Data analysis for process optimization in additive manufacturing
The increase in additive manufacturing production volume in recent years has led to a need for an increased level of productivity, quality control during the build job, as well as the optimization of other crucial steps in each respective technologies process chain. Current industrial additive manufacturing production processes such as selective laser sintering and multi jet fusion face major challenges when facing the part quality and throughput requirements of the automotive industry. This dissertation analyzes the use of machine sensor data such as temperature sensors, thermal images and build job data within the multi jet fusion and SLS process to better understand the relationship between fusing energy, packing density, part location within the build to part quality. Visualization techniques, statistical analysis tools and machine learning were used to communicate the results, draw correlations, and optimize these processes in a production environment. A reduction in porosity as well as an increase in powder cake molecular weight was observed with an increase in fusing power, which also resulted in improved mechanical properties. This correlation was also verified by a detailed analysis of internal part temperatures throughout the build, by means of a thermal imaging camera. Additionally, an in-depth analysis on build job cooling times using external temperature sensors and build job data was done, resulting in the development of a gaussian process regression model which predicts the required cooling time for any respective polymer powder-based build job with a significantly higher accuracy than that of the manufacturer. The implementation of the optimized cooling model at BMW’s AM production facility leads to increased transparency, leaner production, and a higher overall economic viability of AM technologies.