Straddle carrier routing optimization at container terminals utilizing telemetry-based prediction

Abstract

Efficient vehicle routing and scheduling for horizontal transport means on container terminals can reduce lead times and travel distances, resulting in fuel savings and productivity gains. We apply machine learning to emulate operational processes, bridging the gap between theoretical optimization models and real-world practices at container terminals. Using telemetry data and predictions based on machine learning, we analyze travel times. This information is then incorporated into an optimization framework for horizontal transport dispatching. In a case study with straddle carrier operations for a container terminal in Hamburg, Germany, we focus on (meta)heuristics for solving related vehicle routing problems. Using a greedy algorithm, local search, and variable neighborhood descent, we achieve a reduction in travel times of 17% compared to current operations controlled by the terminal operating system.