On the Application of Genetic Algorithms for Optimization of RTM Process Parameters

Abstract

Abstract : Resin infusion processes are finding increasing applications in the manufacture of composite parts that have geometric and material complexities. In such cases, the placement of gates and vents is nonintuitive and may require expensive repetitive experimentation. Finite element-based resin-flow simulation codes have been successfully used for modeling and analysis of the mold-filling process. Such filling simulations, when coupled with a search algorithm, can also prove useful for optimal design of the filling process. Genetic algorithms (GAs) mimic natural selection and can efficiently evolve near-global optimal solutions from a large number of alternative solutions. In this paper, GAs are used to optimize gate and vent locations for the resin-transfer molding (RTM) process in order to minimize fill times and dry-spot formation. A process performance index, or cost function, is defined, which incorporates the fill time and dry-spot formation as primary variables. A part having material and geometric complexities was chosen for a case study. GA and mold-filling simulations were used interactively to search for optimal gate and vent locations to locate near optimal solutions. The GA was able to find good solutions using less than 1% of simulations of the possible permutations of gates and vents. The case study was also repeated in the presence of race tracking channels. Again, the optimal locations were found by the GA using less than 1% of all possible combinations.