The use of anisotropic material models in the structural simulation of fiber-reinforced plastics is state of the art today. The errors regarding stiffness and strength are too large if the direction-dependent material behavior due to the fiber orientations is ignored.
The creation of anisotropic material models for the structural simulation of plastic components requires an iterative calibration process in which results from specimen production (injection molding) and material testing (tensile tests) are compared with corresponding simulation results (filling and structural simulation).
This procedure requires considerable know-how from the user and generates significant costs. The aim of the current project is to replace the currently required simulation programs with suitable AI models and thus to provide the user with a fully encapsulated and automated tool for creating and validating anisotropic material models.
Using simulation results from filling and structural simulations for different materials, two AI models will be developed and trained to predict both the local fiber orientations in injection molded specimens and the mechanical behavior of these specimens in tensile tests. The prediction quality of the AI models will be validated using practical measured data and additional simulations. The substitution of numerical FEM solvers by AI models is possible here because the geometry and boundary conditions of the specimens considered vary only slightly.
The project results are to be incorporated into commercial software (MatScape) in the form of an additional module which will considerably simplify the use of holistic simulation or even make it possible in the first place and thus contribute to more effective utilization of the material potential.