Tuesday, January 8, 2008

TSP

I got asked a question about how to analyze transit signal priority at traffic signals and to answer the question, I would say that simulation is a good tool for the assessment. I would also say that you want the simulation to reflect what's going on in the field and for this to be true you should use the controller software that is operating in the field. This leads us to a new development called software in the loop simulation which you might have heard about. The University of Idaho has worked with PTV to integrate various controllers into their model with varying degrees of success. I am helping the U of I as a peer review member for their training materials they are developing.

With respect to setting TSP to 10 seconds or an amount of time to give, I would say that is normally a good place to start, but at the same time look at the policy of the agencies involved, the cycle length (10 seconds means a lot more in a 60 second cycle), and the traffic impacts you and the other agencies are wiling to allow. Now if ever there may be conflict, this is where it is, but that's what life is all about. I am spending a lot of time helping WMATA in Washington DC figure this all out with their partnering traffic agencies.

This is okay to be over capacity momentarily, as long as people are willing to accept that. I would model person delay in VISSIM and use that as a starting point. The transit agency service is a cost that we're all investing in, so even motorists have a stake in making transit effective.

Beyond the 10 seconds, if it is in the middle of the night or other periods of low traffic volume intensity, let them be more aggressive with the signal timing because there's no reason not to be if there's no traffic. Keep in mind the public (and UTA) are still paying the driver the same amount per hour, so there is savings in getting them through the intersection quickly.

One quick aside is that one of the other problems with a 10 second rule is that it is a rule that only makes sense if it serves our purposes. Why aren't we using the detectors and the information we know in the field (green times) to estimate delay. Our NCHRP 3-79 research focused a bit on that and developed some algorithms. This is something transit agencies could use to sell the aggressiveness of TSP to the agencies. For instance, if your signals could tell you that they were operating at LOS B and that by giving 10 seconds you would only go to LOS C on the side street, wouldn't you say, it is okay to go to LOS D if you knew it was only going to be this one cycle?

As you can probably tell, this is where I love to debate both the details of signal timing and the policy level of transit performance and what is best for all of us, so I'd be happy to talk more. I am off to TRB on Thursday, but if you want to, drop me a line. I think we may even have a UDOT flexible services contract that we never use, so if you want me to look into that or help provide some review of the work that's been going on, I'd be happy to help. You know us consultants, always looking for someone to work for :)... it must be nice on the other side.

The example that I am passing along is a study that is a demonstration of a corridor review where we used Synchro as a first cut to estimate the impacts and benefits. The impact assessment is better than the benefit assessment, but from UDOT's perspective this may be your primary focus. I will say that we weren't allowed to be very aggressive in this example and we were trying to show limited impacts.

No comments: