A multi-body head and neck model for low speed rear impact analysis

12th IFToMM World Congress, Besançon (France), June18-21, 27 A multi-body head and neck model for low speed rear impact analysis S. Himmetoglu *, M. Acar, A.J. Taylor, K. Bouazza-Marouf Mechanical and Manufacturing Engineering Loughborough University Loughborough, LE11 3TU, UK Abstract In this study, a 5 th percentile male head and neck model has been developed using multi-body dynamics to analyse low speed rear impacts and the resulting whiplash injury effects. The model has been validated by using data from JARI (Japanese Automobile Research Institute) rear impact sled tests conducted with 7 volunteers in normal driving posture on a rigid seat without head restraint and seat belt. The developed model is simple, biofidelic and computationally very efficient. The results show that the model can represent the rear impact response of a human with a high degree of accuracy. Keywords: multi-body head and neck model, whiplash, rear impact, volunteer sled tests I. Introduction Injury to the human neck is a frequent consequence of car accidents and has been a significant public health problem for many years. The term whiplash is used to describe these injuries in which the sudden differential movement between the head and torso leads to abnormal motions within the neck, causing damage to its soft tissue components. It has been estimated that 8% of personal injury claims made against British Insurers are related to whiplash injury, costing around 1 billion annually with this figure rising yearly [1]. Whiplash can occur in all impact directions, but is most commonly reported as a consequence of rear impacts. The term Delta-V is typically used to classify impact severity, and it is defined as the area under the acceleration-time curve of the struck vehicle over the course of the impact. The mean or peak acceleration value of this curve is also specified in addition to Delta-V, to better indicate the severity of rear impacts. The most common rear impact configuration in which whiplash injuries occur has been recorded to be full overlap straight ( to 5 angled impact) with a Delta-V between 1 and 2 kph. Around 5% of whiplash injuries have been observed in low speed (or severity) rear impacts which are classified as impacts of Delta-V upto 15 kph [2, 3]. Rear impact head and neck motions have been analysed by various multi-body models in the literature. In lumped models, all the mechanical behaviour of soft tissues is lumped into the intervertebral joints [4, 5]. In detailed *E-mail: s.himmetoglu@lboro.ac.uk models, the head and vertebrae are modelled as rigid bodies and soft tissues (intervertebral discs, facet joints, ligaments, muscles) are usually modelled as massless spring-damper elements [6-8]. In head and neck models muscles are modelled to have passive and active properties. Passive muscle behaviour can be considered to represent cadaver type response, whereas active muscle behaviour represents the muscle contractions as seen in initially relaxed human volunteers subjected to rear impact sled testing [9, 11]. Detailed models are able to simulate both passive and active muscle behaviour, whereas lumped models in the literature do not have the complexity to separately represent the effects of muscle contractions. Numerous volunteer and cadaver experiments are published in the literature, however, the extent of studies that are suitable for validation of rear impact models is limited. In most cases information about the experimental set-up and the complete time history of responses are not given. Volunteer sled tests can only involve low severity impacts for the safety of subjects. Cadavers are not able to fully represent actual human behaviour since they usually represent older, fragile subjects and have no muscular tone and reflexes. Considering these limitations, JARI volunteer sled test data [1] has been found to be the most suitable for rear impact validation of human body models. The head and neck model developed in this research has lumped properties, but it has the ability to represent the effects of muscle contraction. This makes it biofidelic, and at the same time computationally efficient. II. The JARI rigid seat volunteer sled tests The model has been validated against published data from JARI rigid seat volunteer sled tests. In these tests seven healthy 5 th percentile male volunteers (25±4 years of age) in normal driving posture were subjected to rear impacts. A rigid wooden seat without head restraint and seat belt was mounted on a sled sliding on a long rail at an angle of 1 with the horizontal, as shown in figure 1. At the end of the rail, the sled engaged an oil damper at an impact speed of 8 kph, resulting in a Delta-V of 9.3 kph [1]. The resulting impact pulse is given in figure 2. Several film targets and accelerometers were attached to the head-neck and torso to record the displacements and accelerations.

