Adaptive automation and defect control in forming processes of composite materials : a robust parametrisation approach to improve simulation accuracy

Abstract

Variations in the parameters of both materials and processes constrain the precision of composite forming processes. Examples of these varying properites include material like Young's modulus, bending and shear modulus, and process properties like temperature, pressure, and tool velocity. It is important to consider these parametric variations while developing a part through composite forming processes to ensure accuracy and minimise defects. Robust parameterisation and optimisation methods or defect control systems are two approaches to improving production accuracy. This research focuses on applying robust parameteric and optimisation methods to enhance simulation accuracy and using defect control systems to identify and manage induced forming defects. Robust parametric and optimisation methods create design variables minimally affected by variations during forming process. Usually, the behaviour of the process is predicted using complex and time-consuming FE simulation models. If there is much design freedom, it is often only feasible to study the complete parametric design space using expensive computational resources. Efficient methods are employed to determine the robust design of these processes with limited computational resources. One approach involves creating an approximation model of the FE model, such as a surrogate or metamodel, and then improving this approximation model by evaluating the FE model multiple times. In this research, an improved sequential robust parametric and optimised method improves the FE simulation accuracy for the forming process. The effect of these sequential robust parametric approaches is supported by genetic algorithm and regression for optimisation. The second approach to improve production accuracy is using defect control methods. Point cloud scanned measurements are used to inspect the part in real time. A challenge in controlling defects during the composite forming process is that the final geometry can have several defects after the forming has been completed. Therefore, it is only possible to measure the final geometry of the part after performing the scanned measurements. The improvements in the scanned measurements depend on the accuracy and fineness of point cloud. This research examines whether scanned point cloud measurements are used to identify and control defects like bridges, gaps, and wrinkles. It also looks at whether these defects can be identified using a finite element (FE) model from the simulation. Both methods are studied, and separate frameworks are proposed and validated. The convergence method discusses the accuracy of both methods employed in a framework using comparative analysis and feature extractions. In summary, the research objectives are the following:

A specialised robust parametric framework that can simulate and predict forming behaviour and optimise to self-improve with real-life data. Physical testing and point cloud-based digital scanning of parts produced through the forming process for defect detection. Comparison of simulation results and physical testing to develop a robust framework for further analysis and improve its accuracy.