Physics-driven digital twin based architecture of wire arc additive manufacturing enabled by fast surrogate calculations

This study presents a physics-driven Digital Twin (DT) architecture for Wire Arc Additive Manufacturing (WAAM) that integrates high-fidelity thermo-mechanical simulations with data-driven surrogate models to support monitoring, prediction, and closed-loop process control capabilities. The proposed multilayer framework defines interactions among physical assets, computational models, intelligent modules, and visualization services through standardized data exchange and scalable workflows. Core components of the DT, including finite-element thermal modelling, neural-network-based surrogate models, middleware integration, and communication mechanisms, are developed and quantitatively validated within the proposed architecture. Validation results obtained using a Ti-6Al-4V WAAM demonstrator show that the surrogate models reproduce the responses of the underlying finite-element models with an accuracy exceeding 95%, while reducing computational time by approximately 85–90%. The results demonstrate the feasibility of integrating physics-based and data-driven models within a unified WAAM-DT framework and establish a foundation for perspective implementation of adaptive closed-loop manufacturing systems.