Optimisation of the VARTM process

Abstract

This study focuses on the development of a multi-objective optimisation methodology for the vacuum assisted resin transfer moulding composite processing route. Simulations of the cure and filling stages of the process have been implemented and the corresponding heat transfer and flow through porous media problems solved by means of finite element analysis. The simulations involved material sub-models to describe thermal properties, cure kinetics and viscosity evolution. A Genetic algorithm which constitutes the foundation for the development of the optimisation has been adapted, implemented and tested in terms of its effectiveness using four benchmark problems. Two methodologies suitable for multi-objective optimisation of the cure and filling stages have been specified and successfully implemented. In the case of the curing stage the optimisation aims at finding a cure profile minimising both process time and temperature overshoot within the part. In the case of the filling stage the thermal profile during filling, gate locations and initial resin temperature are optimised to minimise filling time and final degree of cure at the end of the filling stage. Investigations of the design landscape for both curing and filling stage have indicated the complex nature of the problems under investigation justifying the choice for using a Genetic algorithm. Application of the two methodologies showed that they are highly efficient in identifying appropriate process designs and significant improvements compared to standard conditions are feasible. In the cure process an overshoot temperature reduction up to 75% in the case of thick component can be achieved whilst for a thin part a 60% reduction in process time can be accomplished. In the filling process a 42% filling time reduction and 14% reduction of degree of cure at the end of the filling can be achieved using the optimisation methodology. Stability analysis of the set of solutions for the curing stage has shown that different degrees of robustness are present among the individuals in the Pareto front. The optimisation methodology has also been integrated with an existing cost model that allowed consideration of process cost in the optimisation of the cure stage. The optimisation resulted in process designs that involve 500 € reduction in process cost. An inverse scheme has been developed based on the optimisation methodology aiming at combining simulation and monitoring of the filling stage for the identification of on-line permeability during an infusion. The methodology was tested using artificial data and it was demonstrated that the methodology is able to handle levels of noise from the measurements up to 5 s per sensor without affecting the quality of the outcome.