Focused and Dispersing Forward and Backward Progression Algorithms were Tested for Adaptive Signal Timing Strategy with Various Trip Dispersion Rates, Grid Size, and Grid Topology.
Statewide, California, United States
Traffic Signal Plans to Decongest Street Grids
Summary Information
Adaptive signal timing strategy can improve traffic efficiency by dynamically adapting to changes in traffic demand and flow. This study developed two signal timing synchronization strategies to address not only steady state traffic conditions, but also heavy congestion and queue spillover conditions. The focus was arterial grids formed by two sets of roughly parallel intersecting streets. Both strategies produced synchronized offsets (forward or backward) in one direction of every link on the network. The selected two strategies were benchmarked with simulations against a fixed, zero-offset strategy and also against a state-of-the-practice computer program. Percent change in vehicle hours delay (VHD) in the network was used to evaluate the two strategies.
METHODOLOGY
The selected network was a non-homogeneous, rectangular grid formed by two intersecting families of roughly parallel, two-way streets with a traffic signal at every intersection. Two signal timing strategies were developed. The first was a static strategy called Focused Forward Progression (FFP), designed for the morning rush hour, aiming to provide perfect forward progression for all intersections on a path towards a reference intersection. The second strategy was adaptive, toggling between FFP and the focused backward progression (FBP) based on real-time traffic density. The area in which the offsets were chosen to be adaptive in the grid network was a 6x6 center of the grid. For the evening rush, the same strategies were implemented for all travel directions pointing away from the reference intersection. They were named DFP (Dispersing Forward Progression) and DBP (Dispersing Backward Progression). Various scenarios were tested to evaluate the impacts of the strategies. Especially, in the Morning Commute with Concentrated Workplace scenario, workplaces were distributed randomly using a Gaussian distribution centered at the center of gravity of the grid network and a standard deviation such that 40 percent lie in the 6x6 grid central district.
FINDINGS
- Timing plans employed by the simulation software reduced VHD by 6.6 percent, while the first and second strategy developed in this study reduced VHD by 20.6 and 32.2 percent, respectively, compared to the benchmark (i.e., zero offset) scenario.
- In the Morning Commute with Dispersed Workplaces scenario a similar analysis was repeated with the same global demand rates but with six dispersion levels in total. The results showed that VHD reduction varied by -2.2 percent to 29.7 percent for the first strategy, and by 0.6 percent to 38.6 percent for the second strategy.
- In the Morning Commute with Multimodal Distribution of Workplaces scenario, two non-overlapping 6x6 center of grids were used for toggling between FFP and FBP. Trips had equal likelihood to go from any home to any workplace. It was found that two center of grids did not have additional improvement in comparison to one center of grid: Reductions in VHD for the first strategy was 17 percent and 18.9 percent for one and two center of grids respectively. Reductions in VHD for the second strategy were 23.2 and 20 percent for one and two center of grids respectively.
- In the Evening Commute with Concentrated Workplaces scenario, the first and second strategies reduced VHD by 25 and 28.2 percent respectively. In addition, the results showed that increasing grid size to 8x8 increased improvement in VHD by 33.6 Percent for the second strategy.
