Elastically Coupled Numerical Fluid Models for Flows Found in Composites Manufacturing

Abstract

Void formation is a common issue in the manufacturing of carbon fibre composites. In particular, composites produced using out-of-autoclave manufacturing, which offers an emerging low-cost alternative to autoclave manufacturing, can experience high porosity (greater than 2\%) due to decreased consolidation pressure. Numerically simulating void evolution during the curing process will allow manufacturers to adjust process parameters to reduce porosity. Previous attempts at this have not included important features such as representative distributions of non-spherical voids and complex boundary conditions.

This work addresses this by creating two new fully coupled multi-phase models. Each model simulates the behaviour of solid carbon fibre reinforcements, liquid resin, and void gas during the curing process. Both coupled models incorporate the same finite difference structural model developed for this project, which simulates very stiff fibres at low cost using a linearised implicit formulation of tension, bending and contact forces. However, the coupled models use different, existing, fluid models. These are a smoothed particle hydrodynamics (SPH) model and a finite volume mass-pressure model. Gas is represented using a volume-of-fluid approach. The coupled models operate at different scales. The coupled SPH model simulates void movement at a very fine resolution comparable to that achieved by CT scans in two dimensions over short time periods, while the finite volume mass-pressure model performs component scale simulations over the entire curing cycle at a coarser resolution driven by the thickness of pre-preg plys.

Several coupling schemes were implemented and evaluated. For the SPH case, monolithic and partitioned coupling schemes for weakly compressible SPH solvers were developed. Of these, the monolithic showed marginally lower computational cost but less scope for further development. The partitioned solver was further adapted to operate using incompressible SPH, providing faster computation at the cost of worse parallel scaling and fluid behaviour at boundaries. Meanwhile, the mass-pressure model uses a partitioned coupling scheme based on quasi-steady fixed-point iterations. Simple test cases were used to validate the coupled models as well as the structural and fluid models operating in isolation. CPU profiling informed changes which led to a significant speeding up of both codes. Finally, the coupled models performed larger simulations of the curing process. In process-CT scan data was used to create representative initial void distributions.

The coupled SPH model performed well in simulating the behaviour of dry areas and inter-ply voids in small two-dimensional external corner test cases, replicating complex behaviour over short time scales. The coupled mass-pressure model simulated three-dimensional component-scale test cases for entire cure cycles. Flat test case simulations showed good agreement with experimental void evolution. External corner simulations allowed the bending stiffness of pre-preg plies to be accurately evaluated and demonstrated the importance of thickness effects at the sample apex in driving void motion. These simulations represent a novel advance over existing methods and provide useful insight into the manufacturing process.

Date of Award	30 Sept 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Thomas C S Rendall (Supervisor) & James Kratz (Supervisor)