What is "perfect" signal timing?

[webny99]

What do you think constitutes "perfect" signal timing?

In my opinion cars should miss the first light after they turn onto a main road. They wait at the next red while thru traffic approaches. Then they become part of the thru traffic, and are in sync to make all green lights thereafter. Obviously this is in a perfect world, and may not be achievable for any number of other factors.

[hotdogPi]

Michigan does it pretty well, as do the one-way avenues in New York City. It doesn't exist where I live.

And yes, the first light should be missed; it's green for people going straight, not people coming from a turn.

[Scott5114]

A mythological creature, akin to a unicorn or dragon. Much like the other two, it cannot be found in Oklahoma.

[Brandon]

Extinct in most of Illinois (downtown Ottawa has it), it is fairly common in most of Michigan.

Examples:

[bzakharin]

Oh man, I wish NJ 73 had signal timing like that. It would truly shave 10-15 minutes from my commute, and I would never need to use I-295 instead.

[tradephoric]

Here is a time-distance for Woodward Avenue in Detroit:


Achieving good signal progression in both directions of travel along a 2-way road comes down to three main variables.... signal spacing, cycle length, and speed.  The posted speed of Woodward is 45 mph and the signals are spaced roughly 3,000 feet apart.  If the speed and signal spacing along a corridor stays consistent you can figure out the cycle length that will achieve good signal progression in both directions.  I put together a chart that figures out the resonant cycles for different speeds and signal spacing.  In the case of Woodward the signals need to run roughly 90 seconds to achieve good signal progression in both directions of travel (the corridor is actually timed for about 48 mph, but most people are going a few miles over the speed limit anyways so it works out well).

[kphoger]

Extinct in most of Illinois (downtown Ottawa has it), it is fairly common in most of Michigan.

Does North Avenue no longer have it out in the western 'burbs?

[Brandon]

Extinct in most of Illinois (downtown Ottawa has it), it is fairly common in most of Michigan.

Does North Avenue no longer have it out in the western 'burbs?

No, not since North Avenue was rebuilt a few years ago.  Too many dual left turns along it from IL-59 east to IL-83.  Sad, IDOT had a chance to emulate MDOT (see the videos above) and blew it.

[jeffandnicole]

Oh man, I wish NJ 73 had signal timing like that. It would truly shave 10-15 minutes from my commute, and I would never need to use I-295 instead.

Hurting NJ 73 (and most NJ roads) is the jughandles.  It would be fine if every intersection has jughandles, or every intersection has left turn lanes.  But when there's a combination of the 2, it's impossible to have the cycle lengths consistent.  At intersections such as 73 and Church Road, extra time needs to be devoted to the cross streets, which takes away time from 73.  An eventual overpass will solve some of that problem, but that's years down the road...

[Rick1962]

A mythological creature, akin to a unicorn or dragon. Much like the other two, it cannot be found in Oklahoma.

Ugh. Tulsa has signal un-timing down to a fine art. Not only do the vast majority of them work independently, there are places where it looks like a traffic signal salesman came through and made them a deal.

Yale avenue between 61st and 71st Streets: Signals at 61st, St. Francis Hospital entrance, 65th, 66th, 68th, and 71st Streets. All operate independently of the others. 

41st Street between Yale Avenue and Sheridan Road: Signals at Yale, Darlington, Hudson, I-44 frontage roads, and Sheridan. Again, all operate independently and cause tremendous backups along this stretch. Oh, and this is the "improved" setup following the rebuilding of I-44 a few years ago. 

Apologies for a bit of topic creep, but the subject is a big sore spot with me, living here in the land progress forgot.

[tradephoric]

IMO, there are 3 main reasons why most roads are incapable of good dual signal progression.

#1.  Irregularly spaced traffic signals:  If the signal spacing between traffic signals is constantly changing along the corridor, then the cycle length that achieves perfect dual progression is constantly changing too.  If a set of lights has a perfect dual progression at a 90 second cycle, but the next set of lights has a perfect dual progression at 130 second cycle, what cycle do you time the corridor for?   Either you compromise and pick the cycle length that works best for the entire corridor or you break it up where some of the signals run different cycles altogether.

#2.  Signals are too closely spaced:  The most obvious example of this would be a two-way street in Manhattan like Broadway.   While the traffic signals are evenly spaced, they are only 260 feet apart.  Anybody driving Broadway intuitively knows that getting a string of green lights in both directions is impossible.  As speed of the corridor increase, the distance between traffic signals also must increase to be able to maintain dual progression.  The Woodward signals are spaced about 3000 feet apart and the speed limit is 45 mph.  Even along a Michigan left corridor that eliminate the need for left turn phases, the 90 second cycle just barely fits along Woodward.  If the signals were spaced any closer (or the speed limit was set any higher), the dual signal progression would be lost. 

#3.  Pedestrian crossings are too long: A side-street pedestrian crossing of 150 feet would need to run 52 seconds to adhere to the MUTCD and safely fit the pedestrians (there are literally dozens of major Orlando intersections with side-street crossings this long).  In general you want your main-street to run just as long, if not longer, than your side-street.  Assuming the main street also runs 52 seconds, the cycle length is up to 104 seconds.  But we haven't accounted for left turn phases yet.  Add 20 seconds for both left turn phases and the cycle length is creeping up to 150 seconds.  Maybe a corridor in Orlando would achieve perfect dual progression at 100 seconds, but since the pedestrian crossing distances are so long it really can't run that short of a cycle.  Technically you can use pushbuttons and only run the side-street the full 52 seconds when the pushbutton is pressed, but then the signal will get out of step with the surrounding signals every time a pushbutton is actuated.

