In 2014, San Francisco Municipal Transportation Agency (SFMTA) piloted a transit-friendly signal progression along Geary Street, a mile-long one-way corridor with frequent bus service.

Due to the presence of a dedicated bus lane, which minimizes variations in bus travel time. SFMTA implemented a signal progression, with dwell time offsets, averaging 18 mph. Using spreadsheet models, planners were able to assess travel time and dwell time distributions, with the goal of not only reducing travel time and delay along the corridor, but also minimizing variability in travel times, which creates increased operational difficulty and frustration for riders. SFMTA utilized automatic passenger counter (APC) and automatic vehicle location (AVL) technology to collect data for the model.

The models found a high amount of variability in dwell times, further complicating implementation of an optimal signal progression. Planners opted for a progression that would favor “most” buses rather than “all” so as not to punish faster buses, or create sub-optimal coordination with other users (especially pedestrians). Finally, the project set a goal of having buses arrive roughly 5 seconds after a signal turned green, with the rationale that a solid green allowed bus drivers to progress through the intersection at a smooth speed without hesitation.

After timing and tuning the signals, the project yielded notable improvements to both transit corridor travel times and variability. Over a single mile, the PM peak travel time fell by an average of 20 seconds, a 3.4% time savings. During the same PM peak, schedule deviation also fell 5.6%, reducing travel time variability and increasing transit predictability.