Microstructure on demand in additive manufacturing through a synergy between control, measurements and simulations – MIFASOL

Obtaining optimal properties in various places of a structure is a major issue in metallic additive manufacturing or repair. The solution is based on an in-depth knowledge of the links between properties and microstructures and their control during the entire process. Moreover the link is present at different time and space scales and controls the solidification process as well as the evolution of the microstructure during the subsequent thermomechanical cycles. The aim of the MIFASOL project is to propose a manufacturing strategy to control jointly geometry and microstructure for direct energy deposition (DED) processes.
However, such strategies come up against three main scientific and technical obstacles. The first is that any control strategy requires predictive simulations of the formation and evolution of the microstructure during the process. The second is due to the real-time control strategies necessary to adjust the process parameters to avoid a drift in thermal kinetics. The third difficulty concerns the definition of the manufacturing strategy and the control of the evolution of process parameters to guarantee geometry and microstructure.
The MIFASOL project therefore proposes: 1) rapid models coupling temperature and microstructure formation / evolution on the scale of the whole process, allowing to establish a manufacturing strategy, 2) in-situ measurements coupled with machine-learning algorithms to correct in real time the manufacturing parameters and 3) precise modeling and control of the kinematics of the material deposition in order to define the manufacturing strategy in the case of complex structures.
The expected results of the project are: 1) an efficient fast calculation tool to simulate heat transfers as a function of all the process parameters as well as the formation and evolution of microstructures, 2) an experimental setup allowing in-situ temperature measurements of a large part during the process as well as a neural network (trained on a large number of simulations) allowing to use this measurement in real time to correct the manufacturing parameters and achieve the desired microstructure and 3) the creation of a digital twin based on the digital additive manufacturing chain, integrating knowledge and models allowing the synthesis of deposit strategies by performing virtual testing of the process or in real time by coupling digital models and in-situ measurements.
The MIFASOL project therefore will clearly work on different complementary analysis paths: measurements and analyzes in real time associated with fast simulations of the process. It is therefore interested in materials and processes, but being resolutely turned towards innovative measurement and control instrumentations, control-command learning techniques by neural networks in order to propose a better integration of additive manufacturing among innovative technologies allowing simultaneous optimization of the material, its microstructure and the manufactured part.
The success of the project therefore rests on the perfect synergy between the project partners and by the recruitment of two doctoral students as part of the project, one responsible for making the link between the fast models and the manufacturing strategies in order to go towards the development of a digital twin and the second responsible for carrying out quantitative in-situ measurements coupled with real-time monitoring by neural network.