Computational modeling is an increasingly important tool in the study of injury biomechanics. This paper describes the development and validation of a seated human body finite element model as part of the Global Human Body Models Consortium (GHBMC) project. The model was developed using LS-DYNA ® (LSTC, Livermore, CA) and is intended for blunt injury prediction. The geometry of the model is based on a protocol that leverages the strengths of three clinical scanning methods;​ computed tomography (CT), magnetic resonance imaging (MRI), and upright MRI (i.e. subject in seated position). The protocol was applied to a living male volunteer (26 years, height, 174.9 cm, and weight, 78.6 kg) who met extensive anthropometric and health criteria. Computer Aided Design (CAD)) data were developed from the images, containing significant anatomical detail. Seventeen sub-substructures of the brain, 52 muscles of the neck, and all major organs of the thorax and abdomen, with associated vasculature, are represented in finite element model. The positioning of the axial skeleton and the location of organs were determined using upright MRI scans to represent the seated posture.

A region-specific development approach was used, with five Body Region Centers of Expertise (COEs) focused on meshing and regional validation of the head, neck, thorax, abdomen, pelvis and lower extremity. The regional models were then integrated into a full body model. Mesh connections between neighboring body regions were assembled using techniques based on the geometry, element type, and anatomic purpose. This consisted of nodal connections for all 1-D beam and discrete element connections (e.g. ligamentous structures), 2D shells (e.g. the inferior vena cava to right atrium), and many 3D tetrahedral and hexahedral structures (e.g. soft tissue envelope connections between body regions). In cases where node-to-node connections were not made, (e.g. 3D muscle to bone insertions), contact definitions were implemented.

The integrated full body model consists of 1.3 million nodes, and 1.9 million elements. Element types in the model are 41.0 % hexahedral, 33.8 % tetrahedral, 19.5 % quad shell, 5.1% tri shell, and 0.6 % others including beam and discrete elements. Non-linear and/or viscoelastic material models were used where appropriate. Simulations were conducted using MPP LS-DYNA R.4.2.1. The model has been validated against a number of frontal and lateral rigid impactor and sled tests. Two of these (a chest impact per Kroell and an abdominal impact per Hardy) are highlighted via computational benchmarking on a computational cluster running Red Hat Enterprise Linux 4.0. Benchmarking tests ranged from 8 to 88 nodes. Reductions in compute times are seen up to 80 CPUs. Using 64 CPUs, solution times for the 60 ms chest impact and 100 ms abdominal impact were 10 hours, 45 minutes and 12 hours, 10 minutes respectively. Through the use of a living subject, comprehensive image data, and extensive geometric validation, this model has the potential to provide a greater degree of accuracy in blunt trauma simulations than existing human body models. It will serve as the foundation of a global effort to develop a family of next-generation computational human body models for injury prediction and prevention.