A digital twin is a virtual representation that accurately replicates its physical counterpart, fostering bi-directional real-time data exchange throughout the entire process lifecycle. This study explores the integration of digital twins into additive manufacturing processes such as laser direct energy deposition of metal wire (L/DED-W). Here, digital twins help to control the residual stress design in built parts [1]. Employing faster-than-real-time and highly accurate surrogate models enables the prediction of altered structural properties, facilitating on-the-fly re-optimizing of the ongoing manufacturing process to achieve desired structural outcomes. Consequently, this approach contributes significantly to realizing the first-time-right paradigm in additive manufacturing. The foundation of successful digital twin derivation lies in the physics-based modeling and prediction of the additive manufacturing process. Predicting final structural properties necessitates mapping input parameters to potentially non-linear part properties. However, a challenge arises from the need to provide faster-than-real-time replications of these mappings through simulations, particularly as the complexity and computational cost of multi-physical simulation models increase. This study addresses the challenge above by investigating the use of a physics-aware machine learning reduced order modeling methodology for thermo-mechanical process simulation. Particularly, stable port-Hamiltonian neural networks [2] are investigated and compared to standard approaches such as neural ordinary differential equations and their derivatives for accurate, robust, and reliable identification of nonlinear dynamic systems. Computational efficiency is demonstrated by characteristic solution times of one-half of a second, imposing negligible processing costs on a single-core processor. References [1] M. Kannapinn, F. J. Roth, and O. Weeger, Digital twin inference from multi-physical simulation data of DED additive manufacturing processes with neural ODEs, 2024. arXiv: 2412.03295. [2] F. J. Roth, D. K. Klein, M. Kannapinn, J. Peters, and O. Weeger, Stable port-Hamiltonian neural networks, 2025. arXiv:2502.02480.