Modelling and optimising micro-nozzle resin injection repair of impacted composites using CFD

Abstract

Resin injection repair is identified to have a gap of knowledge and rigour in the modelling and execution of the process. We outline the strategy of our proposed predictive modelling strategy of ‘reconstruction-simulation-injection’ to simulate real cases to improve repair outcomes. We model the damage zone using Darcy’s law and determine permeability using two methods applied on the Kozeny-Carman equation. We then discuss how we evaluate porosity and detail two proposed methods on reconstructing the porosity field. We verify the model through simulation, and demonstrate verification using a novel comprehensive 2D porosity liquid-ideal gas phase flow model after deriving the analytical solution, which is a contribution of our work. Next, we apply the now-established model to reconstruct real damage cases using the two methods and compare them. We also calibrate the permeability parameter for the model by comparison to a simulation accuracy index, and also calibrate an ultrasonic scanning parameter to minimise reconstruction artefacts as well as the sensitivity of the reconstructed geometry characteristics to scan parameter variations. Then, we validate the model by simulating real repair cases and comparing them to the experimental outcomes, achieving simulation accuracy indices of about 70% or more. We demonstrate the application of the resin injection model by applying resin injection in a proof-of-concept simulation and use it for a case study, and examine the importance of hole configuration, vacuum usage as well as resin flow behaviour between inlet and outlet holes in the context of a given damage area geometry. It is important to maximise the total length of resin flow paths available, through carefully placing inlet and outlet holes, to allow resin to infiltrate the damage zone as much as possible. Vacuum increases the minimum achievable filling, and it is still invariably better to use vacuum with an optimal hole placement, instead of one or the other. In a second case study, we improve the predicted outcome by the model after intentionally changing the hole configuration to maximise resin infiltration, demonstrating that filling can be improved by placing holes intelligently (e.g. by using gathered information on the damage area, together with knowledge of how resin would flow). Using this, we conduct an optimisation study of the resin injection model by first setting up the optimisation strategy and carefully determining the methodology. The optimisation procedure is verified by using one and two degree-of-freedom optimisation cases, with known optima. Then, the optimisation strategy is applied to reconstructed repair cases to demonstrate and assess the efficacy of the optimisation procedure, with average reductions in unfilled volumes of approximately 28% compared to initial configurations.