Integrating 3D printing, simulations and surrogate modelling: A comprehensive study on additive manufacturing focusing on a metal twin-cantilever benchmark

Abstract

Additive Manufacturing by powder bed fusion of metals using a laser beam (PBF-LB/M) is constantly growing as an advanced technology to produce metal components. It offers greater design freedom compared to conventional processes and allows the production of complex, lighter geometries with numerous applications in a variety of industries. However, the time and cost required to achieve production readiness present significant challenges to the widespread adoption of new parts development. Success in builds is not reliable until tested, with common issues including distortion, and warpage. The expensive costs of physical iteration to optimize parameters calls for digital simulation to mitigate build failures. This paper presents the successful development of a surrogate model for predicting distortion in a PBF-LB/M metal part. The methodology is grounded on a design of experiments, additive manufacturing tests, finite element modelling playing a critical role, alongside reduced order methods to achieve a surrogate model for improving the additive manufacturing process. The reduced order method for creating the surrogate model is based on tensor decomposition and designed for easy integration into a digital twin, while preserving the underlying physics by retaining the effects of input variables on the final output. The validity of the proposed approach is demonstrated through a benchmark example involving the manufacturing of a metal twin-cantilever part using different laser power, scan speed, and preheating conditions. The twin-cantilever surrogate model developed embeds physics-based simulations and facilitates efficient estimation of deflections. It offers accurate results useful during process setting calibration and improves understanding of how the process parameters affect the final built part.