A digital twin based decision support system for dynamic vehicle routing problems

The research on transportation problems is strongly motivated by real problems. One important industry is the courier, express, and parcel (CEP) market. Recent developments are the growth of the CEP market, the increasing expectations in customer services, and the new technological advances like tracking information, data storage and mobile communication. Modern decision support systems (DSS) are therefore expected to provide real-time decision-making and require fast reactions to changes in the system. Although literature of dynamic vehicle routing problems (DVRPs) exists, further challenges have to be considered when dealing with a real-time DSS. Thus, besides the development of sophisticated optimization approaches, it is essential to provide a concept representing the real world and its dynamic behaviors. In this work, we outline a digital twin based DSS with simulation capabilities for DVRPs. We present various aspects which must be considered in a real-world DSS, in order to provide high-quality solutions within short response times. We show the implementation of a background optimization task and discuss a solution for synchronizing the current system state with the planned routes. Then, we demonstrate the advantages of integrating a simulation model into the DSS. Most importantly, this allows extensive analysis of the optimization algorithm performance in advance and facilitates systematic evaluation and algorithm tuning under simulated real-world conditions. Additionally, we show how the simulation can be used to support a human decision maker during day-to-day operation. Finally, the connection to the digital twin and its advantages are demonstrated by an example application.