Digital Twin Development and Long-Term Simulation of a Cyber-Physical Production System

Abstract

This thesis presents the development and verification of a digital model for the WS Conveyor system located at the SII-Lab. The conveyor system, used for drone assembly, consists of six docking stations arranged along an oval belt where pallets carrying products circulate continuously between workstations. The digital model was implemented in Python as a discrete-time simulation that replicates the control logic specified in the original CODESYS PLC documentation, with every parameter traceable to a specific page in the reference document. The simulation models pallet movement along a 9.4-meter belt at configurable speeds, sensor-based detection at each docking station, and timer-controlled autopass behavior. All physical measurements and operational details were confirmed through on-site visits and a semi-structured interview with the system architect. Verification was carried out through 95 automated tests using the pytest framework, achieving 100% code coverage across all simulation modules. Face validity was established through a structured review with the system architect following the procedure described by Sargent (2013). A 26-scenario parameter-sensitivity analysis revealed that station dwell time is the active-period bottleneck governing system throughput: reducing the auto-pass timer from 10 to 2 seconds nearly doubled throughput, while varying belt speed from 40% to 100% had no measurable effect. Worker processing speed was found to be the dominant factor in shift production: an all-experienced workforce (2–2.5 min per pallet) produced 58% more output than an all-beginner workforce (3–5 min per pallet). The pallet-count relationship saturated beyond 8 pallets when workers became the binding constraint. The system is classified as a digital model in the Kritzinger et al. (2018) sense, with manual data transfer from the physical system. The thesis discusses the path toward a fully connected digital twin using the ISA-95 automation-pyramid framework, and identifies camera-based pallet tracking, process mining of event logs, and predictive maintenance as future extensions.