Inhalt des Dokuments
Bachelor's / Master's thesis
For computational efficiency reasons, MATSim scenarios are often run with sampled populations, i.e. a randomly chosen subset of the whole population of the study region.
It is clear that network capacities need to be adjusted accordingly, such that congestion patterns remain the same also in case a sampled population is used.
While it is obvious that the flow capacity (i.e. the amount of vehicles a link can handle in a given time interval) should be reduced by exactly the same scale as the population, the choice of other network parameters — although reasonable "rules of thumb” exist — remain to some extent as open research questions.
Does a storage capacity (i.e. the number of cars that fit on a link spatially) that is reduced by the same scale as the population lead to overly strong backlog? Should the storage capacity thus be reduced sub-linearly?
What is the effect of the stuckTime (a parameter that describes after what time a car that does not have space on the next link of its route is still moved to that link, inspired by “squeezing in” that can be observed in reality)? How does a stuckTime interact with a reduced flow capacity? Should this parameter be adjusted as well in scenarios with a downscaled population?
It might make sense to include further parameters and effects into the evaluation.
Java Programming skills are required. Basic MATSim skills are advantageous.
The thesis may be written in English or German.
Contact: Theresa Ziemke, M.Sc.
Please direct all requests to firstname.lastname@example.org