This paper discusses a delay model for variable demand conditions. The model is applicable to the entire range of expected operations, including highly oversaturated conditions with initial queues at the start of the analysis period. The model is used to clarify issues related to the determination of the Peak Flow Period, as well as the periods immediately preceding and following the peak. While the model is for signalized intersections, the issues discussed about variable demand modelling apply to other types of intersection generally. Flow rates for the Peak Flow Period and the non-peak periods are determined using the Peak Flow Factor (PFF), a generalization of the well-known Peak Hour Factor (PHF) parameter. The location and duration of the Peak Flow Period are determined from the demand profile. Average delay is estimated for the Peak Flow Period, non-peak flow periods, post-peak oversaturation period when applicable, and the Total Flow Period, for a variety of operating conditions. It is found that the average delay to vehicles arriving in the Peak Flow Period appears to be a reasonable estimate for the corresponding average delay in the Total Flow Period (based on average delay calculated for all flow periods). The paper concludes that single-period analysis is adequate provided that the Peak Flow Period is determined with due attention to peaking in the Total Flow Period, and the use of PFF (PHF) parameter appears to be sufficient for this purpose even when oversaturation persists beyond the Total Flow Period. On the other hand, the use of the average degree of saturation with no consideration of peaking can lead to significant underestimation of delay, particularly when operating at or near capacity conditions. These findings are confirmed by comparing the model results with other models found in the literature.
AKÇELIK, R. and ROUPHAIL, N.M. (1993). Estimation of Delays at Traffic Signals for Variable Demand Conditions. Transportation Research 27B (2), pp 109-131.