Hybrid delay models for unsaturated two-way stop controlled intersections

A new methodology for modeling delay at undersaturated two-way stop controlled intersections was developed. Total delay was divided into service delay and queue delay. Empirical models were developed to estimate service delay as functions of conflicting traffic volumes, and variance of service delay as a function of conflicting traffic volumes and the average service delay. These models were used as inputs to the Markovian (random) arrivals/generally distributed service times/one server model to estimate the queue delay. The Markovian (random) arrivals/generally distributed service times/one server model provided significantly better queue delay estimates than the Markovian (random) arrivals/Markovian service rate/one server model. The models were built based on 28 h of field data from eight intersections. The total delay model was also tested using data from another location. The proposed models estimate delays that are closer to the field data than the 1994 Highway Capacity Manual delay model. The new models are for unsaturated two-lane two-way approaches with random patterns and a major street speed limit of 48 km/h (30 mph).