Development of a digital twin for real-time 3D temperature field monitoring: Application to the Jominy end-quench test

This paper presents the development of a digital twin designed for “real-time” monitoring of the three-dimensional temperature field during the Jominy end-quench test. This test, commonly used to assess the hardenability of steels, generates a pronounced axial thermal gradient, making it an ideal case study for validating a digital twin approach. The core of the methodology relies on the construction of a surrogate model based on Proper Orthogonal Decomposition (POD), derived from Finite Element simulations (FEM) performed with Forge NxT®. This reduced-order model enables the reconstruction of the full temperature field from a limited number of surface measurements acquired via an infrared camera. An inverse formulation, based on least-squares optimization, is used to estimate the modal coefficients required for reconstructing the thermal field. The approach is validated both numerically and experimentally. Temperature deviations between simulated and reconstructed fields generally remain below $$\:\pm\:20\:^\circ\:C$$ , with larger localized errors near the air/water transition zone. A real-time visualization interface, developed in UNITY 3D®, allows the operator to monitor thermal evolution, including within the core of the specimen. This work demonstrates the feasibility of “real-time” thermal monitoring paving the way for better monitoring of thermomechanical paths during hot forging, particularly for superalloys.