Parametric Finite Element Model of Sport Utility Vehicle
Perhaps the most drastic automotive design verification comes from vehicle collision tests. Not only do they bring perspective to everyday driving, but also make one more appreciative of challenges facing vehicle designers. The amounts of energy and deformation involved in collision are enormous and must be well understood to be harnessed into mechanisms which will protect vehicle occupants. In recent years considerable effort has been directed towards development of computational methodologies for simulating the mechanical response of automotive structures in collisions. It has become a standard practice in vehicle design to evaluate new and existing vehicles using computational simulations using large and complex models based on finite element method (FEM).
Computational vehicle models need to capture complex deformation and interaction of vehicle parts and subsystems occurring during impact. The accuracy with which the crash behavior of the vehicle is simulated can be related to the density of the computational cells (finite elements) employed in the vehicle regions that experience significant permanent deformations. Therefore, a single vehicle crash model that would accurately simulate a wide range of impact conditions (frontal, side, offset, oblique impact, roll-over) would require high density of finite elements over the entire vehicle. Such a model, aside from being computationally tractable only on supercomputers, would generally not yield a better accuracy than a set of models that are tuned to specific impact conditions.
The problem then becomes one of generating a modeling environment that can be used for rapid generation of crash situation-specific models. These models have finite element discretization tuned to specific impact scenario, required accuracy, and computational resources that are available for the simulation. Such models will then execute much faster than a uniformly-detailed discretized model and correspondingly, allow for simpler and more efficient visualization and analysis of simulated events.
The modeling environment has been developed at ORNL that enables researches to access the vehicle models over the network through Virtual Reality Modeling Language (VRML) interface. Users are able to analyze the model, modify it by changing the provided control parameters, and generate new FEM models suited to their specific simulation objectives.
Another distinctive advantage of the parametric approach to the finite element mesh discretization is that convergence of simulation results can be easily tested by proportionally increasing or decreasing the FEM mesh density employed for the model. The FEM models that have only single finite element discretization cannot generally provide a measure of computational convergence and usually require excessive FEM mesh to ensure that the convergence is achieved. Parametric FEM model can in addition be used to develop measures that can be used for determining the type, extent, and distribution of FEM discretization for different impact situations.
Research sponsored by the National Highway Traffic Safety Administration, U.S. Department of Transportation, under Interagency agreement DOE No. 2117-I027-AI, NHTSA No. DTNH22-95-X-07229, under contract DE-AC05-00OR22725 with UT-Battelle, LLC.