Prediction of process parameters in selective laser melting

Abstract

Selective laser melting (SLM), also called laser-based powder bed fusion (L-PBF), is an additive manufacturing technology that consolidates powder material in layers into three-dimensional parts by the local application of high-intensity laser radiation. This technology has evolved from a prototyping to a production technology within just two decades. Major applications in the aerospace, energy, tooling, and medical sector have now been explored and have demonstrated both the great economic potential and freedom of design. For now, additive manufacturing allows a profitable production of highly complex and low to medium volume parts. Most of the research on this process is focused on material and application development as well as process simulation. However, there is a distinct gap between the high-detail numerical simulation work and the trial-and-error based application or process development. In particular, the link between major processing parameters, powder properties, and resulting part density is missing, which would allow a significant decrease of trials during process development. Hence, the focus of the present work is to establish a foundation for the prediction of resulting part density in relation to main process parameters, material, and powder properties. As a starting point, a thorough investigation of the influence of the primary processing parameters (laser power, focus diameter, scanning speed) and secondary parameters (e.g. hatch distance, layer thickness, scan vector length) is conducted. Furthermore, a correlation between deviation of weld pool dimensions and particle size distribution of powder as well as spatter particles is established. It is shown that both are not only a practical limit for the recoating process, but also for the resolution and stability of the process. Additionally, it was observed that the size of spatter particles can be approximated by the surface tension of the used material within the investigated process parameters. As a result, three process parameter zones are established whose main influencing factors are the basis for the process model development. The derived models are able to estimate the resulting part density based on the processing parameters, material, and powder properties. A matrix-based, a semiempirical, and an empirical model for the estimation of relative density are introduced; all three share the same approach but allow different levels of complexity in terms of their inputs and outputs. The method is composed of analytical weld pool width, depth and cross-section approximation, estimation of the number of melts, and molten powder fraction, from which the relative density can be calculated. Estimation of the analytical weld pool size is supported by a model for calculation of the local preheat temperature. Approximation of the resulting relative density is achieved by combining the number of melts with the molten powder fraction. Depending on the model and processing zone, these factors are weighted against each other. The proposed models are verified with selected nickel, iron, aluminum, titanium, and copper-based materials. Accuracy of the proposed models is mainly driven by the precision of the estimation of the weld pool dimensions, but an overall good fit of experimental and calculated density is achieved. Additionally, factors such as humidity as well as particle shape for very irregular powders are discussed. Moreover, the impact of shielding gas crossflow on part density and mechanical properties is demonstrated for Inconel 625. In conclusion, the present work provides a basis for increased process understanding and the link between process parameters, powder properties, and resulting part density, thus enabling prediction of process results.