Ahmad Mahmoodjanlou

Doctor of Philosophy, (Statistics)
Study Completed: 2020
College of Sciences


Thesis Title
Modelling and Inference for Dynamic Traffic Networks

Day-to-day dynamic assignment models play a critical role in transport management, modelling and planning. Mr Mahmoodjanlou examined the differences between deterministic and stochastic day-to-day dynamic assignment models to determine when deterministic models can be a good approximation for the average of day-to-day stochastic models. The results for deterministic models are fixed however the results of stochastic models can vary so the mean of these results is compared with the deterministic model outcomes. This research is important as deterministic models typically produce much faster solutions for large-scale networks when compared with stochastic models. Stochastic models are more realistic but working with deterministic models is easier. Mr Mahmoodjanlou found that for a system with unique equilibria, these models are comparable however for multi-equilibria systems, the situation is more complicated.
A measure called the coefficient of reactivity can summarise the reaction of a traffic system to a disruption, such as an accident, and is useful in predicting the accuracy of approximation models. Mr Mahmoodjanlou found the fine detail of how well the approximations work alongside the value of the coefficient of reactivity depends to a modest degree on properties of the network such as its size and number of routes and so on. Mr Mahmoodjanlou refined the definition of the coefficient of reactivity to take better account of variation in historical flows. Extension of the definition of the coefficient of reactivity to allow for assessment of the impact of longer disruptions to the system is another original contribution of this thesis.

Professor Martin Hazelton
Associate Professor Jonathan Marshall
Dr Katharina Parry