The post showing Dundas Street as it was in the mid 60s spawned many comments about traffic signal operations. That’s the sort of site this is. We get started on one topic and find ourselves on another.
To clear out the clutter, I have created this post as a separate thread to hold those comments.
Steve Munro 
Interesting side discussion on the advance green lights. There also used to be (often) signs opposite the advance-green direction saying something to the effect of “delayed green signal.”
The flashing greens are being phased out (along with the last 8-8-8 traffic signals) in Toronto to avoid confusion as to their purpose. I am well aware of the BC usage, but there, it is a much slower flash sequence. Most other places with BC’s style of crossing have motorist-facing signals with a flashing yellow light that switches to a solid yellow, then red, for pedestrian-actuated crossings, which is less confusing.
About hurrying drivers into the turn: Waterloo Region uses flashing green arrows to draw attention to the advanced phase, which I find interesting. (Waterloo is also big on roundabouts, which don’t work in dense urban settings, and is also switching to Hamilton-style black and yellow signal housings, something I would advocate here).
This is a great series, Steve. Thanks!
LikeLike
The other problem is how the walk/signal lights are programmed in relation to the law. This is especially apparent in the suburbs where an artery and a side street with little traffic meet.
I’ll explain: The artery’s traffic light goes green, the white walk sign appears, the count down timer begins and counts down to zero, with no side street cars waiting the signal ‘resets’ and the artery’s traffic light remains green (instead of turning red), the pedestrian crossing now shows a white hand, and the cycle repeats.
This means although the artery has always had a green light and it remained green, a pedestrian would have been breaking the law if they started to walk against the flashing hand, even though the walk signal later turns to white.
Perhaps people would obey the traffic rules more often, if the people responsible for the infrastructure actually paid attention (or showed some passing interest) with how it operates in relation to laws on the books.
Steve: This problem happens right downtown, a block from where I live. It is also a very confusing way to operate a traffic signal as the countdown implies an impending change, but none occurs. This can foul up pedestrians on the side streets who have not pushed the “walk” button assuming that the light is already in the process of changing. The fact that some intersections won’t even start a countdown without a request to cross (or an auto on the detector loop) further confuses the issue.
LikeLike
Robert Wightman wrote,
This either misunderstands what I was suggesting, or perpetuates the apparent idea with Toronto road services that traffic light control logic should be somewhat static.
In some cases I have seen, one direction gets a left arrow at the start, while the other gets it at the end. Depending on traffic flows at a particular intersection, this can REDUCE the chance of the left turn lane getting filled up. Whether the left time is at the beginning or at the end in simple theory should not matter, but the arrival pattern of vehicles can result in the lane filling up sooner in one case versus the other. A little more intelligence in the controller logic should be used to make dynamic decisions in the matter.
Also, another way would be to give a left arrow to one or both directions at the beginning of the green time. When one or both directions did not receive it at the beginning but needed it by the end, it would be provided then. Again, more intelligence in the controller is needed.
LikeLike
Calvin Henry-Cotnam says:
Most major intersection have “smart signals” that determine the amount of traffic in the turn lanes and do not give the signal unless required. The geometry comment is mainly about streets in the old city of Toronto. On Spadina and other old streets left turning cars often back up into the left through lane. If you put the turn signal at the end you would lose use of the left lane for the entire green phase. There is simply not enough time at most intersection in the old city to have a turn at the end as well.
The close street spacing requires shorter cycle times so there has to be a time limit on the turn phase to allow the through phase to be long enough. Another problem, especially on streets like Spadina, is that the length of time for the cross street is determined by the time required for slow pedestrians to cross. I believe that the length of the flashing hand phase has to be 1 second for ever metre of road width and the length of the walk phase has to be long enough to let the screaming hordes get into the intersection. At Spadina I believe that the Dundas green phase has to be almost 40 seconds to allow the pedestrians to safely cross the intersection.
There are a few intersections, notably Front and Spadina, Lakeshore and Spadina, and Bathurst and Lakeshore that have suburban length long cycle times and this puts delays into the Bathurst and Spadina cars. You cannot run a 2 minute headway if the cycle time is 3 minutes. In 6 minutes there would be 3 transit vehicles but only 2 green phases. You have to get bunching as a result of the incompatible timing.
I didn’t miss-understand your comment but I wanted to point out that it is not always possible to do what you want. Again we have to decide whether we want to design for the automobile or transit and pedestrians.
