Traffic signal optimisation in disrupted networks with re-routing

This paper describes a modified bi-level framework to simulate disruptive events in urban road networks with different levels of disruption severity and duration. This framework combines the Cross Entropy method to optimise traffic signals and the quasi-dynamic user equilibrium assignment model embedded in SATURN software package. This enables simulation of short-term closures with less computational effort and running time than fully dynamic models. We have applied this model to the Cambridge (UK) network and demonstrated how the degradation at one node affects the optimal signal settings at that node and nearby nodes. The computational results show that for different disruption severities, as traffic starts to divert to other routes, the optimal traffic signal settings changes, to minimise the travel time.