Measuring Signal Platoon Flow

Traffic‐flow models are often based on an assumption of random vehicle arrivals. Arrivals are not random where upstream flow interruptions, such as traffic signals, cause platoons to form. Some way is needed to identify and describe periods of high and low flow rates if these traffic models are to be more accurate. A statistical procedure is developed to predict these flow rates. The procedure is demonstrated with five sets of data from unsignalized intersections affected by platoon flow. The demonstration includes identification of the high‐flow period, determination of confidence limits for flow rates, and testing for randomness of the vehicle arrivals. Ramifications of using the procedure in the analysis of unsignalized intersections and other types of traffic‐flow problems are then discussed.