Digital twin–enabled model predictive control in additive manufacturing: critical review, research challenges, and future directions

Abstract

This review provides a comprehensive analysis of Digital Twin–enabled Model Predictive Control (DT–MPC) in Additive Manufacturing (AM), highlighting its potential to enable predictive, adaptive, and sustainable production. Digital twins–constructed using finite element models, data-driven surrogates, and hybrid physics-informed neural networks–offer dynamic, real-time process representation, while MPC provides optimal control under complex, multi-physics constraints. Comparative studies from 2015–2025 reveal significant advances in melt-pool stability, defect mitigation, and energy efficiency, though industrial adoption remains constrained by computational cost, synchronization lag, and limited interpretability. The proposed research roadmap identifies short-term goals in real-time scalability, mid-term objectives in uncertainty management and explainable AI, and long-term priorities in lifecycle-aware, eco-efficient manufacturing. Bridging these challenges requires standardized digital frameworks and interdisciplinary collaboration. DT–MPC emerges as a cornerstone of Industry 5.0—transforming AM into an intelligent, self-adaptive, and sustainable manufacturing paradigm.