Stochastic Resin Injection Simulation of the High-Pressure Resin Transfer Molding for an Automobile Floor Using Adapted Polynomial Chaos Expansions

Abstract

Carbon fiber composite materials are being used at an increasing rate in the automotive industry due to their low densities, superior mechanical properties and great design flexibilities. The recently developed High Pressure Resin Transfer Molding (HP-RTM) process enables high resin injection rates and thereby shortens composite molding cycle times to meet the demanding performance and volume production requirements of automotive components. However, variability is endemic in composite materials and their processing. The development of an accurate injection simulation model of the HP-RTM process is critical for robust process design. In this paper, uncertainty quantification (UQ) for the simulation of the resin injection process of a carbon fiber composite is developed to predict the resin flow and thus the outcome of the composite molding. The uncertainty interplay is modeled using polynomial chaos expansions (PCE). Several material and molding process parameters have been considered for stochastic analysis. This multi-variable stochastic manufacturing problem is challenging to solve computationally, and a previously developed basis adaptation scheme has been used to reduce the computational effort. The UQ development has been implemented on the GM high performance computing system, and the coupled UQ toolbox PAM-COMPOSITETM injection simulations have been conducted for an automobile underbody floor. The numerical results show an excellent convergence of the PCE methodology for the chosen injection problem. A hypothesis of predicting the occurrence of potential dry spots indirectly, by monitoring the information about the resin arrival time, the value of pressure increase, and the value of saturation pressure at a location in the proximity of the flow front closure has been proposed and validated.