Publication

BACKGROUND: Accidents are among the top leading causes of death worldwide. In the European Union, there are about 43.000 reported deaths [1] and 1.3 millions casualties because of road traffic accidents [2] each year. The annual costs to the European Society due to these accidents are more than 160 billion Euros, which is twice the entire budget of the European Union [3]. To prevent, diagnose and treat injuries it is vital to understand the mechanics of injuries in biological tissues as well as the biomechanics of the injured human body. Finite element (FE) models of the human body present a valuable complement to other models, such as animal models and experimental human studies. Today, human body models (HBM) are recognized as important tools within traffic safety research at universities and in the automotive industry [4]. In pre-crash simulations, the occupant’s muscle tonus, bracing, and muscle reflexes have a large impact on the occupant kinematics. Therefore, it is necessary that HBM have biofidelic representations of both the active and passive properties of the musculature. HBM can be used to develop advanced restraint systems, in reconstructions of real life crashes and to develop injury criteria and thresholds. Several HBM with active muscles for impact simulations have been developed using multi body [5, 6], FE [7, 8, 9] or a combination of multibody and FE methods. In all these models, the muscle activation levels are pre-defined for each load case. Recently, researchers have developed multibody models where the muscle activation levels are governed by closed loop control. Fraga [10] has an active neck using PID-control of a motorcycle rider, Budsziewski [11] used PIDcontrol for the upper extremity and Cappon et al [12] stabilized the spine of an HBM using PIDcontrolled torque actuators for each vertebra. Methods to estimate feedback parameters have been developed using continuous perturbations [13, 14]. For traffic safety research it will be valuable to incorporate the active muscle response in a human full-body FE model to predict pre-crash kinematics and restraint interactions as well as injury due to crash loading. Therefore, the aim of this project is to evaluate the feasibility of closed loop control of active musculature in an FE environment. METHOD: The object version of the LS-DYNA FE code [15] was used for this study. The upper extremity of the THUMS v3.0 HBM [16] provided the skeletal geometry. The model of the musculature was developed based on guidelines from anatomical textbooks [17, 18]. The muscles where implemented as single beam element musculotendon units with non-linear contractile and parallel elastic components, see Figure 1.To use the muscle element as an actuator in a closed loop system its output force must be controlled. This is achieved by changing the muscle activation level. A PID-controller for the muscle elements was implemented with a subroutine for user control functions, written in FORTRAN. MAIN RESULTS: The developed musculoskeletal FE model was numerically robust and could simulate simple arm movements in flexion and extension. The model could maintain its original posture when gravity was applied. The material properties of the upper extremity musculature were successfully tuned towards experimental data. The implemented PID-controller was sucessfully tuned to counteract other external disturbances such as impact like perturbations and to simulate bracing. Validation was performed using isometric data from the literature, dynamic data from volunteer experiments using continous perturbations [14], and new experiments with brief impact like loading. The method with user defined subroutines to control muscle activation in the material models in LS-DYNA is a very promising technique for active FE HBM. This technique will be applied to a human full-body FE model to implement posture maintenance and bracing.