The aim of this document is to provide an assessment of possibilities for simulation approaches, both microscopic and macroscopic, that do not require detailed data on all aspects of the simulated scenario, for example for incompletely specified railway freight traffic.

In a macroscopic simulation, only a choice of aspects is included, e.g. stations and stops, but no detailed track plan between these network nodes. In that case, a stochastic approach is required because failures of different infrastructure elements of the same type on a specific macroscopic network-edge can have different consequences for train travelling times. This approach ignores the causal connection between disturbance, re-routing, possible waiting times and the resulting delay, and simply assigns a probability to different delay times given a type of disturbance.

The simulation tool PRISM uses a macroscopic model. It makes it less important to have a high level of detail in the data. Missing such data is thus less problematic than for microscopic simulation tools such as RailSys. In either case, it is prudent to start by establishing a well-calibrated scenario, before scenarios with missing observations are run.

However, missing data makes the comparison to real data more difficult. Especially if many assumptions are used or the input is generalised to a large amount, the results will not be comparable for enabling an accurate analysis. In that case, other ways of validation have to be chosen.

The conclusions reached in this report are that there are certain situations where the available information is not sufficient for performing a simulation, but that there are ways to overcome these shortcomings.