Fig. 1. JARI Volunteer test setup (Adapted from [1]) 5 equivalent resistance of soft tissues at each joint. The model was developed using MSC VisualNastran 4D multi-body dynamics simulation package. The initial configuration of the model corresponds to a 5 th percentile male occupant s head and neck in normal driving posture with the head looking forward. As shown in figure 4b, an arc having a 19 mm radius and a sector of 37, as drawn between OC (occipital condyles, the joint between the skull and the first vertebra) and C7 lower end plate, was shown to be a good approximation for the curvature of the neck [4]. The drawn arc represents a common type of neck curvature in occupants with a normal driving posture [12]. Sled x-acceleration [m/s 2 ] 4 3 2 1-5.5.1.15.2.25.3 Fig. 2. Average sled x-acceleration [1] Electromyography (EMG) measurements indicated that most of the volunteers were relaxed before the impact. EMG activity for the sternocleidomastoid (SCM) muscles was found to be substantially higher than the other muscles, hence this muscle group was dominant in active muscle behaviour. A typical EMG response for the SCM muscles obtained in JARI tests [9, 11] at an impact speed of 8 kph, is shown in figure 3. Fig. 3. Typical EMG response of SCM muscles in JARI tests [9] III. Model development The multi-body model developed in this study is composed of a head, seven neck segments and a body representing T1 (the first thoracic vertebra), as shown in figure 4a. The neck segments have identical geometry and represent the vertebrae C1 to C7 from top to bottom. The inertial properties of each neck segment represent the equivalent mass and moments of inertia of the vertebra and the surrounding soft tissues. The system is driven by specifying the motion at T1. The bodies are connected by revolute joints producing resistive torques opposing the motion. These intervertebral joint torques designate the (a) Fig. 4. (a) The multi-body head-neck model in its initial configuration (b) The initial neck curvature (Adapted from [4]) The relative orientations, inertial properties and the positions of the centres of gravity of the human neck segments are given in reference [6] (available online) and can be found in summarised form in Table I. The neck segments of the model are designed to have the same geometry for simplicity, as shown in figure 5. The angle θ indicates the relative orientation of a body with respect to its adjacent body below. The relative orientations of the human neck given in Table I and the arc specified in figure 4b are used to position the neck segments. The last column in Table I indicates the relative orientations in the multi-body head and neck model. Compared to the human neck, the relative orientations of C1 and C2 were slightly modified such that the arc approximately cuts the neck segments in half. This geometry resulted in a height of 17.4 mm for each neck segment. The intervertebral joints are placed at the intersections of the arc and the neck segments. The segments have the inertial properties given in Table I, but are assumed to have a uniform density distribution. This assumption accurately approximates the positions of the centres of gravity of the neck segments. It is difficult to define the axis of rotations between adjacent vertebrae. Several researchers [14] recorded the locations of instantaneous axes of rotation (IAR) by using lateral X-rays when subjects voluntarily moved their heads between full flexion and full extension configurations. However, in rear impact conditions, the sudden differential movement between the head and torso forces the head and neck to behave in an unnatural way. (b)

