Powder bed fusion (PBF) is a widely adopted additive manufacturing (AM) process category that enables high-resolution fabrication across metals, polymers, ceramics, and composites. However, its inherent process complexity demands robust modeling to ensure quality, reliability, and scalability. This review provides a critical synthesis of advances in physics-based simulations, machine learning, and digital twin frameworks for PBF. We analyze progress across scales—from micro-scale melt pool dynamics and mesoscale track stability to part-scale residual stress predictions—while highlighting the growing role of hybrid physics–data-driven approaches in capturing process–structure–property (PSP) relationships. Special emphasis is given to the integration of real-time sensing, multi-scale modeling, and AI-enhanced optimization, which together form the foundation of emerging PBF digital twins. Key challenges—including computational cost, data scarcity, and model interoperability—are critically examined, alongside opportunities for scalable, interpretable, and industry-ready digital twin platforms. By outlining both the current state-of-the-art and future research priorities, this review positions digital twins as a transformative paradigm for advancing PBF toward reliable, high-quality, and industrially scalable manufacturing.
Łach et al. (Wed,) studied this question.