Common or variable cycle length policy for a more efficient network performance?

To coordinate signals along a corridor or in a network, the conventional wisdom of using a common cycle length for all coordinated intersections has been widely used and has led to very desirable results. On the other hand, using variable cycle length for this purpose may result in good signal coordination as well, however, the effectiveness and appropriateness of this method remains unknown. In this study, we compare the effects of using a common cycle length, to the effects of using a variable cycle length on signal coordination in a network of oversaturated intersections. For this comparison, we have used our microscopic-simulation-based Intelligent Dynamic Signal Timing Optimization Procedure (IDSTOP) that uses Genetic Algorithms to find near-optimal signal timing parameter in a network of oversaturated intersections. IDSTOP determines near-optimal signal timing parameters on a network of oversaturated intersections with a) fixed cycle length policy, and b) variable cycle policy. In both cases, we have made sure that the network was working at its near-optimal condition. In the first policy, IDSTOP determined a common cycle of 144 seconds and offsets and green splits so that in each corridor the through movements were coordinated. For the common cycle policy, the network processed 4854 vehicles that traveled 482 vehicle-miles, and on average each vehicle experienced 1.27 minutes of delay. The analysis of offsets showed that through traffic movements were coordinated in each corridor of the network. In the second policy, IDSTOP determined cycle length that ranged from 136 to 156 seconds. This shows that the cycle lengths were not similar but the range of that was narrow. The network processed 5298 vehicles, that travelled 501 miles, and the delay was 1.23 minutes. With the variable cycle policy the network processed 8.4% vehicles more than the first policy, vehicle miles travelled were significantly increased, and the average delay per vehicle did not change significantly. The analysis of the offsets showed that through traffic movements were coordinated when needed. These figures show a significant improvement over the traditional common cycle policy. The finding, as an example of other conditions that were analyzed, indicated that using variable signal timing policy results in a considerably higher number of vehicles processed by the network as well as a significantly lower delay in the network compared to a common cycle length.