Urban Network Safety Assessment under Uncertain Demand

This paper presents a method based on robust optimization for assessment of road network safety in urban areas under limited budgets and different demand scenarios. Future traveldemand in urban areas depends on forecasts for land uses and household demographics, other socio-economic factors, and policies set by decision makers for demand management; thus, it always carries some levels of uncertainty. Demand uncertainty would producefluctuations in traffic flow of network links, which in turn as a measure of exposure, leads to different crash frequencies. This means safety planning would be affected by demand uncertainty. Robust optimization is a novel approach to model optimization problem under unce rtaintyconditions. Expected crash frequencies in each scenario are estimated based on link traffic flows derived from equilibrium assignment of demand matrix. Genetics Algorithm isutilized and calibrated for solving the model problem. The model is tested for arterial road network of an actual urban area and expected crash frequencies and safety priorities are determined for the horizon year of 2016. Results for applying this model in different levels of demand uncertainty are presented.