A Part-Scale Process Simulation App for Laser Powder Bed Fusion
The laser powder bed fusion (LPBF) process involves a highly localized laser heating, resulting in a large thermal gradient that induces residual thermal stress and deformation in the finished build. The excessive residual deformation in the direction normal to in the process plane could lead to recoater jam and cause the process termination. This issue is usually resolved by designing an optimal support structure using the rule-of-thumb and trial-and-error routine. In this effort, a COMSOL® application, ITRI AMSIM, was developed to predict the residual stress and deformation of the part manufactured by LPBF. It provides users a virtual process prediction to cut the time and cost from the trial-and-error routine. The key feature in the app is based on the inherent strain method which was first established for fast prediction of residual stress and deformation in welding problem and recently was used to solve metal additive manufacturing (AM) problem. In short, the residual stress and deformation in an AM part can be estimated by a mechanical analysis, instead of the complicated thermo-mechanical analysis. By doing so, an experimental calibration has to be conducted first to extract the correct inherent strain vector that changes with different scanning strategies and materials as well as the laser parameters (laser power, beam size, scanning speed, hatch size, etc.). Five different materials have been calibrated with our ITRI AM and are ready for app users to virtually preview the outcome of their AM process. For the users using a different AM system, the calibration experiment is required in order to obtain accurate prediction for their AM process.
In this simulation application, a non-linear solid mechanics module is employed in the AM process simulation with the layer activation feature, and an optimization module is utilized in the experimental calibration for different materials. During the app designing stage, the method record feature is quite helpful as it can significantly shorten the time evolving to application from the simulation model. Furthermore, COMSOL Compiler™ enables us to distribute the standalone app to users and customers in a short time.
In this paper, an injector of 17 cm height manufactured by LPBF for a hybrid rocket engine is simulated to provide the information needed to design appropriate support structure that reduces the warpage of the part to avoid the process termination as the recoater is jammed. The manufacturing time of the injector is about 4 days, while the simulation time is within an hour. A user can virtually print his part, improve the support structure design, and diminish the unnoticed build risk. For this case of 3DP injector, the present approach can reduce the production time to one fourth and cut the cost to one sixth, compared to the trial-and-error routine without simulation.