Information on IARs under rear impact loading conditions is also very limited in the literature [9, 13]. Considering the satisfactory responses of BioRID II P3 (Biofidelic Rear Impact Dummy) [1] and the lack of information on dynamic IARs, it should be an acceptable approximation to place the intervertebral joints on the specified arc, which in fact passes close to the voluntary (normal) flexion/extension IARs specified in reference [14]. Body Mass (kg) Principal moments of inertia (kg.cm 2 ) Relative orientation θ(deg) I xx I yy I zz Human MultiB. neck neck Head 4.6 18 24 221 5. C1.22 2.2 2.2 4.2-1.2 C2.25 2.5 2.5 4.8-3.8 C3.24 2.4 2.4 4.6-5.3-5.3 C4.23 2.3 2.3 4.4-4.7-4.7 C5.23 2.3 2.3 4.5-5.2-5.2 C6.24 2.4 2.4 4.7-5.6-5.6 C7.22 2.2 2.2 4.3 2.8 2.8 T1 -- -- -- -- TABLE I. The initial configuration and inertial properties of a 5 th percentile male human head-neck (Adapted from [6]). using a time varying damping coefficient and a scaled version of the nonlinear stiffness function, in which the torque values are multiplied by a factor of.75. It was observed that damping played a crucial role in obtaining satisfactory responses. Therefore, a time varying damping coefficient function based on the EMG response time history such as the one shown in figure 3, was found to realise the JARI volunteer responses throughout the motion. Increasing damping in accordance with the muscle activity pattern realises active muscle behaviour and it has the potential to reduce head rotation and head angular velocities at the same time. Figure 7 shows five different phases of the damping coefficient function. Between to 75 ms, no significant active muscle response occurs. At around 75 ms, muscle discharge (mainly SCM) starts. Maximum EMG response is achieved at around 1 ms and maintained until 15 ms. The muscular discharge decreases between 15-25 ms and almost disappears after 25 ms. Torque [Nm] 2 15 1 5-5 Extension -1-15 -2 Flexion -.2 -.1.1.2.3 Angle [rad] Fig. 6. Rotational stiffness of the intervertebral joints [5] Fig. 5. (a) The neck curvature (OC to head C.G. distance given in [6]) (b) Neck segment reference frame axes and orientation Rotational stiffnesses for the intervertebral joints were derived from the nonlinear torque versus angle relation presented in reference [5], as shown in figure 6. This function was based on quasistatic experiments and adjusted to integrate the contribution of muscles. However, it doesn t include the effects of muscle contraction. The viscoelastic feature and the dynamic stiffening behaviour of the soft tissues are realised by rotational damping coefficients. For simplicity, the same stiffness and damping properties have been used for all the intervertebral joints. During validation, a good agreement between the JARI volunteer responses and the model has been achieved by Fig. 7. Damping coefficient variation IV. Model validation A. Kinematic responses The developed multi-body head and neck model has been validated against JARI volunteer responses by simulating the rear impact sled testing conditions explained in Section 2. Figure 8 shows the mean values of T1 accelerations and rotation obtained from JARI volunteer sled tests, together with their upper and lower limits which corresponded to mean ± SD (standard deviation). The mean values have been calculated and used as inputs for T1 motion.

A detailed set of responses that compare the responses of JARI volunteers and the multi-body head and neck model are shown in figure 1. The complete set of JARI volunteer responses are provided in reference [6] and are shown by the grey lines. For validation, several responses have been analysed. T1 trajectory and rotation are displayed with respect to the sled. Head angle is displayed both with respect to the sled and T1. OC with respect to T1 displacements are expressed in the T1 anatomical coordinate system (figure 4a). Head C.G. x-acceleration and head angular acceleration are expressed in the head coordinate system (figure 4a). Displacements of the head with respect to the torso, accelerations, intervertebral motions and neck forces/moments can provide good predictions for whiplash injury. These variables are in fact used in various injury criteria that have been reported in the literature [15]. T1 x-acceleration [m/s 2 ] 5 4 3 2 1 mean + SD mean mean - SD -1.1.2.3 5 T1 z-acceleration [m/s 2 ] 4 3 2 1-1 -2-3.1.2.3 by a parameter variation analysis in which the pre-stress, reflex time and activation levels of flexor, extensor and SCM muscle groups were tuned. B. Occipital condyle (OC) loads In volunteer tests, the loads experienced by the soft tissues of the neck cannot