A capacity analysis approach intended as an alternative to the traditional Highway Capacity Manual (HCM) method was evaluated. One- and two-stage models of pre-queue and queue discharge flow (each of which might be thought of as representing “capacity” in some sense) were developed and compared with one another and the HCM method. Two-stage models related capacity flows to intervening variables, including average time gaps (average time separations between the rear of a vehicle and the front of one following it) in the critical lane (that with the highest flow rate) and the critical lane flow ratio (the flow in the critical lane divided by the average flow per lane), and then related these intervening variables to the geometric, vehicle population, and driver population characteristics of bottleneck sites. One-stage models involved direct relationships between capacity flows and site characteristics. Differences in capacity flow among study sites were primarily the result of differences in average critical lane time gaps; however, critical lane flow ratios were also important. The performance of the one-stage and twostage models was similar. For the sites used to develop the models, both were better able to predict pre-queue and queue discharge flows than was the HCM method. In particular, the HCM method tended to overestimate actual bottleneck flows, especially in queue discharge. However, neither type of model was successful in explaining variations in capacity flows at additional sites used for verification. Once apparently anomalous data were eliminated, the only significant explanatory variable in the models was the number of lanes. Consequently, it is recommended bottleneck capacity analyses continue to be based on existing HCM methods but that these be supplemented by use of a look-up table based on the means and standard deviations of pre-queue and queue discharge flows for sites with particular numbers of lanes. (Author/publisher)
Abstract