Three methods frequently reported in traffic literature for obtaining flow-concentration-speed relationships were selected for an experimental validation study. In the first method, data are grouped into short successive time intervals for computations; in the second, cars are classified by computed virtual concentration; and in the third, cars are classified by their speeds. A standard method was developed against with these 3 methods were compared. The method used as a standard is based on the isolation of periods of constant traffic flow. The data used for the study were collected by the federal highway administration's traffic analyser and were taken at 12 unidirectional two-lane sites. Results show substantial agreement between the methods based on short time slices, constant flow intervals, and speed classes. However, the method based on virtual concentration disagrees with all of these in the high concentration range. Because the method based on virtual concentration yields flows much higher than any ever observed in practice, it should not be considered a valid method for obtaining flow-concentration-speed relationships. (Author/publisher)
Abstract