Thermal twinning for induction processing of composites

Abstract

Composites manufacturing faces significant challenges in thermal management, where precise temperature control is critical to part quality and process efficiency. Traditional methods such as ovens and autoclaves rely on conduction and convection, resulting in thermal lag due to low resin thermal conductivity and heat losses to tooling and air. Cure control is further complicated by the highly exothermic nature of thermosetting resins. Electromagnetic induction heating offers rapid, volumetric heating, but is difficult to apply due to the nonlinear relationship between current and resulting temperature distribution. This PhD research presents a real-time temperature control methodology for induction-heated composites, built around a Digital Twin combining heat transfer, cure kinetics and induction heating models. It predicts through-thickness temperature profiles from limited sensor input, assuming relatively uniform in-plane heating without requiring ideal conditions.
The model incorporates key physical properties (thermal conductivity, heat capacity, density), resin behaviour (total heat release) and process parameters (convective coefficients and Joule heating effects). Parameter estimation and model calibration are achieved pre-cure using pulse excitation, Nonlinear Least Squares (NLSQ) and Latin Hypercube (LH) sampling. During cure, a recursive scheme updates evolving properties in real time. These feed into a Model Predictive Control (MPC) framework that dynamically adjusts induction current to minimise cure time and prevent overheating. Validation through simulations and physical experiments on CFRP panels, using both induction and mat heating, confirmed temperature prediction accuracy within 5 ℃ and 1.5 ℃, respectively. The methodology also demonstrated effective process control, simulating optimised composite curing. While focused on induction, the framework extends to other thermal processes, including heating mats and alternative energy sources. Overall, it offers a promising avenue to improve efficiency, reduce energy use and accelerate manufacturing through precise, model-based thermal control.

Date of Award	9 Dec 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Dmitry Ivanov (Supervisor), Janice M Barton (Supervisor) & Jason Zheng Jiang (Supervisor)