
Time (s)
27.8
%MVC
(a) Lower limb Fig. 8 Reaction force profiles with muscle activities (MR EMG)
S.D. mean
60
4
30
26.3
22.8
22.4 28.9
21.6
Extensor (Rectus Femoris)
Flexor (Biceps Femoris)
Extensor (Rectus Femoris)
0.9 m
Flexor (Biceps Femoris) 1.0 m
Sled height
Ankle 150
%MVC
The volunteer’s pattern of bracing in sled test was quantified by computing his ratio of joint muscle activation levels to the maximal voluntary contractions from dynamometer test mentioned in previous section. Fig. 9 shows the %MVCs, the ratios of RMS of MR EMG signals between sled and dynamometer tests. The higher sled at 1.0m height tends to induce from 5% to 20% more muscle activations in both extensors and flexors than the lower sled at 0.9m height, -except the knee joint. But quite same ratios of activations between extensors and flexor were produced from both sled heights. The extensors were significantly more activated than flexors in elbow, wrist and ankle joints while the opposite tendency found in shoulder and knee joints. There were relatively large standard deviations in the sled test comparing to the dynamometer test since the styles of the bracing might have differed greatly between volunteers.
26.6
20.7
0
77.1
70.3
S.D. mean
75 13.6
64.2
17.4
68.4
17.1
14.5
0
Plantarflexsor (Soleus)
Dorsiflexor (Tibialis Anterior)
Plantarflexsor (Soleus)
0.9 m
Dorsiflexor (Tibialis Anterior) 1.0 m
Sled height
Fig. 9 %MVC of joint muscles from sled test The average axial forces measured at steering wheel and pedal as shown in Fig. 10 also indicate that volunteers braced more at the higher sled drop resulting in larger reaction forces.
Choi 5
Steering wheel & Pedal Force
400
S.D. Mean
95
Force (N)
131
200
79 47
274
253 151
122
0 Steering wheel
Pedal
Steering wheel
0.9 m
Pedal
1.0 m Sled height
Fig. 10 Reaction forces at steering wheel and pedal Elbow joint extension and subsequent rearward rotation of the upper body developed more contact pressure on seat back as displayed in Fig. 11. The average net normal reaction force at seat back, the area integration of increased pressure by bracing is shown in Fig. 12.
Before bracing After bracing Fig. 11 Measured contact pressure distributions between volunteer and seat back Seat back reaction force
S.D. Mean
1000
Force (N)
crashworthiness simulation [5]. Each body segment in the H-model, in a version aiming for muscle tensing simulation, was defined as a rigid body and was linked by the anatomical joints and with the relevant skeletal muscles represented by bar finite elements [5]. The incorporated sixteen major skeletal muscles modeled by Hill type one dimensional bar elements [6] are listed in Table 3. The articulated joints were modeled with kinematic joint elements whose characteristics were designed to have no resistance within the range of motion such that only muscle forces could be the source of joint torques. Seeking the average of active isometric muscle force-length relations of the model, the maximal forces (Fmax) of each muscle at various lengths with different joint angles were computed based on the maximal isometric voluntary joint torques obtained from dynamometer test in Fig. 4. In the case when multiple muscles were involved for the same articulation DOF, e.g., biceps brachii, brachialis, and brachioradialis for elbow flexion, an optimization algorithm was adopted to determine the likely distribution of the muscle forces (design variable) by minimizing the active muscle energy (objective function) for static equilibrium (constraints). The sequential response surface method in HyperOpt [7] was selected for the optimization process. Fig. 14 shows the result of computed isometric maximum muscle forces for shoulder, elbow, knee, and ankle joints.
233 500
181 604
486 0 0.9 m
Sled height
1.0 m
Fig. 12 Reaction force at seat back NUMERICAL STUDY Numerical investigation of muscular activation was performed subsequent to the experimental study presented in the first part of this paper.
Anterior view Lateral view Posterior view Fig. 13 H-model with skeleton and muscles Table 3 Skeletal muscles in H-model for the simulation of bracing Flexors
Extensors
Human body modeling The H-model, shown in Fig. 13, is a finite element human body model representing the 50% male anthropometry. This model is widely used for
Choi 6
Shoulder Flexion
3 1
Anterior deltoid Pectoralis major (upper) Flexion torque (mean± S.D.)
