Méthodes numériques pour la prédiction des déformations et l'anticipation des retraits thermiques en fabrication additive

Swiss partners

• HES-SO, HEIG-VD: Eric Boillat (main applicant), Siddartha Berns

Partners in the MENA region

• Ecole Nationale d’Ingénieurs de Sfax, Université de Sfax, Tunisie: Sana Koubaa (main applicant), Omara Ayadi, Hamza Rahali, Slim Abbes, Slim Abdelkafi

Presentation of the project

This project aimed to improve the perfomances of a 3D printing process: laser powder bed fusion (or L-PBF). With a L-PBF process, the piece is manufactured layer by layer. At every stage, a bed of metal powder is deposited and then selectively melted with a laser. This process is different from other traditional production technologies (machining, foundry) because it allows to manufacture more complex pieces (for example, a rocket combustion chamber) without feasibility constraints.

In selective laser melting, the part is manufactured whilst hot and then cools down to reach operating conditions. However, as it cools, the material contracts and changes shape slightly. If the part is manufactured according to the plans, the result will therefore be non-compliant: it is therefore necessary to print a part that is slightly distorted. Currently, the distorted shape is generally derived from the desired geometry through successive trials, which are costly in terms of both time and materials. The challenge addressed in this project was to develop a strategy for determining the shape to be manufactured based solely on calculations.

The project was to be carried out in three phases.

1. The first phase involved learning how to use a thermo-mechanical simulation programme (Stratus/Warpus) to determine the shape of the manufactured part based on the printed model. The originality of this work lies in the fact that the L-PBF process produces the part layer by layer. It was therefore first necessary to understand how deformations accumulate from one stage to the next.
2. In the second phase, the aim was to use the simulation protocol to test the geometry to be printed and modify it accordingly until a part conforming to the customer’s requirements was obtained at the machine’s output.
3. The final step was to automate the numerical iteration procedure described in point ii) and move directly from the drawings of the desired part to the geometry to be printed without the operator having to intervene.