In the current Highway Capacity Manual (HCM), the capacity and delay of unsignalised junctions are predicted using gap-acceptance theory, which is very sensitive to the size of the junction's critical gap. This paper re-examines the relationships between critical gaps estimated by maximum likelihood estimation (MLE). One of its findings is that the effect of traffic volume on critical gaps is rather weak. This rather counterintuitive result supports the HCM's procedure for capacity and delay estimation; it is quite possible that drivers adjust their behaviour in both high and low traffic volumes. A warning is given that this conclusion is obtained by assuming flows below 700vph for major roads and 400vph for minor roads at the junction. Therefore, studies for high-flow conditions are strongly recommended, to substantiate this conclusion. The paper uses empirical data from the T junction of South Canal Road and Synagogue Street in Singapore. South Canal Road is a one-way road with four 3m wide traffic lanes, with an exclusive left-turn lane on its near side. Synagogue Street is a two-way street with two 3m wide lanes. The junction is very flat and has full visibility. Another finding is that Troutbeck's method of handling inconsistent drivers in the MLE procedure produces more consistent results.
Abstract