Review on modelling of thermoset and thermoplastic composites: constitutive equations, processing modelling and multiphysics modelling

Abstract Fibre-reinforced composites have emerged as pivotal materials in advanced structural applications across aerospace, automotive, marine, and renewable energy sectors, due to their superior specific strength, stiffness, toughness, recyclability, and rapid manufacturing potential. Composites manufacturing involves complex multiscale, multiphysics phenomena, including melting, crystallisation, flow, fibre orientation, and residual stress generation that significantly influence the final component quality and performance. Process modelling has become an indispensable tool to address these challenges, enabling the prediction of manufacturing-induced defects, residual stresses and dimensional distortion, and reduction of reliance on physical trials. This review examines current process modelling approaches for thermoset and thermoplastic fibrereinforced composites, with emphasis on the constitutive equations, process simulation methodologies and multiscale modelling frameworks used to predict process-induced phenomena. Particular attention is given to predictive frameworks for temperature evolution, crystallinity development, fibre orientation, residual stress generation, and process-induced distortion. Modelling approaches ranging from conventional finite element methods and molecular dynamics to meshless methods and machine-learning-based techniques are critically reviewed and compared with respect to their capabilities, limitations and application domains. Emphasis is given to the integration of process simulation with experimental monitoring and to the transfer of process-induced material states and defects to structural-level performance models. Key challenges that have been identified include uncertainty quantification, computational efficiency, model transferability, and the limited integration of physics-based and data-driven approaches. Future developments are expected to focus on validated multiscale process-structure-performance frameworks, uncertainty quantification, hybrid physics-informed data-driven modelling, and digital-twin-assisted process monitoring to enhance predictive reliability and accelerate the industrial adoption of composite process modelling and simulation.