This paper presents the development of a digital twin for an extruder, leveragingDigital twinExtrusionMachine learning advanced modeling and computationalEfficiencyComputational materials techniques to enhance the efficiencyEfficiency and reliability of extrusionDigital twinExtrusion processes. The digital twin integrates real-time data acquisition, machine learningMachine learning (ML) algorithms, and physics-based models to create a comprehensive virtual representation of the extruder. The paper focuses on the simulationSimulation models developed to replicate the extrusion processExtrusion process of thermal sheets from glass fibre reinforced polymer material along with the data generation to make surrogate models for the digital twin. A neural networkNeural network (NN)-based model was developed to simulate the extrusion process within a digital twin framework, utilizing real-time sensor data to predict defects in thermal sheets. Additionally, we investigate the potential benefits and challenges of deploying digital twins in industrial settings, and we explore the possibility of optimizing energy usage for such energy-intensive processes.