As the Green Line opened on June 14th, most of us were just excited to finally have a new service plying the busy Central Corridor area, but many were keeping an eye on their clocks and watches, checking its speed against the schedules that had been posted a couple of weeks earlier. On opening weekend, many trains were taking more than an hour to make it from end to end. The signal priority system that’s necessary for trains to operate smoothly just wasn’t working right. But hey, it was opening weekend, when 107,000 people turned out for free rides over two days—things were bound to be worse than normal.
Sunday turned to Monday, the first day of revenue operation, and things were…still slow. Fortunately, signal timing has been refined through the University of Minnesota campus and in downtown Saint Paul since then. Riding today is much less aggravating than during the first few days, but trains still aren’t hitting their scheduled times, so it’s clear that there’s room for improvement.
It has been difficult to find good information about how signal priority systems work in general and particularly how they’re set up on the Green Line. I sent a bunch of questions to John Siqveland of Metro Transit to try to clarify the situation. Personnel from Metro Transit and Saint Paul Public Works were kind enough to answer them.
Q: When did work begin on setting up signal priority?
A: Transit Signal Priority (TSP) was identified as the strategy to be used for traffic signal operation during the project scoping and design strategies in 2008 or 2009.
The Transit Signal Priority timing values were developed with the signal timing plans for the corridor in 2012 and were implemented during the train testing phase and continue to be reviewed and adjusted.
Q: How much progress has been made?
A: Since partial implementation began in February, we believe we have perhaps 10-15% more refining to do at this stage.
Q: How long is it expected to take to get signal priority functioning as well as possible?
A: This is a continuous process and transit signal priority adjustments are made as we determine them. We expect another step forward within the next few weeks.
Q: Is signal priority being implemented in downtown Minneapolis?
A: No. It operates the same as Blue Line. LRT is however accounted for in the 5th street signal timing plans.
Q: Is signal priority being implemented on the University of Minnesota campus?
A: Yes.
Q: Is signal priority being implemented on University Avenue?
A: Yes.
Q: Is signal priority being implemented in downtown Saint Paul?
A: Yes.
Q: Are there any particular places that have been problematic?
A: We have been dealing with some issues with equipment and are working through a troubleshooting process to try and pinpoint the issue to determine if it is software or hardware. The City has recently installed a solution to this problem that appears to be working. We have been focusing on making adjustments to minimize stopping at lower volume intersections along University Avenue.
Q: Are there places on the line where signals are working particularly well?
A: Yes – The University of Minnesota campus area.
Q: Do traffic signals detect cars at intersections?
A: Yes.
Q: Are traffic signals set up to create “green waves” for cars, especially on University Avenue? If so, how fast are they supposed to propagate along the roadway?
A: The traffic signal timing plans are developed to create “green bands” to facilitate movement of all vehicles in both directions on University Avenue. Adjustments are being made to provide good progression for cars at 30 mph and for trains at their posted speeds.
Q: In what ways are controllers able to change the phasing/timing of signals? Is green extension the only method available, as indicated on this Metro Transit page about route 10, or are the options broader, such as on this list on Wikipedia?
A: The controllers are using transit signal priority to provide green extensions and early greens and at locations where LRT has its own exclusive phase, there is phase insertion.
Q: Do the TSP emitters/detectors use radio or optical (infrared?) communication? Some other method?
A: There are loop detectors embedded in concrete between rails that are hard wired to the signal controllers.
Q: Are signal controllers able to detect trains via the tracks?
A: Yes. Via the loop detectors.
Q: How close does a train need to be to an intersection before a controller detects it?
A: Any detector is able to provide a detection call to any traffic signal in the system. This allows a call to be placed to a traffic signal well in advance of a particular intersection.
Q: Is the TSP system able to understand how far away a train is from the next intersection and how soon it will arrive, or does it simply aware of the presence/absence of an oncoming train?
A: Yes. The loop detectors send a signal to the controller when the LRV crossed the detector. Based on the spacing and estimated travel time, the timing plans are set.
Q: Can a signal controller know that a train is coming if it is more than one intersection away?
A: Yes.
Q: Does the detection of a train trigger a “green wave” for the train across multiple intersections, or does the detection only work one intersection at a time?
A: No. The signals are timed to provide a green band on the street to allow trains and cars to travel through on a green light. The train detection tells the signal the train is approaching which allows the controller to implement TSP by bringing up an early green or extending the green time if it is needed.
Q: Have signals been changing too early for trains because they get detected when they’re too far away (such as before pulling in to station stops)?
A: Every situation may be different, but if the train has to slow unexpectedly or someone holds the doors at a station, for example, it can have this effect.
Q: Can signal controllers at neighboring intersections communicate with each other?
A: Yes.
