Optimisation of the stamp forming process for thermoplastic composites

  • Simon L Wilkinson

Student thesis: Doctoral ThesisEngineering Doctorate (EngD)

Abstract

This thesis aims to reduce the costs associated with optimising a part, process, and tool design in order to avoid defects when using the thermoplastic stamp forming process. This was done by investigating the defects that can occur when using the process, the mechanisms by which they occur, and how they can be avoided. This knowledge can then be used to avoid defects in other designs, hence reducing design optimisation costs. This work focuses on an omega geometry that was stamp formed from a carbon fibre PEEK laminate in a series of manufacturing trials. Wrinkling, corner thinning, edge squeeze out, and fibre breakage defects were observed in the trials. Hypothetical mechanisms that caused these defects were proposed, and design feature modifications that prevented each of them were identified. However, the modification that prevented wrinkling caused fibre breakage, so an optimum design that prevented both wrinkling and fibre breakage was not tested. However, this optimum design was proposed based on the hypothetical defect mechanisms.

The use of process simulation was also investigated as this has the potential to reduce design optimisation costs by predicting whether defects will occur in a given design. A forming simulation was conducted for three of the designs tested in the trials, and for the proposed optimum design. A verification and validation process was conducted in order to establish and improve the credibility of the simulations. It was initially found that there was uncertainty in the defect predictions due to the use of fibre stress to indicate defects. However, this uncertainty was reduced by using basic methods that accounted for the compaction and temperature reduction of the material. The fact that the defects were predicted with reasonably low error and uncertainty suggested that the models used in the simulations could be used to predict defects in similar designs.
Date of Award28 Sept 2021
Original languageEnglish
Awarding Institution
  • University of Bristol
SupervisorDmitry Ivanov (Supervisor) & Ivana K Partridge (Supervisor)

Cite this

'