The suburbs with their wide streets and wider intersections can have the extra turn phases but this results in very long cycle times. I believe that this is a major cause of red light running; people don’t want to have to sit for another 2 minutes waiting for the next green. The suburbs could use smart signal to determine if there is any traffic coming on the street with the green phase and if there isn’t then give the green back to the other street. I have often sat for 2 minutes waiting for a green while only two cars have used the other street. If this didn’t actually speed up traffic it would at least make the drivers waiting at the red happier.
As I said the geometry of the road plus the street spacing determines where the turn phase goes and how long it can be. There is only so much time in each cycle and it needs to be used wisely. Your ideas will work in some suburban areas but not in very many in the downtown
LikeLike
Robert Wightman wrote,
Good that you quoted “smart signals”. As I previously mentioned, the spec is to only provide the turn signal when there are at least three cars, but its implementation is to provide the turn signal when there is a third car. I try to be the third car when there is no first or second. Furthermore, these “smart signals” simply provide a turn signal for a fixed duration. If that duration is long enough for eight cars to make it through, but there are only three, other traffic waits. My point once again is that more intelligence is needed in our controllers. This is not a pipe dream, as these things are done in other jurisdictions. Then again, our “we must think of it ourselves” attitude probably makes it a pipe dream.
Robert Wightman added,
In the city core, during rush hours there should be severe turn restrictions during rush hours at intersections where a dedicated turn lane is not possible. Many have this now, but there are the odd examples where this isn’t the case. The point about losing the left lane for the entire green phase is out the window when there is a turn arrow or advanced flashing green and the first four or five cars are all going straight. The first car needing to turn misses the advanced turn and blocks the left lane for the rest of the cycle.
Robert Wightman added,
I never suggested adding more time to the cycle. If a direction has green time that is 40 seconds, but the first 10 is for the turn signal, that 10 seconds is used for turns in both directions and then both directions get 30 seconds of full green. If one direction only needs it at the beginning and the other direction needs it at the end, the first direction gets a full green with a green arrow at the beginning. That full green lasts for 30 seconds with its arrow going out after 10. The other direction gets 30 seconds of full green starting after the first direction’s arrow goes out, but gets a turn arrow during its final 10 seconds. This portion of the cycle is still 40 seconds.
Robert Wightman said,
Exactly, just the way what is being done now does not work at all intersections the way the traffic planning department seems to think. What I have suggested is NOT a universal solution to every intersection. In some cases, what I am suggesting could possibly be the worst thing to do while what is being done now might be the best. Each intersection and its surroundings has to be looked at individually.
LikeLike
Calvin Henry-Cotnam says:
You left out the other two sentences which change the meaning of the sentence. In its entirety I said:
Calvin said:
I was only referring to streets in the city core like Spadina that had left turn lanes. These lanes are usually short and hold at most five cars. IF the lane is full and IF you put the turn signal at the end and IF more cars show up, THEN they wait in the left through lane and block it. If you watch at the intersection left turning cars often show up after the turn phase has ended and wait in the turn lane until the next turn phase instead of blocking the through lane. I have seen this happen in Brampton where there are long turn phases and even more cars coming than can be handled by the green arrow. This either removes the left though lane from use as it is blocked or removes the left turn lane going in the opposite direction as the cars back up to the previous intersection.
Calvin said:
Be careful about putting end of phase turn signals in where there are lots of pedestrians, in large groups they tend to have herd mentality and do dumb things, like walk in front of turning cars. At intersections that do not have left turn lanes I would be inclined to abolish all left turns during rush hour.
LikeLike
This is probably getting well beyond the scope of the original discussion, so this will be my last comment on this topic.
Robert Wightman also wrote,
My point about tailoring the operation of the signals to each intersection is buried in that. Each intersection must be considered with more care than they do now.
Robert Wightman added,
I NEVER said that the turn signal should be at the end of the cycle for this situation. I said that there should be intelligence in the controller that gives a left turn signal at the end of a cycle WHEN there was not one at the beginning and the need is there at the end.
Robert Wightman wrote,
I would agree with that, and also seriously consider banning more right turns as well (like Yonge/Dundas where all turns are banned all the time, or like Yonge/Bloor where even right turns are banned at peak times).
Steve: And can this be the last word on the subject, at least in this thread?
LikeLike
Steve, please consider accepting into the thread another comment pertaining to intersection traffic management. It’s not to deal with procedural details, but with philosophy in a certain sense.
One theme broadly represented in the discussion is that today’s already dense yet still intensifying traffic creates, at intersections, a need for flow management that exhibits both peculiarity per intersection, and real-time adaptability per intersection, in light of analogous peculiarities ascribable to intersection geometry and patrons’ varying patterns of demand.