Shoulder
Muscle force (kN)
4
150
3 100 2 50 1
5
Joint torque (Nm)
4
2
NA
8
0
0 30
6
45
60
75
90
120
Joint angle (degree)
Elbow 7
10
9
Knee
13
12
(a) Shoulder joint
11
Elbow extension
11 14
Ankle
16
15
100
Tric eps Extension torque (mean± S.D.)
3
Joint torque (Nm)
Muscle force (kN)
4
2
1
0
0 10
30
1. Anterior deltoid 2. Pectoralis major (upper part) 3. Posterior deltoid 4. Teres major 5. Biceps 6. Brachialis 7. Brachioradialis 8. Triceps 9. Biceps femoris 10. Semitendonous 11. Gastrocnemius (lateral & medial) 12. Rectus femoris 13. Vastus intermedius 14. Tibialis anterior 15. Extensor digitorium 16. Soleus
50
70
90
115
Joint angle (degree)
Elbow Flexion
Shoulder extension
NA
0 45
60
75
90
120
Muscle force (kN)
Joint torque (Nm)
Muscle force (kN)
40
30
Bic eps Brac hialis Brac hioradialis Flexion torque (mean± S.D.)
80
50
1
0
0
0
100
Joint torque (Nm)
2
Teres major Pos terior deltoid Extens ion torque (mean± S.D.)
1
10
30
50
70
90
115
Joint angle (degree)
Joint angle (degree)
(b) Elbow joint
Choi 7
Knee extension
200 2 100 1
0
0 10
30
50
70
90
Muscle force (kN)
3
Soleus Gastrocnemius (lateral) Gastrocnemius (medial) Plantaflexion torque (mean± S.D.)
3
2
100
1
50
0
0
110
-20
Joint angle (degree)
150
Joint torque (Nm)
300
Joint torque (Nm)
4
Muscle force (kN)
Ankle plantaflexion
Rectus femoris V as tus intermedius Extension torque (mean± S.D.)
-10
0
10
20
30
Joint angle (degree)
Knee Flexion
Muscle force (kN)
4
200
(d) Ankle joint 150
3 100 2 50
1
Joint torque (Nm)
Bic eps femoris Semitendonous Gas trocnemius (lateral) Gas trocnemius (medial) Flexion torque (mean± S.D.)
5
0
0 10
30
50
70
90
Joint torque
Muscle force
Fig. 14 Computed isometric maximum muscles forces (Fmax) (wrist joint was not performed) Simulation of bracing occupant
110
Joint angle (degree)
The seat, floor panel and steering wheel of the sled system were added to the H-model with driving posture as shown in Fig. 15. Sliding contact interfaces were defined between the seat and the skin part of the H-model. Translational motions of hands and feet were respectively tied to steering wheel and pedal such that the forces generated from the muscle bracing could be transmitted.
(c) Knee joint Ankle dorsiflexion
Tibialis anterior Extens or digitorium Dors iflexion torque (mean± S.D.)
50 40 30
0.5 20
Joint torque (nm)
Muscle force (kN)
1
10 0
0 -20
-10
0
10
Joint angle (degree)
20
30
Displayed with skin without skin Fig. 15 Configuration of H-model for the simulation of bracing occupant In the simulation, the average values of %MVC in joint muscles, the ratios of RMS of mean rectified EMG signals between sled and dynamometer tests which are listed in Table 4 were applied as activation levels of the bracing muscles. The reaction forces at the steering wheel, pedal, and seat back were then computed until they statically equilibrated with the imposed bracing muscle forces. The simulation results correlate quite well with the experimental measurements as shown in Fig. 16
Choi 8
Table 4 Average volunteer’s muscle activation levels used for bracing simulation 1.0 m Ratio* Extensor Flexor 0.46 3.88 0.55 0.75 4.43
11.3 41.6 16.3 20.7 68.4
25.8 10.6 31.5 26.6 17.1
Ratio* 0.44 3.93 0.52 0.78 3.99
313
Muscle force (N)
Sled height 0.9 m Muscles Extensor Flexor Joint Shoulder 10.7 23.0 Elbow 35.3 9.1 Wrist** 15.4 27.8 Knee 21.6 28.9 Ankle 64.2 14.5
Shoulder
400
0.9m
278
1.0m
230 203 200
51
58
55
54
0 Anterior deltoid
Pectoralis major
Posterior deltoid
Flexors
Teres major
Extensors
*: Ratio=Extensor/ Flexor, ** Wrist joint is not included in the model Elbow 1200
Reaction forces
800
1134
0.9m
980
Force (N)
604
Test (mean±S.D.) 486 400 274
253 122 144
Muscle force (N)
1.0m Simulation
82
151 227
540
178
242
586
Pedal
Seat back
Steering wheel
Pedal
Seat back
0.9m
116 139
100
28
34
0
0 Steering wheel
600
Biceps
Brachialis
Brachioradialis
Triceps
Flexors
Extensor
1.0m
Knee 700 Muscle force (N)
Fig. 16 Comparison of reaction forces between volunteer test and simulation The simulated driving posture altered by muscle tensing is illustrated in Fig. 17. There is a noticeable straightening of arms and an elastic penetration into the seat surface in the bracing position.
