The current work presents the results of an investigation focused on the influence of process parameters on the melt-track stability and its consequence to the sample density printed out of NdFeB powder. Commercially available powder of Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 alloy was investigated at the angle of application in selective laser melting of permanent magnets. Using single track printing the stability of the melt pool was investigated under changing process parameters. The influence of changing laser power, scanning speed, and powder layer thickness on density, porosity structure, microstructure, phase composition, and magnetic properties were investigated. The results showed that energy density coupled with powder layer thickness plays a crucial role in melt-track stability. It was possible to manufacture magnets of both high relative density and high magnetic properties. Magnetization tests showed a significant correlation between the shape of the demagnetization curve and the layer height. While small layer heights are beneficial for sufficient magnetic properties, the remaining main parameters tend to affect the magnetic properties less. A quasi-linear correlation between the layer height and the magnetic properties remanence (J_r), coercivity (H_cJ) and maximum energy product ((BH)_max) was found.

In today’s complex supply networks sharing information between buyers and suppliers is critical for sustainable competitive advantage. In particular, for both business partners, cost information is highly relevant in purchasing situations. According to empirical studies in literature, artificial neural networks (ANNs) are expected to have a great potential to reveal cost structures by machine learning (ML). In digitally enabled supply chains this information can contribute to cost reduction and operational excellence and lead to win-win situations in supplier relationship management. Nevertheless, authors do not thoroughly investigate how ANNs may support cost estimation for purchasing decisions. Based on a case study from the automotive industry, we evaluate ANNs regarding their capability to gain cost structure data. In an additional comparative study, we benchmark ANNs for cost estimation in purchasing against other promising ML algorithms. Thereby, we apply the cross-industry standard process model for data mining projects. The findings of the studies show that some ML algorithms outperform ANNs regarding accuracy. The research results give indications for choosing the ML approach that promises the best outcome for cost estimations and cost structure information to support decision-making in buyer–supplier relationships.

Modern industrial production systems are facing the challenge of interconnecting and exploiting the increasing amount of data from smart products, software systems, and IIoT devices. One of the main tasks in this context is the implementation of a data-continuous traceability system, which interconnects the heterogeneous data objects over the entire production flow. While traceability maturity is already high in regulated and process-driven industries, the established traceability solutions cannot be successfully applied to customized and volatile industries. In this research, a graph-based traceability modeling methodology is presented that allows the systematic development of data continuity and integration in manufacturing. Based on the traceability methodology, the graph data model and architecture are specified, which are aligned to the requirements of customized and complex manufacturing systems. The model is implemented and validated based on a Neo4j graph database for the use case of the manufacturing process of automotive electrical systems. This research overcomes the shortcomings of state-of-the-art traceability models by shifting the focus to the relationships between traceability-relevant data objects. The proposed graph-based traceability model is able to capture multi-hierarchical product structures and is less dependent on physical object identification, making it more applicable to customized and complex manufacturing.