Statistical process optimisation and probabilistic performance assessment of resin infused carbon-epoxy composite laminates

Abstract

The Resin Infusion under Flexible Tooling (RIFT) has always been assumed to be a cost- effective alternative for manufacturing composite structures. However, there appears to be limited consensus on defining an optimised RIFT manufacturing process configuration for manufacturing a given structure. This had led to many costly trial and error approaches to the problem. In addition, the exact cost-performance advantages of an infused structure are also not widely known. The objective of this research is to assess and present an optimised process configuration for manufacturing a flat RIFT laminate using woven carbon fabrics and a liquid epoxy resin. A statistical process optimisation technique, namely the Design-Of-Experiments method was used for this assessment. It was found that the ideal process configuration for low cycle times does not necessarily result in a laminate having ideal mechanical properties. An optimised process configuration for RIFT of flat laminates was obtained and presented. However, the mechanical properties of the optimised RIFT laminate are still slightly lower than the equivalent Resin Transfer Moulded (RTM) laminate. The second objective of this research is to quantify the cost-performance advantages of manufacturing a flat laminate with RIFT compared with RTM. In this respect, RTM was found to be a better choice for manufacturing better quality laminates. RIFT on the other hand becomes exceptionally more economical than RTM especially for low volume manufacturing, for example prototyping or composite tool manufacture. There appears to be some scope for further automating both processes and hence reduce its labour dependency, which was found to be the dominant cost element in both processes. The third objective of this research is to investigate the applicability and shortcomings of currently available analytical material models for predicting the compressive strength of RIFT and RTM laminates. The fibre kinking model and the shear instability method, developed for unidirectional prepreg were found to be highly accurate (to within 5%) for predicting the compressive strengths of the infused woven laminates. In addition to the three main objectives set at the beginning of the project, the reliability of both the RTM and RIFT laminates were also computed using the Monte-Carlo Simulation Method. The RTM laminates were found to be more reliable than the RIFT laminates and further uptake of the probabilistic strength analysis into industry is recommended.