Spatio-temporal model predictive control of workpiece temperature in additive manufacturing

Additive manufacturing of metals has increasingly attracted the attention of industry and science due to its ability to produce complex geometries, enable rapid prototyping, and facilitate the cost-effective production of customized small batches. A key challenge in this context is ensuring consistent component quality. This is attributable to the strong interactions between process parameters and thermodynamic variables. Thermal gradients, as well as solidification and cooling rates, directly influence the microstructure, residual stress state, and crack susceptibility of the workpiece. Conventional strategies using static process windows cannot adequately model this dynamic manufacturing process and require complex calibrations as well as extensive expert knowledge.

This project addresses these issues through the use of modern predictive control algorithms. The research approach follows the following development path: First, the control methods are designed in virtual models and systematically validated. This is followed by implementation and experimental testing on a real production system such as the LMD^2 research facility. Using a physics-based process model and sensor data collected in real time (thermal imaging camera, pyrometer, and coaxial visual RGB camera), spatially and temporally dependent predictions of the workpiece’s thermal temperature are to be calculated. The focus here is on the predictive, model-based adjustment of optimal process parameters during the manufacturing process. This control is intended to ensure consistent quality assurance and to sustainably increase the reproducibility and efficiency of additive metal manufacturing.

Emre Öztürk, M.Sc.

November 2025 - October 2028

University of Kassel