640
610
0.9m 1.0m 299 318
350 138 128
250 236 127 120
0 Biceps femoris Semitendonous Gastrocnemius Rectus femoris
Extension
Flexors
Flexion
Extensors
Ankle 700 Muscle force (N)
Initial posture and joint torques (before bracing)
Vastus intermedius
617
653
0.9m 1.0m 299 318
350
38
45
23
27
0 Tibialis anterior
Extensor digitorium
Dorsiflexors
Gastrocnemius
Soleus
Plantarflexors
Fig. 18 Computed bracing muscle forces CONCLUSIONS After bracing Displayed with skin without skin Fig. 17 Simulated bracing posture The computed peak muscle forces during the bracing, which are proportional to the activation levels multiplied by the isometric maximum voluntary forces at corresponding joint angles, are shown in Fig. 18. The net joint torque generated by tensing of each muscle depends on the effective moment arm of the muscle with respect to the corresponding joint center.
Vehicle occupants tend to brace in anticipation of a crash and this pre-crash muscle tensing can change the kinematics and kinetics of the occupants. The pattern of extremity bracing, i.e., shoulder, elbow, wrist, knee and ankle joints was quantitatively analyzed by volunteers EMG test. For shoulder, elbow and ankle joints, activations of extensors were substantially higher than those of flexors. However, an opposite trend was found at wrist and knee joints. The reaction forces at steering wheel, pedal and seat back were also measured to identify the degree of muscle tensing. Numerical simulation of muscle tensing was
Choi 9
performed to verify the finite element human body model. The simulated muscle tensing behavior of the model such as amounts of reaction forces at the steering wheel, pedal and seat back correlated quite well with the test results. It was the first step in the development of human body model to investigate the effect of muscle tensing on occupant kinematics and kinetics. A crash simulation with likely dynamic muscle activations taken into consideration would follow as a next step. ACKNOWLEDGMENT The authors would like to thank Prof. Richard Kent at Center for Applies Biomechanics of UVA for reviewing the manuscript and providing a constructive commentary. The authors also would like to thank Dr. Eberhard Haug for his advice on the simulation process. This study was supported, in part, by Korean ministry of science and technology. (Project No.: M10139080002-04L1008-00210) REFERENCES [1] Begeman, P., A. King, R. Levine, and D.C. Viano. 1980. “Biodynamic response of the Musculoskeletal System to Impact Acceleratio” Proc. 24th Stapp car crash Conf., SAE paper 801312 [2] Klopp, G. S., J. R. Crandall, E. M. Sieveka, and W. D. Pilkey. 1995. “Simulation of Muscle Tension in Pre-impact Bracing”, IRCOBI Conference on the Biomechanics of Impact [3] Gordon, S., P. ORticke, J. Prince, and R. Mcmeekin. 1977. “Dynamic Characteristics of Human Leg Joints" Proc. 21st Stapp car crash Conf., SAE paper 770924 [4] Haug, E. Allain, J.C., and H. Choi. 2001. “Modeling of Ergonomics and Muscular Comfort” Journal of KSME, Vol. 15 No. 7, pp 982-994. [5] Handbook of Numerical Analysis, Volume XII: Computational Models for the Human Body – Human Models for Crash and Impact Simulation, pp 297-361, Elsevier, 2004 [6] PAM-CRASH/SAFE USERS AND THEORY MANUAL, VERSION 2004, ESI GROUP, FRANCE [7] HYPER-OPT, ALTAIR ENGINEERING. INC, USA
Choi 10