Q: Are the interlocking signals able to communicate with traffic signal controllers?
A: No. Interlocking signals take priority (just like Blue Line).
Q: Is communication between train emitters and stationary detectors being affected by weather?
A: No.
Q: Is communication between train emitters and stationary detectors being affected by obstructions such as catenary poles and wires or signal poles?
A: No.
Q: What are the government entities that own and manage the hardware involved in signal priority? (City of Minneapolis? City of Saint Paul? Hennepin County? Ramsey County? MnDOT? others?)
A: The hardware is owned and/or operated by all of these entities plus Metro Transit.
Q: In late May, it was reported that the City of Saint Paul had just received some new hardware that was necessary for the correct TSP operation. Can you explain what that device was, and how it relates to TSP?
A: There was some equipment (C-11 connectors) that needed to be procured for implementation of part of TSP (Dual phase insertion)
Q: Who are the manufacturers of the signals, controllers, detectors, and emitters involved with TSP?
A: City of St. Paul utilizes Econolite (ASC 3) signal controllers and software. The loop detection and cards are by Reno. Minneapolis utilizes Peek Controllers with Northwest signal software.
Q: Who are the contractors involved with installing and configuring TSP?
A Traffic signal construction was done by Ames-McCrossan and Walsh. TSP programming has been installed by the cities with assistance from Central Corridor Project Office and their consultants.
Q: How many levels of priority/preemption are available with these signal controllers?
A: These controllers are able to be configured for many levels of priority and preemption. For instance, today the controllers handle emergency vehicle preemption, TSP for LRT, and regular vehicular traffic operations.
Q: Is the TSP detector integrated into the controller, able to directly alter the signal phases, or does it just send messages to a “black box” signal controller where it doesn’t have any insight into the signal controller’s operation?
A: TSP is operated by the traffic signal controller.
Q: Could an emitter on the rear end of a train cause problems with TSP?
A: Trains do not use emitters.
Q: Have LRV operators been forgetting to either enable or switch the direction of the TSP emitters on the train?
A: There are no emitters. Train operators do not have to do anything to enable TSP.
Q: There are dozens to hundreds of cars passing through each intersection for each train that goes through. Are signal controllers being overwhelmed by the number of cars?
A: Not now and not before.
Q: Is signal preemption from emergency vehicles causing problems?
A: We do not believe so.
Q: Are there any laws or regulations preventing more aggressive signal priority, such as a requirement to maintain previously-existing levels of service (LOS)?
A: No.
Q: Are there other comments you would like to make?
A: No. Thanks for the opportunity.
I was glad to get these responses, in part because it proved that some of my assumptions had been wrong. Did these questions help you better understand the behavior of car and train signals along the Green Line? What would you ask about?
Wow. Thanks so much Mike. This is deliciously nerdy.
“Open the Green Wave doors, Hal.”
“I’m sorry Mike. I can’t do that.”
Fascinating, nice work Mike. Now I need to go make friends with some TSP controllers. 😉
Can someone translate this for me?
A: The controllers are using transit signal priority to provide green extensions and early greens and at locations where LRT has its own exclusive phase, there is phase insertion.
And I don’t understand how this answer relates to the question – “no” to an either or question. What’s the difference between a “green wave” and a “green band?” Help!
Q: Does the detection of a train trigger a “green wave” for the train across multiple intersections, or does the detection only work one intersection at a time?
A: No. The signals are timed to provide a green band on the street to allow trains and cars to travel through on a green light. The train detection tells the signal the train is approaching which allows the controller to implement TSP by bringing up an early green or extending the green time if it is needed.
Seconded — could someone do a glossary of technical terms being used here?
Those do definitely seem to be in conflict, and I’m not quite sure how to resolve them. That’s something to follow up on.
As for “green band”, that term is new to me, but I believe it has to do with what it looks like to plot out the signal phase over time for a number of intersections along a corridor. An example is figure 6-1 here: http://ops.fhwa.dot.gov/publications/fhwahop08024/chapter6.htm
A bunch of bar graphs next to each other, where one axis is time and the other axis is relative position. When the signals are working correctly, there will appear to be bands of green sloping forward in time as you move along the corridor. Hopefully that makes some sense.
For the first answer, my understanding of his answer is that green extensions/red delays are used regularly. But at the locations where the LRT needs it’s own exclusive phase (e.g. reds in all directions at Cedar & 12th) they need to insert that phase.
I am not sure what “dual phase insertion” is. My guess would be that it applies to using phase insertion at two signals simultaneously (e.g. so the train doesn’t get stopped on the Cedar Ave Bridge)?
Regarding green bands, see page 26 of this document: http://dspace.mit.edu/bitstream/handle/1721.1/84303/50194954.pdf?sequence=1. I think speaking of bands rather than waves is more useful when you are talking about bidirectional traffic flows.
