Abstract

Possible applications of metal additive manufacturing (AM) in the railway industry have been continuously assessed at the Knorr-Bremse (KB) Group in recent years, as metal AM could deliver significant benefits compared to traditional manufacturing processes when it comes to lower volume production, generative design and design simplification. Although the potential is huge, there are still obstacles to be overcome to realize AM applications on an industrial level. As the process is basically a welding process, FEM-based process simulations created using the inherent strain method became state-of-the-art. However, material knowledge for accurate prediction of stresses and distortion, for example, to compensate deformation, predict process failures, or optimize supports, is crucial. Thus, there is the need to consider further material-related effects that occur on preheated processes and real geometries. Metal additive manufacturing has been performed at Knorr-Bremse Rail Systems Budapest since 2018 with an EOS M290 machine. By applying laser powder bed fusion (LPBF), several different parts have been manufactured from the aluminium alloy, AlSi10Mg, since then. Although extensive metal AM experience has been built, printing failures have occasionally been experienced. The goal of avoiding printing failures and enabling first-time-right metal AM process was formulated. To support this goal, an AM simulation validation project was initiated to assess the capabilities of the Amphyon software to predict printing failures and optimize print jobs. An advanced calibration method for the inherent strain approach was developed, including the efficient modelling of the creep effect of AlSi10Mg. Print jobs with printing failures encountered in practice were collected along with high-resolution images created during the printing processes. The validation of simulation on printing failure prediction was performed by comparing simulation results of these print jobs with the high-resolution images, focusing on the primary printing failures of re-coater interference and support failure. The print job optimization capabilities (i.e., orientation assessment, support structure optimization, pre-deformation of geometry) were validated via the actual 3D printing of jobs optimized by simulation.