Abstract
<title>Abstract</title> <p>Stevedoring in Roll-on Roll-off (RoRo) terminals is a highly dynamic process, driven by heterogeneous cargo and heavy reliance on human labor. Uncertain processing times make it hard to predict a vessel’s departure time, which is essential for managing shore power and timing of engine preheating to minimize at-berth emissions. While a Digital Twin (DT) can provide real-time decision support, many operators are constrained by the high cost of sensors and complex computing. This creates a critical trade-off between the DT fidelity required for good decisions and the costs of developing and maintaining the system. To explore this, this paper presents a “pragmatic” DT architecture based on discrete-event simulation and rule-based heuristics. Using real-world data from the Port of Kiel, this study systematically reduces the DT’s validity (accuracy) and granularity (data availability) to identify the minimum required fidelity needed for effective decision-making. The results show that the DT is remarkably robust. Decision quality remains high even when data is updated via snapshots rather than in real time, and the system effectively recovers from structural disruptions, such as broken tugs. While the model is sensitive to systematic errors in processing time assumptions, it consistently outperforms static planning methods in predictive accuracy. For practitioners, these findings suggest that the barrier to adopting DTs is lower than expected: a simple, rule-based system provides significant value without massive infrastructure investment. Future research should focus on adaptive DTs that self-calibrate parameters and explore the impact of the interaction between the DT and human behavior.</p>