Stochastic nature of freeway capacity and its estimation

Abstract

The main purpose of this study was to investigate the meaning of freeway capacity, in particular to explore its stochastic nature and to estimate its distribution. The evaluation is based on traffic data from three busy urban freeway sections over 3 full days. Momentary capacity is defined as the intersection of the best-fit regression lines calculated for the dense- and unstable-flow regimes close to the maximum flow. An algorithm was developed for the selection of the relevant pairs for each regression line and is discussed. It is argued that momentary capacity values are stochastic in nature and distributed according to the shifted gamma distribution. Estimation of the parameters of this distribution for the three urban freeway sections is studied. It is proposed that the 5th percentile of the distribution, found to be approximately 2330 vehicles per hour per lane for the prevailing conditions of the basic section, could be adopted as the representative design value of capacity. Another conclusion is that the average distribution of capacities for all three through lanes is relatively close to the distribution of capacities in the middle lane.Key words: freeway flow, capacity, flow breakdown, dense flow, unstable flow.