Enabling Process Mining In Global Production Networks

In today's globalized economy, global production networks (GPN) play a central role in ensuring competitiveness. The number of interconnections and size of the networks lead to an ever-increasing complexity. With the help of data-driven methods, such as process mining, transparency over material flows in GPNs can be increased, making the complexity manageable. However, the adoption of process mining is associated with challenges. In GPN, different components are produced across various production stages, each with their own characteristics. In the course of the material flow, the components are then assembled into component groups, which in turn represent intermediate products or the finished end product. Consequently, converging material flows are the result. These circumstances make it difficult to build unique case identifiers and a comprehensive process data model. Furthermore, the identification of the right data sources is a challenge, since in GPN often different information systems are available in individual production lines or locations. We present an approach that supports overcoming these challenges and enables process mining in GPNs. For this purpose, the presented approach contains a process data model that takes into account all GPN specifics and requirements of process mining. This includes the handling of different component types and their properties as well as the merging of process instances in converging material flows. Through this, the individual sub-components of a finished product with their respective subprocesses are traceable. Further, we provide a procedure that supports the user in data identification and extraction by describing common data structures in information systems, which are usable for the generation of process mining event logs in GPN. Based on these data structures, we provide standardized input data tables into which the identified raw data can be extracted. Subsequently, the transfer into the presented process data model is automated by a data transformation algorithm, which takes into account the specifics of material flows in GPN. The approach was applied, implemented and validated in the GPN of an automotive manufacturer.