Eight-node, as well as some six-node solid elements were used to construct the bony structure of the model. Soft tissues such as pleura, fascia, muscle layers and blood vessels were modeled using membrane elements. The heart and lungs were also modeled using solid elements. The rest of the structures in the mediastinum were modeled by shell elements. This includes the superior and inferior vena cavas, the left and right brachiocephalic veins, the pulmonary trunk and the aorta, the brachiocephalic trunk, the left common carotid and subclavian arteries, the trachea and its bronchi, and the esophagus. Posterior intercostal arteries, which fix the aorta to the posterior wall of the thoracic cavity, were modeled using beam elements. Two layers of shell elements were built between the mediastinum and the lungs, one on either side, in order to represent the pleural reflection on the mediastinum. The parietal pericardium was considered to be part of those two membranes, while the visceral pericardium was modeled as if it were part of the heart.

     The diaphragm was also modeled using shell elements. Another layer of shell elements situated directly below the diaphragm was built to represent the gross stiffness of the viscera in the abdominal cavity. The vessels, trachea and esophagus were modeled as if they were connected to either the neck or the diaphragm. Including the pendulum, the model consists of 15,671 nodes, 4,333 solid elements, 45 beam elements and 11,075 shell elements.

     Free-to-separate and frictionless-sliding contacts were used for the interfacing of organs with each other during an external impact. This is quite appropriate for the interfacing of the organs within the pleural and pericardial cavities, where the lungs and heart can freely slide against their parietal pleurae. In order to accommodate the contact assumptions, other tissues that fill in between organs, such as areolar tissues and pleurae, were assumed to have very little or no effect on organs during high speed impact. A total of 25 contact interfaces were specified among organs, other tissues, and the pendulum. Figure 1 shows the whole model for a lateral pendulum impact simulation and the cross-sectional view of the model at the level of the lower sternum. Figure 2 displays the individual models of the organs in the mediastinum.

     As in most previous models, linear elastic properties were used for most of the tissues in the model in order to reduce the model complexity. Young’s Moduli were approximated from their respective nonlinear stress strain curves. Hard tissues such as bones were most readily modeled as linear elastic materials. Soft tissues, such as muscle layers, vessel walls, airways and esophagus, were simply approximated to be linear elastic. These approximations were made while maintaining the secondary high stiffness stress-strain response region for the Young’s Modulus, while ignoring the initial low stiffness-large strain physiological regions. A list of the material properties used in this model is provided in Table 1 and has been adopted primarily from references to literature by Granik and Stein ([7]), [22], [16], [9], [5], [6], [12].

     The only two organs that still retain their non-linearity are the lung and the heart. During impact, they are expected to be stiffer in compression than in tension. Hence, the effect of the non-linear stress-strain relationship would be significant. Because the dynamic response data were not available for the lung, the quasi-static uniaxial test data for the lung parenchyma from Vawter et al. ([16]) were used as a basis for the material properties of the lung. However, the magnitude was increased by a factor of 10 in order to account for the possibility of strain rate effects experienced during high-speed impacts. For the heart, the basic data were adopted from Yamada ([22]) and were treated in the same manner as the lungs.

     The model was further simplified by eliminating the hemodynamic effects of the cardiovascular system and the aerodynamics effects of the respiratory system. Additionally, the bending stiffness of the ribs may be lower than was expected, this is due to the selection of a single layer modeling technique. This model did not include the head and neck. Lower torso and extremities. The inertial effects of the head and neck were approximated by distributing their masses to adjacent bony structures. For example, the masses of the shoulder and arms were distributed to the first three ribs, and the masses of the lower abdomen and the extremities were distributed to the lumbar and pelvic regions. The model was used to simulate previous testing by Viano ([20]) in which a pendulum was used to impact the thorax. A 150 mm diameter pendulum was used to provide loading. Gravity was assumed to have no effect on the model. A commercially available three dimensional explicit finite element code, LS-DYNA3D (Livermore Software Technology Co., Los Angeles, CA) was used for model analyses.