Optimizing Finite Element Material Models in a Crash Test Dummy

Finite element representations of crash test dummies are widely used in the simulation of vehicle safety systems. The biofidelity of such models is strongly dependent on the accurate representation of the nonlinear behavior of the constituent rubber, plastic, and foam materials. Advanced material models are often needed to capture the dynamic response of the various parts of the dummy. The process of calibrating a finite element material model is resource-intensive, as it involves optimizing model parameters to achieve good correlation with test data under different loading conditions and rates.

altAn automated optimization workflow for the calibration of material parameters was developed using component models from the World-wide Side Impact Dummy (WorldSID) finite element model. Calibrated material parameters were verified by comparing the response of several dummy sub-assemblies with experimental data.

The WorldSID Anthropomorphic Test Device was developed by the WorldSID Task Group so that a globally harmonized side impact dummy would be available. Offering greater biofidelity than existing side impact dummies, the technologically advanced WorldSID will eventually replace the variety of dummies used in regulation and other testing.1

Numerical simulation is a key part of the process of developing vehicle passive safety systems. It is thus essential that every hardware dummy have a virtual counterpart. During a vehicle program, hundreds of simulations are performed, and reliable dummy models provide large cost savings by reducing the need for physical prototyping and testing.

The WorldSID50 finite element model has been developed in cooperation with the Partnership for Dummy Technology and Biomechanics (PDB), a consortium that includes German automobile manufacturers Audi, BMW, Daimler, Porsche, and Volkswagen. The consortium defined the requirements for the finite element dummy model and approved the release of the dummy for commercial use. The consortium also defined all material, component, and full dummy tests and requirements.

Finite Element Model and Analysis

altThe WorldSID50 mesh was provided by PDB and is based on CAD data published by ISO1, technical drawings of the hardware WorldSID 50th percentile male model, and 3D scans of the dummy parts. The WorldSID50 model consists of approximately 260,000 elements and more than 20 different materials (see Figure 1) including rubber-like compounds, super-elastic alloys (Nitinol), foams, plastics, and vinyl.

The rate-sensitive foam material finite element model is used for the foams in the dummy. This model allows for direct input of stress-strain data at each strain rate for both tension and compression. A user material that simulates super-elastic behavior is used to model both the inner and outer Nitinol ribs. Hyperelastic material models are used in conjunction with a viscoelastic model (to account for strain rate effects) to simulate the rubber-like materials.

Most of the material samples used for testing were cut directly from the hardware dummy parts, and the rest were taken from material sheets. Experimental data were available in either nominal stress-strain format or pressure-compression ratio format (for volumetric tests). For each material, data from a variety of tests — quasi-static and dynamic (strain rates from 20/s to 400/s), tension and compression, volumetric compression, shear, and biaxial tension — were used for numerical material model calibration. Overall, data from over 400 material tests were used for the calibration.