There may be that the approach to traffic management, in question, can optimize traffic flow even if not perfect traffic flow. I wonder, though, if extensive diversification, of procedures and of signalling, might introduce some effect to compromise safety by similarly compromising intersections’ operational predictability from both the motorist’s point of view and the pedestrian’s point of view. For people don’t always perform as well as they might, when constantly needing to accommodate themselves on an ad-hoc basis, over short terms of time, to unfamiliar circumstances rather than being able to enjoy stability and predictability.
One correspondent appears to have expressed the view that, under certain circumstances, extant flow-control arrangements at intersections may be encouraging red-light-running because drivers otherwise going to be delayed, may not want to wait through two red-light phases at a particular intersection.
Traffic management issues are addressable not merely in terms of technical parameters. Modern Toronto’s motorists must learn to exhibit patience, because no technical approach to traffic management is going to present the entirety of a solution to congestion. On many accounts, I find Toronto’s modern population to be unduly aggressive and to lack the grace of being restrained both in deference to others and in deference to their formally ordained social responsibilities, compared to Toronto’s population of yesteryear. The observation has application to motorists, if not exclusively to motorists; and, in my opinion, has validity as a basis for concern, notwithstanding that the challenges accruing to one’s psyche, on account of traffic congestion, in recent years may certainly have become more imposing than before, on a city-wide basis. Indeed, under those circumstances an even more superior form of self-control is needed to be cultivated, not a lesser one.
I’ve witnessed today’s drivers demonstrating unwillingness to wait at an intersection for even one cycle of a stop signal — as greatly as by entering intersections long after signals’ red aspects have begun to be displayed; and with no competing traffic in the vicinity; and with drivers having enjoyed ample opportunities to properly come to stops, for their having been far up-street when the signals turned to yellow. Even so, on some occasions, when still having been far up-street when the signals turned to red.
Moreover, I’ve narrowly escaped being struck by a red-light-runner, a driver and his heavy truck not only not having slowed but possibly somewhat exceeding the speed limit while transiting the intersection in question, upon my starting to cross at the pedestrians’ crossing with a green light in my favour, in Scarborough at Lawrence Avenue East and Burnview Crescent — a signalled intersection nearby one elementary school and favoured by the local school children.
More lengthy yellow-light phases are needed. Respect for the yellow-light phase’s meaning is also needed: a matter representing change-for-the-worse, from yesteryear to the present. Therefore, socio-spiritual approaches to traffic’s challenges much imposing upon a person’s store of composure, might have to carry the day in terms of safety, anyway, if engineering’s approaches cannot.
Steve: I agree that traffic signals need to behave in a way that motorists can understand, broadly, that any intersection they come to exhibits one of a few set behaviours. That said, we also expect motorists to pay attention and anticipate that not all locations will work the same way. A simple example where there are now only three intersections operating this way are the all-walk phases at Dundas/Yonge, Bloor/Yonge and Bloor/Bay. These are the only locations where one street’s green does not reasonably closely follow the other’s. However, there are many locations where some sort of special turn phase exists between the green phases of two streets, and so it is reasonable to expect motorists to wait for their turn, not just gun ahead. As we well know, some motorists get caught short at places where there is a turn phase, especially if this phase is only displayed at certain times of the day, or if there is traffic requiring it sitting on a detector in a dedicated lane.
The other problem is purely technical. There are only so many “standard” configurations of traffic signal controllers, and the city engineers don’t just dream up new control strategies on a whim for each new location. The problem them becomes to ask whether all of the capabilities of what we have are actually being used, or if there are more sophisticated controllers available. A good example is the problem with closely spaced intersections. As Toronto designs things, each set of lights is controlled by its own set of detector loops and algorithms related to transit priority. There is no capability of creating “green waves” where a transit vehicle is cleared through multiple signals to reach the next transit stop without being held up in between.
Some of the problems are philosophical — should we have long, suburban style green times for motorists, not to mention provision for slow-moving pedestrians? What would the effect on the road and transit network be of a major change in the philosophy underlying how green time is managed for all users?
I am sure there are traffic engineers out there who would love to play with this sort of thing both in simulation and on the road equivalent of a full scale train set. Some of this work may already have been done, although it will need to apply to urban areas with strong pedestrian presence, transit and parking — in effect a lot of local demand of various flavours, not just cars whizzing through abandoned downtowns.
LikeLike
A couple of comments on short green times for cross streets and flashing don’t walk signals that disappear.
