Wednesday, April 2, 2008

TSP Question from San Francisco

It is an interesting question that you pose. You were interested in “the physical operation of TSP, its strength and weaknesses, limits, and the range of flexibility for modification to fit certain traffic conditions”.

To answer this, I would simply say that TSP enables a communication between the traffic signal and the transit vehicle to provide an opportunity to adjust the signals to improve the efficiency of transit movements. Obviously, knowing when a bus is going to stop to pick up passengers and when it is going to move through a traffic signal is of key interest, thus the typical interest in far side stops.

Given the assumptions, I will give you some advice on what I would say having implemented TSP in Portland with some corridors that are similar (signals every 260 feet in some cases with stops every other block).

He bases his assumption on the following facts: 1) On a 2.3 mile corridor Van Ness has 29 intersections, 2) cross traffic on several streets cannot be impacted too much because the traffic is very heavy so Van Ness doesn’t have as much control re. signal timing as may be needed,

** Pedestrians may be more important than the traffic, given what I know about the City’s policies. Jack Fleck and Bond Yee from the City are two that I have talked with about this in the past. This largely depends on the performance measures that are used, so I would say I am not sure that I would agree with this in all cases. If the signals are actuated, it may be possible to benefit the bus even at near side stops.

3) the likely infrastructure plan will use exclusive bus lanes so traffic congestion is either a minor issue or not an issue at all as buses move in the corridor, 4) buses will be able to average a consistent speed on the exclusive lanes that will be a major part of the signal timing plan, 5) the likely headway of buses in the corridor will be about 2-3 minutes at peak times in both directions, and 6) if a bus falls off the signal progression “schedule” it automatically resets when they get the first green – and then they should see green lights for the next 2-4 blocks until a stop is made, and if they complete the stop in the allotted time, they will continue to see green at that light and 2-4 more. He developed a similar signal / near side transit stop plan for F-Market Streetcar service on Market Street, and his assumptions were proven mostly correct. The end result shaved 3-4 minutes (10%-15%) off the run time on the route in one direction and 1-2 minutes off in the other direction.

** On most of these points I agree, but one issue that is different than streetcar service in many cases, is that there are situations where buses aren’t stopping at every stop. Given this case, it may be beneficial to have TSP at select locations that could reduce delay associated with bus stopping or not. A standard plan such as what was implemented on Market Street (which I have seen operate and is an extremely good plan), works well if the stops are consistent. The bus service would benefit during periods when bus stop frequency at a particular stop is variable. I know the signal vendor that the City is working with on the 3rd Street light rail project and one of the innovative steps that they took on that corridor was to use the TSP system and detection of multiple intersections upstream of the stop to reduce the likelihood that any stop would occur. This sort of predictive priority is a powerful tool, considering the potential for bus stop dwell times that are variable. One of the reasons the Market Street corridor looks good is that it leaves the stop when the signal goes green. In many cases, the dwell is unnecessary and activation of TSP may improve that operation adjacent to the stop and at the downstream intersections.

** The reliability of the service may also be improved with TSP, if it is used only for vehicles that are late. This would help maintain the 2-3 minute headways and result in bus bunching, but it is the case at that as the headway increases to that sort of level (near the signal cycle length) that changing the signal timing may be best and TSP becomes less effective.

No comments: