Learning by Doing: Experience with a Practical Course on Data-Driven Simulation and Digital Twins

Digital Twin is a virtual replica of a physical asset, system, or process, updated in real-time with data generated from its real-world counterpart. In manufacturing, Digital Twins enable continuous monitoring and what-if simulation, supporting productivity gains and cost reduction. Despite their growing importance, the practical integration of Digital Twin development into engineering curricula remains limited. In this paper, we report on our hands-on course that addresses this gap by teaching the practical foundations of Digital Twins. In the course, students design production lines using LEGO® SPIKE™ kits, equip them with sensors, and implement data pipelines to collect streaming event data. The resulting event logs are utilized to extract Stochastic Petri Net (SPN) models through process mining for workflow discovery, statistical fitting of timing distributions, and, where appropriate, machine learning to capture complex behaviors. The resulting SPN models are simulated and subsequently validated against predefined Key Performance Indicators (KPIs) from the physical system, enabling (near) real-time updates of the Digital Twin models aligned to their physical counterparts. ... mehrWe present the course workflow, approach, and outcomes, offering a template for integrating data-driven simulation and Digital Twin concepts into engineering curricula.