It is standard design to allow 1 s per m of street width for pedestrians to cross. The length of the “walk phase” is set to allow all the pedestrians waiting at the intersection to get into it before the flashing “don’t walk” sign starts. This is why the “walk” signal is longer at Spadina and Dundas than at Lawrence E and Ellington.
At many intersections in Toronto it used to be common for minor cross streets after the p.m. rush only to get a green signal if someone pushed the walk button or if a car showed up at the intersection. Many of these intersections were set up only to insert the green for the cross street on what would be every second cycle. Adjacent side streets would be set up to turn green on alternate cycles. If there was only a car then the walk signal would not activate and the green phase would end a couple of seconds after the last car left the detector loop. This usually resulted in a 5 s green phase. This only allows for a 5 m wide cross street for pedestrians using the one second per metre cross time.
If the main street is 3 lanes in each direction with right and left turn lanes that results in an intersection that has 8 lanes of traffic or is about 20 – 24 m wide. This means that the green phase has to be about 20 – 25 seconds to allow for pedestrians. If there is only one car waiting then it is a lot of wasted cross street time. When a car occupies the detector on the side street the walk phase on the main street has to be turned off and a safe amount of time allowed for a pedestrian who just entered the intersection to finish crossing it. If the car makes right hand turn before the flashing phase finishes then the walk signal comes back on and the cross street green is omitted. This is better than having a 5 s green phase for no traffic. If there is a button to push for the walk signal then push it and don’t assume it will happen automatically.
It used to be quite common for cross streets to be on a demand only cycle in off peak hours. This meant the green phase would last until someone triggered the green for the cross street. Putting these intersections into the standard cycle full time means that you can be the only car in sight and the cross street will get the green and you sit there for 30 – 40 s while nothing uses the cross street. It is things like this and really long cycle times that cause cars to run the red so they don’t have to “waste their valuable time waiting.”
In Asia many of the walk signs are animated LED’s and show the person actually walking. When you get to what would be the flashing hand the walking figure changes colour and starts to walk faster. Just before the green ends the figure is in an all out sprint. They also had count down timers for the red phase; it would tell you the number of seconds until the next green. The two outside my hotel were 181 and 205 seconds. As the timer got closer to zero you could hear the cars revving up and the squeal of tires when it hit zero. Not many people tried running a late amber.
Steve: Of course, living in a city where we can’t tell people to “walk left” for fear that this might generate lawsuits by escalator users (who are supposed to stand still), I am sure than a “sprinting man” graphic would have the legal beagles frothing at the mouth.
LikeLike
There needs to be a study done that use the “Principles of Unintended Consequences” Many of the changes that have been made that were supposed to increase safety and speed up traffic flow have had the opposite effect.
1) The introduction of the ALL RED phase was suppose to cut down on intersection collisions because there would time a dead time to let the intersections clear. What has happened is that drivers have interpreted the yellow light as a warning to speed up and get through the intersection and that the all red phase will protect them. This has resulted in more T bone accidents by motorists running reds.
2) The introduction of third high mounted brake light was to help prevent rear end collisions because you could see what the driver two in front was doing and brake sooner. What has happened is that drivers get closer to the car in front because they can see the car two in front and will brake when it does. According to a Dutch study it actually increased the number of rear end collisions.
3) The lengthening of traffic signal cycle time was supposed to speed up traffic flow because there would be less amber time which was time when traffic could not move in either direction. Amber time is about 4 seconds so there is 8 seconds in each cycle when no traffic would be moving. On an 80 second cycle it is 12.5% of down time. Going to a 160 second cycle reduced this by half to 6.25%. A consequence of this meant that when you stopped you were stopped for a lot longer. This caused motorists to become frustrated and since we had that all red phase for protection they would speed up and run the amber or red. This causes more collisions and ties traffic up more.
Traffic simulations seem to be designed for “well behaved” traffic flow which would obey all the rules. When this happens the above changes would be beneficial. Unfortunately traffic flow is not “well behaved”, especially the motorists, and things which should improve safety actually make it worse.
I remember in the early 70’s doing computer simulations that assigned mathematical formula and queuing theory to predict flow rates for different intersection timings. There was no input to allow for how it affected driver behaviour and how that in turn affected traffic flow. From empirical observation I believe that there would be an improvement in traffic flow and safety if a number of intersections were taken out of the computer controlled cycles in the off peak and went back to a strictly demand actuated signal. I have no statistical data or computer simulations to back this up, just a gut feeling.
LikeLike