Identifikation und Analyse von Wechselwirkungen in einem soziotechnischen System: Eine agentenbasierte Modellierung und Simulation am Beispiel einer Trafostationfertigung

The increasing complexity in modern production systems presents significant challenges for companies, particularly about the identification and control of internal interdependencies between man, machine and organization. In socio-technical manufacturing environments, dynamic interactions between operational decisions, logistical dependencies and strategic objectives mean that traditional, deterministic modelling approaches quickly reach their limits. Since these models are often only able to depict the non-linear and adaptive processes within a production system to a limited extent, new methodological approaches are required to enable a realistic analysis and optimization. This thesis therefore examines the suitability of agent-based modelling and simulation as a tool for better capturing and controlling such systems in an industrial environment. The use case is a newly established production facility for transformer stations in an industrial company, whose production processes were implemented using agent-based simulation in the AnyLogic software environment. The modelling approach was based on an iterative bottom-up development in which the industrial partner was closely involved in the process at an early stage. This proved to be essential to ensure a realistic representation of the production dynamics and to push forward the progress of the simulation in a targeted manner. The focus of the simulation study was on investigating the influencing factors that contribute to the production performance and stability of the system. Particular attention was paid to the control logic of prefabrication, the availability of materials and the organizational framework conditions in assembly. The results of the simulation show that it is not only the availability of resources that determines production performance in the examined system, but rather the way in which material flows and work processes are coordinated. It became particularly clear that an overarching target that uniformly aligns all levels of the company is not sufficiently anchored. As a result, potential for process optimization remains untapped and decisions are prioritized differently depending on the perspective. Furthermore, the simulation study also illustrates how challenging it is to precisely capture the actual process in a dynamic production environment. These complex interdependencies within production are difficult to grasp, even for those responsible on site. The validation of the simulation proved to be a valuable means of realistically mapping the actual process and creating a reliable data basis. The insights gained illustrate the strength of agent-based modelling in the analysis and optimization of socio-technical manufacturing processes. In addition, the work shows that agent-based simulation is not only applicable for large companies with extensive resources but can also be implemented effectively under limited framework conditions.