DIGITAL TWINS IN METAL ADDITIVE MANUFACTURING

Abstract

The review focuses on digital twins (DT) in metal additive manufacturing, in particular laser powder bed fusion (LPBF; also powder bed fusion, PBF) and directed energy deposition (DED), including wire arc additive manufacturing (WAAM). The paper examines the role of digital twins as integrated computational and data-driven representations of physical manufacturing systems operating under layer-by-layer fabrication conditions. Digital-twin architectures are discussed at the “machine–process–part” levels together with in-situ monitoring data sources, such as optical and infrared (IR) cameras, photodiodes, acoustic, and electrical signals, which are used to capture process behavior and part formation characteristics during printing. The functional interaction between these architectural levels and their contribution to process understanding and quality assessment are considered. The paper systematizes physics-based models of heat transfer, melt-pool hydrodynamics, residual stress and distortion formation, and microstructure evolution, taking into account the multiphysical nature of metal additive manufacturing processes. Hybrid “physics+data” approaches and surrogate models that enable near real-time inference are summarized, emphasizing their role in reducing computational cost while maintaining consistency with physically grounded simulations and experimental observations. Typical DT tasks in additive manufacturing are highlighted: defect prediction (porosity, lack-of-fusion, cracks), process-parameter control, virtual quality qualification, and linking process parameters to microstructure and part properties. Based on the analysis of the literature, a conceptual algorithm for defect prediction in a near real-time mode is proposed, incorporating a feedback loop for updating the digital twin and (optionally) adjusting process parameters to maintain process stability and quality consistency.