That specific problem, where drivers with conflicting movements get the green at the same time, is something that signal engineers work very hard to prevent. Signals are designed from the ground up to prevent that type of problem from happening. If that happened because of a fault in the system, the most likely outcome is that the signals would detect a problem and start flashing red, like we sometimes see after a storm has knocked out power.
Near as I can tell “green band” and “green wave” mean more or less the same things, band is the conceptual term on paper, and wave is the term as it functions on the street. As for LRT having it’s own exclusive phase, that’s another puzzler. It’s not like the Blue LIne where a train will cause gates to come down. It’s possible that LRT changes the phase somehow, and that’s referred to as an “exclusive” phase, this more applies outside of downtown. In the core downtown areas all the streets in both direction are timed as part of a system; you can’t do much to alter the timing without causing gridlock for cars.
Interesting that Minneapolis apparently does their own thing in signal-land in yet another way, they apparently use 170/2070 series controllers, rather than NEMA like St. Paul and the rest of the state. They’re a lot more difficult to program but you buy the software separately so you can use any vendors, or even write your own, rather than what came with the controller. Programming these things is fantastically complicated- it’s not to hard to set up a basic intersection, I even managed to do a basic 4 phase intersection on mine- but when you get involved in coordination and preemption and doing so at a whole string of intersections, some of which are themselves also coordinated the other direction, it’s more complicated than noticing that the train gets caught at the red on Hamline and just clicking a button on a computer to lengthen the green time.
I’m with Janne in that much of this is a bit too technical for me to make sense of. But it sounds like it’s basically good news because the plan is for there to be signal preemption, at least between the two downtowns. Is that right? I thought there was a conflict between transit planners who wanted signal preemption, and county road planners who did not want it because it would (presumably) add to traffic and backups on the north south roads in StP.
John Siqveland does not allude to this problem, and implicit in all of his responses is that this is simply a technical problem that needs to be fixed and smart people are fixing it. All the powers-that-be want it to be fixed. So there will be a time that in the not so distant future when the Green Line will only stop at its actual stops.
If this is all true then my day is made. Do I have it right?
No. Hard preemption, where no matter where in the cycle the light is it turns green for the train and the flashing white light goes on, would be absolutely devastating to car traffic- witness how long it took them to get Hiawatha somewhat decent, and that’s with a lot fewer intersections. What they’re trying to do is allow variations from the normal cycle. During a green wave setup, you try to get it so all the cars go through on the green band for the main street, and there’s few cars on the red, where you allow cross traffic to go. What they’re trying to implement is enable the greens to be lengthened, so you take some time outside the green band that would normally be devoted to other phases and instead enable a LRT vehicle to go through on the main street.
Give them a few months before everyone starts screaming at them. Getting a green-wave right for cars is an exact science- there’s charts online for given speeds, signal spacing, and volumes. There’s no chart to try to get a LRT vehicle through the same signals, so they have to observe in the field and fix things based on their observation.
Besides soft preemption, another concession to LRT was having protected turns at the same time. Normal practice for coordination nowadays is to have one at the start and one at the end of the main phase, since that was determined to be the most efficient way for cars and flashing yellow arrows eliminate the yellow trap, but that also means the time a train go through would be a lot shorter.
One of the things that emerged about the Hiawatha signals after several years was that the originally-installed signal controllers weren’t able to pick up from where they left off after being preempted (“hard preemption”, in your phrasing) — a pretty glaring issue.
Done correctly, i think a full preemption scheme on University Avenue would be okay. We’re still talking about an average of one train every 5 minutes, and in some cases two trains cross at (nearly) the same time, making it seem more like one train every 10. Cross traffic has the bigger flows here, rather than paralleling the tracks. Sure, there will be groaning at peak times, but most of the day isn’t going to be hurt that much. The big question is whether it can be set up to trigger reliably and still give pedestrians enough of a chance to clear the intersection.
I get the sense that you may not have spent much time watching signals along University yet — I think you should, and you’d probably observe some things that many of us miss. They are still making some effort to have leading/trailing left turns like you mention, though they aren’t consistent.
Anyway, it’s more complicated than we would like it to be — certainly not impossible to get working well, but trickier than many of us ever realized at first.
Insertion, I believe, is when the cycle normally goes A –> B, where A is one street and B is the cross street, but the insertion is that the LRT would be detected and C is triggered. In this case, they likely then prioritize C, so once A is done it goes to C. If not prioritized, it would simply go to B then C, or if already on B would go to C.
Correct, except that on the Green Line, phase insertion is only being used where the train requires an exclusive phase. In other words, the cycle is normally A -> B -> A -> B …. And C is a phase that only occurs when a train triggers it.