Estimating local thickness for finite element analysis

  • Authors:
  • Vânio Ferreira;Luís Paulo Santos;Ricardo Simoes;Markus Franzen;Omar O. Ghouati

  • Affiliations:
  • CCTC, Universidade do Minho, Portugal;CCTC, Universidade do Minho, Portugal;I3N, Universidade do Minho, Polytechnic Institute of Cavado and Ave, Portugal;Ford Research & Advanced Engineering Europe, Germany;Ford Research & Advanced Engineering Europe, Germany

  • Venue:
  • MACMESE'10 Proceedings of the 12th WSEAS international conference on Mathematical and computational methods in science and engineering
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Within the development of motor vehicles, crash safety is one of the most important attributes. To comply with the ever increasing requirements of shorter cycle times and costs reduction, car manufacturers keep intensifying the use of virtual development tools, such as, for crash simulations, the explicit finite element method (FEM). The accuracy of the simulation process is highly dependent on the accuracy of the model, including the midplane mesh. One of the roughest approximations typically made is the actual part thickness which, although most frequently modelled as a constant value, can, in reality, vary locally. Availability of per element thickness information, which does not exist explicitly in the FEM model, is one key enabler and can significantly contribute to an improved crash simulation quality, especially regarding fracture prediction. Although not explicitly available, thickness can be inferred from the original CAD geometric model through geometric calculations. This paper proposes and compares two thickness estimation algorithms based on ray tracing and nearest neighbour 3D range searches. A systematic quantitative analysis of the accuracy of both algorithms is presented, as well as a thorough identification of particular geometric arrangements under which their accuracy can be compared. These results enable the identification of each technique's weaknesses and hint towards a new, integrated, approach to the problem that linearly combines the estimates produced by each algorithm.