Buildup strategies for additive manufacturing by direct metal deposition

Abstract

In recent years, additive manufacturing of metals has turned from a laboratory technology into a competitive and industrially applicable production method. However, the success story of the powder-bed technologies could not be transferred to deposition welding technologies so far, although they show advantages in terms of the higher build rate and larger design space. Since the welding principle and the applied materials are similar, there must be other reasons that inhibit a breakthrough. This thesis deals with direct metal deposition (DMD), a technology where metal powder is blown into a melt pool that is generated by a laser beam. The ability to fabricate complex parts with a high geometric accuracy and a dense microstructure depends not only on the control of the material behaviour, but also on a thorough understanding of the multi-layer buildup process. In here, the interplay of the energy input, powder delivery, tool path, and part geometry is analysed from a systems perspective. Buildup strategies for multi-layer DMD are developed and validated exemplarily with the materials steel type 1.4404 and Inconel(R) 718. These strategies can be divided into three areas: First, the importance of a suitable tool path is revealed by an experimental comparison of conventional tool path approaches. Two specific path strategies are proposed and optimised for five-axis processing. Second, the influence of the part geometry on the process behaviour is considered. An analytical melt pool model points out the general necessity for a process control. The development of a geometry-based adaptation of the laser power aims to achieve a steady-state process and takes the heat accumulation in slim parts into account. This feedforward control is performed by an algorithm that analyses the local part geometry and correlates it with experimentally determined process parameters. The validation proves that a satisfactory geometry and a dense microstructure can be obtained, reaching an average porosity of 0.14% for specimens made from Inconel(R) 718. Third, the mechanism of a self-stabilising DMD process is examined, showing the trade-off between the powder catchment efficiency, the robustness towards disturbances, and the geometric accuracy. A recursive algorithm calculates the layer height based on the local powder catchment efficiency. It predicts the waviness of surfaces due to stability and instability, reaching a high agreement to experiments. The machine systems must be capable of multi-layer DMD and execute the calculated tool path. Therefore, the measured dynamics of the machine axes are considered by an algorithm, optimising the tilt and turn movements of the processing head to ensure a constant scan speed. An existing method to measure the powder stream is refined and enhanced for a 3D representation of the powder distribution. Measurements with two nozzles reveal quality deficiencies of the stream and allow a calculation of the powder catchment efficiency for different standoffs and melt pool sizes, serving as an input for the layer height model. The suitability of a nozzle to generate a self-stabilising process is evaluated with experiments and a model. All strategies are implemented in a fully automated CAM software which considers the machine capabilities and generates a feasible and collision-safe tool path. Deficiencies of the DMD machine are compensated by software measures. The performance of the CAM software is demonstrated by the fabrication of turbine blades. With the optimised "raster/contour" strategy, smooth freeform surfaces with R_a = 30 mu, an intended oversize between 0 and 2 mm, and a minimum trailing edge radius of 2.5 mm are achieved. With the "advanced raster" strategy, the radius can be decreased to 1.7 mm, but the surface roughness increases to R_a = 65 mu. Finally, the implications of the process combination of DMD and milling for the part accuracy are discussed. A case study reveals that distortion due to DMD exceeds common milling inaccuracies by more than factor ten. In order to be able to fabricate internal features in an alternating process of DMD and milling, the concept of provisional protrusions is proposed, protecting finished surfaces from the negative impact of DMD. The fabrication of a multi-step cylinder succeeds with an almost smooth surface transition with a step of 8 mu.