Impacts of Intercycle Demand Fluctuations on Delay

This paper demonstrates that in addition to intracycle demand fluctuation, which is already a consideration in many delay models, intercycle demand variance also impacts average delay at signalized intersections. Webster-type delay models treat demand fluctuation over the whole analysis period, often $15\min$ or longer, as if it were just within a single cycle. Such an approach is fine if used judiciously, one might presume. However, results from Monte Carlo simulations with the incremental queue accumulation (IQA) method indicate that Webster-type delay models will underestimate the average delay under heavy traffic conditions. As unutilized capacity at a signalized intersection cannot be saved or carried over to be used by succeeding cycles when demand surges due to normal fluctuation, better understanding of the patterns of intercycle demand variance is important. Simulation results demonstrate that different patterns of intercycle demand variance can result in different levels of average delay. A low-to-high demand pattern will cause a higher average delay than a high-to-low pattern would, even though the overall demand level is exactly the same. It is therefore clear that neglecting intercycle demand variance may lead to significant inaccuracy and, hence, suboptimal signal timing decisions.