be measured directly by any means, therefore they can only be calculated. Dynamic equilibrium equations are applied to find the OC loads F shear, F normal and M defined respectively along the x, z and y axes of the head coordinate system. The forces and moments are calculated as if OC takes all the loads as shown in figure 9. In reality, the connection between head and upper neck is much more complex involving ligaments, muscles and the intervertebral disc. Thus, the values of neck forces and moments may vary depending on how they are defined or calculated. Figure 11 shows the OC loads of the model and some other studies for the impact conditions in Section 2. The response denoted by Avg. Volunteer is the average or the most representative response of the volunteers recorded by Ono et al. [9]. The initial value of F normal corresponds to the weight of the head. T1 rotation [deg] -5-1 -15-2 -25-3.1.2.3 Fig. 8. T1 rotation and accelerations (mean±sd) (Adapted from [1]) The responses of the developed model have been compared with the responses of other models and dummies which were subjected to the same impact conditions, as described in Section 2. The comparative graphs are also presented in figure 1. All the responses of other models and dummies are displayed where available. BioRID II P3(1) and H III (Hybrid III: frontal crash test dummy) responses are provided by Davidsson et al. [1]. BioRID II P3(2) responses are given by Viano and Davidsson [16]. TNO responses are the results of a detailed head-neck model by van der Horst [6], which was built in Madymo and integrated into the TNO human body model (Madymo is a widely used multi-body and finiteelement package of TNO Automotive). The displayed responses for TNO indicate the best agreement achieved Fig. 9. Calculation of OC loads C. Description of motion phases The motion phases and variation of OC loads were described in detail in references [9] and [11], based on two JARI rigid seat rear impact test series with identical conditions. As given below, the model behaviour accurately matches the detailed description given by these authors. The motion of the head and neck model is also illustrated in figure 12. Phase 1 (-5 ms): Both the head and T1 accelerations start at around 25 to 35 ms. Changes in T1 displacements indicate the development of spine extension and straightening. No significant head or neck motion occurs in this phase. A small amount of neck bending moment is observed. Phase 2 (5-1 ms): Spine straightens strongly and torso ramps. Head retracts wrt T1 due to its inertia and an S-shape develops. Thus, flexion in the upper vertebrae and extension in the lower vertebrae is observed. The S-shape becomes very distinct at around 8 ms. The maximum axial compression force occurs at around 5 ms as the

head retracts without any significant rotation with respect to the sled. Maximum T1-z acceleration also takes place at around 5 ms and a quick rise in T1-z displacement is seen. During this phase, shear force increases gradually. Phase 3 (1-15 ms): T1 rotation reaches its maximum at around 13 ms. The largest T1-x and z displacements with respect to sled occur at around 15 ms. Around 13 ms, upper neck flexion starts to transform into extension and head extension with respect to T1 starts to become significant. Therefore, head and neck extension can be considered to start at around 13 ms. The highest rate of head and neck extension takes place at around 15 ms. In other words, the highest relative angular velocities of adjacent vertebrae and the highest angular velocity of the head with respect to T1 are found around this time. In this phase the acceleration of the head becomes maximum at around 13 ms. Also, shear force reaches its maximum between 125 and 15ms. Phase 4 (15-3 ms): During this phase, head extension angle becomes maximum. Head acceleration starts to drop gradually after 15 ms, but then remains roughly constant in the later stages. Neck bending moment, shear and axial forces also start to drop gradually, but this trend changes around 22 ms in accordance with the variation in T1 accelerations. Fig. 1. Comparison between the developed model and other studies

