This talk investigates nodes in a transport network that do not handle transport items but the vehicles circulating to carry passengers or goods between different node work nodes along the network arcs. These nodes are called depots and serve as parking lots or maintenance yards.

Internal depot processes within transport networks face several challenges, like limited vehicle hosting capacity and restricted operation times to name just some predominant examples. The primary need of a depot manager is to coordinate the arriving vehicles with the currently stored vehicles and to ensure that vehicles can leave the depot so early that subsequent vehicle deployments can start as planned. Every depot is a production system comprising several workstations connected by a road or rail infrastructure. Often a depot turns out to be a bottleneck in the transportation processes within a network.

A significant driver for the complexity of planning depot operations arises since vehicles must travel through this production system in a vehicle-specific machine sequence considering vehicle-specific availability and due dates.

We propose a generic mathematical model that aims to connect the vehicle arrivals in front of a depot with vehicle departures from the exit gate by proposing internal vehicle movement processes inside the depot area. The central idea is to set up a time-space network graph where vehicle operations take place. We report initial results from the model application in the context of commuter rail yards confronted with heterogeneous requests to guide vehicles through the depot facilities under time constraints.

In order to manage the demand and control the flow of cargo arriving at terminals, the port sector created a dynamic Truck Appointment System. However, disruptive events can cause a delay or an early arrival of trucks at the port terminal, leading to long waiting times, queues, and the need to reschedule trucks in other time windows when they arrive off the scheduled time. Smart technologies have potential to deal with uncertain scenarios and create a flexible context for the use of TAS. In this context, the main objective of this study is to compare regression and classification Machine Learning algorithms to predict truck arrival times. By comparing the predictions with the original appointment, a flexible Truck Appointment System is designed. Four different ML approaches were evaluated, which have been implemented in Python: Linear Regression, Random Forest, Gradient Boosting Regression, and Decision Tree. Considering the disruptive arrivals, we identified that the classification algorithms performed better than the regression algorithms predicting the ex-act arrival time, but worse than the regression algorithms that predict the time window of truck arrivals.

International rail transport plays an essential role in reducing carbon footprint, with Eurasian rail transport as a successful example. However, multiple interfaces across totally different rail systems increase the complexity of information exchange. The quality of information directly influences the quality of planning. The impact of information exchange and its integration into international rail transport planning activities has not yet been studied extensively. This paper presents an Integrated Planning Method (IPM) for international rail transport focusing on information integration. Advanced mathematical models for Decision Support Systems (DSS) based on information integration are also suggested.