Fig. 11. Occipital condyles loads, model versus other studies Fig. 12. The motion of the developed model V. Conclusions The rigorously validated multi-body head and neck model presented in this study has a neck with lumped properties and does not contain separate muscles but, using a set of stiffness and damping parameters for the intervertebral joints, it is able to represent true muscle behaviour by simulating the effects of muscle contraction as a function of time. The model is capable of producing biofidelic behaviour and its response shows the precision of a detailed head and neck model. It also shows superior responses overall when compared with the currently used models and dummies, and yet, the presented approach is simple, effective and computationally very efficient. Therefore, the model can represent low speed rear impact response of a human with a high degree of accuracy, hence it can be economically used as the head-neck section of a rear impact human body model to accurately compare crash scenarios and to predict injury. References [1] THATCHAM, Whiplash Research.THATCHAM.ORG, the motor insurance repair research centre, http://www.thatcham.org/, 25. [2] Langwieder, K., and Hell, W. Proposal of an International Harmonized Dynamic Test Standard for Seats/Head Restraints. Traffic Injury Prevention, 3(2):15-158, 22. [3] IIPWG International Insurance Whiplash Prevention Group Minutes, Report of the 5th Meeting, Avila, Spain, pp.1-7, 23. [4] Linder, A. A new mathematical neck model for a low-velocity rearend impact dummy: Evaluation of components influencing head kinematics.accident Analysis and Prevention, 32(2):261-269, 2. [5] Jakobsson, L., and others. Analysis of Different Head and Neck Responses in Rear-End Car Collisions Using a New Humanlike Mathematical Model. In Proc. of Int. IRCOBI Conf., pp.19-125, Lyon, France, 1994. [6] van der Horst, M.J. Human head neck response in frontal, lateral and rear end impact loading: modelling and validation. PhD Thesis, Eindhoven University of Technology, Eindhoven, The Netherlands, 22. (http://library.tue.nl/csp/dare/linktorepository. csp?recordnumber=55447) [7] van Lopik, D.W. A computational model of the human head and cervical spine for dynamic impact simulation. PhD Thesis, Loughborough University, Loughborough, UK, 24. [8] Stemper, B.D., Yoganandan, N., and Pintar, F.A. Validation of a head neck computer model for whiplash simulation. Medical & Biological Engineering & Computing, 42(3):333 338, 24. [9] Ono, K., Kaneoka, K., Wittek, A., and Kajzer, J. Cervical Injury Mechanism Based on the Analysis of Human Cervical Vertebral Motion and Head-Neck-Torso Kinematics During Low-Speed Rear Impacts. SAE Paper No. 97334. [1] Davidsson, J., Ono, K., Inami, S., Svensson, M.Y., and Lövsund, P. A Comparison between Volunteer, BioRID P3 and Hybrid III performance in Rear Impacts. In Proc. of Int. IRCOBI Conf., pp.165-178, Sitges, Spain, 1999. [11] Ono, K., Kaneoka, K., and Inami, S. Influence of Seat Properties on Human Cervical Vertebral Motion and Head/Neck/Torso Kinematics During Rear-end Impacts. In Proc. of Int. IRCOBI Conf., pp.33-318, Göteborg, Sweden, 1998. [12] Klinich, K.D., and others. Cervical Spine Geometry in the Automotive Seated Posture: Variations with Age, Stature and Gender. SAE Paper No. 24-22-14. [13] Kaneoka, K., Ono, K., Inami, S., Hayashi, K., and Bogduk, N. Motion Analysis of Cervical Vertebrae During Whiplash Loading. Spine, 24(8):763-769, 1999. [14] Dvorak, J., Panjabi, M.M., Novotny, J.E., and Antinnes, J.A. Invivo flexion/extension of the normal cervical spine. Journal of Orthopaedic Research, 9:828-834, 1991. [15] López-Valdés, F., Mansilla, A., Martín, R., and Muñoz, D. A Study of Current Neck Injury Criteria Used for Whiplash Analysis. Proposal of a New Criterion Involving Upper and Lower Neck Load Cells. In Proc. of 19th ESV Conf., Paper No. 5-313-O, Washington DC, USA, 25. [16] Viano, D.C., and Davidsson, J. Neck Displacements of Volunteers, BioRID P3 and Hybrid III in Rear Impacts: Implications to Whiplash Assessment by a Neck Displacement Criterion (NDC). Traffic Injury Prevention, 3(2):15-116, 22.