[bzakharin]

Oh man, I wish NJ 73 had signal timing like that. It would truly shave 10-15 minutes from my commute, and I would never need to use I-295 instead.

Hurting NJ 73 (and most NJ roads) is the jughandles.  It would be fine if every intersection has jughandles, or every intersection has left turn lanes.  But when there's a combination of the 2, it's impossible to have the cycle lengths consistent.  At intersections such as 73 and Church Road, extra time needs to be devoted to the cross streets, which takes away time from 73.  An eventual overpass will solve some of that problem, but that's years down the road...

Why do they do that anyway? How do they decide on jughandle vs regular intersection?

[Scott5114]

Here is a time-distance for Woodward Avenue in Detroit:

I don't think any of us has any clue how to read this graph. I've seen several of them posted without comment from others, which I think is because none of us know what we're looking at here. It's like reading a weather balloon sounding if you've never done such.

IMO, there are 3 main reasons why most roads are incapable of good dual signal progression.

You missed the biggest (and most common) reason:

#0. The controlling agency shows no motivation toward achieving good signal progression. Signal progression doesn't magically happen on its own; the agency must actively work toward it. If there's no buy-in from the agency, you won't have interconnected signals and you won't have any attempt to space the signals to work toward it. Even if the road is otherwise a perfect candidate for good signal progression, someone has to be assigned to come up with the timings and program them into the controllers. That doesn't happen unless the higher-ups make it happen.

You could get a decent signal progression going along State Highway 9 in Norman, Oklahoma, but neither ODOT nor the City of Norman (not sure who's responsible for the signals) are inclined to make it happen. So we all suffer in traffic.

[7/8]

^ I learned about those graphs in school last semester. The x axis is time, while the y axis is distance along the road. Green, yellow, red are the phases for each intersection on the road. The blue lines represent southbound cars, while the red lines represent northbound cars. Basically, each blue and red line represents the movement of a car along the road. If it passes through a green, then the line will continue straight. If the line hits a red light, it will go to the right until the green phase begins (which represents waiting for the signal). Since this example shows most of the lines (cars) passing through multiple greens without stopping, it shows that the signal progression is good.

[GaryV]

I will agree that many major thoroughfares in Michigan have timing down pretty well (until traffic volumes increase so much that the roads can't handle it in general).  Many roads also have "smart" traffic lights that monitor the traffic flow and adjust the timing.

But in smaller towns and suburbs, sometimes they set the lights so you have to stop for most of them, as "calming".  Until you realize that if you go about 10 or 15 mph over the limit, you make the lights!

[tradephoric]

You missed the biggest (and most common) reason:

#0. The controlling agency shows no motivation toward achieving good signal progression. Signal progression doesn't magically happen on its own; the agency must actively work toward it. If there's no buy-in from the agency, you won't have interconnected signals and you won't have any attempt to space the signals to work toward it. Even if the road is otherwise a perfect candidate for good signal progression, someone has to be assigned to come up with the timings and program them into the controllers. That doesn't happen unless the higher-ups make it happen.

There can always be improvements with signal timings along a corridor, but ultimately the geometry of the roadway dictates how well a corridor can be timed.  When you get to the situation where you have 7 traffic signals along a 1 mile stretch of road, throwing any timing into the signals will work almost as well as a totally optimized timing (since the optimized timing will still stop a lot of cars at red lights simply due to how closely the traffic signals are spaced.. and how many there are).   

You could get a decent signal progression going along State Highway 9 in Norman, Oklahoma, but neither ODOT nor the City of Norman (not sure who's responsible for the signals) are inclined to make it happen. So we all suffer in traffic.

The spacing of the signals on State Highway 9 east of I-35 isn't great.  The signals aren't evenly spaced and they are spaced too closely together.  You could get good progression in one direction, but the other direction will suffer.   The problem is cities don't care how a proposed traffic signal will affect signal progression along major corridors. They just want the tax base of major developments in their city.  If a strip mall warrants a traffic signal at the entrance, they'll just put it in even if there are already several other traffic signals in the surrounding area.  What you get is drivers stopping at every other red light.  And in reality there aren't that many traffic signals along State Highway 9, but there are enough to screw up progression.  I wouldn't be surprised if you get stopped at McGee Drive often and that's the light you curse out the most.

[hotdogPi]

I mentioned this in the "single ideas" thread, but if you can't time it for both directions, what about timing traffic lights in one direction, and then in the other direction for a parallel or near-parallel road?

The example I used was MA 114 and MA 62 between I-95 and MA 128, but I don't think most people are familiar with those roads.

[Jardine]

Lightly traveled side streets on main thoroughfares with lights that only trigger if someone pulls up on the side street are a good idea EXCEPT for folks on the main thoroughfare who turn off onto the lightly traveled side street and swing wide making the turn and thus tripping the lights for NOBODY.

You dirty bastards !!!

I hate you I hate you I hate you !!!   I want to hurt you, and keep on hurting you.  From hell's black depths I will expend my last breath HATING you !!!

[tradephoric]

Lightly traveled side streets on main thoroughfares with lights that only trigger if someone pulls up on the side street are a good idea EXCEPT for folks on the main thoroughfare who turn off onto the lightly traveled side street and swing wide making the turn and thus tripping the lights for NOBODY.

You dirty bastards !!!

I hate you I hate you I hate you !!!   I want to hurt you, and keep on hurting you.  From hell's black depths I will expend my last breath HATING you !!!

LOL tell us how you really feel!  They can put a few second delay on the detector so the call doesn't go to the controller until someone is physically stopped over the detector for a few seconds.  This is usually enough to solve the problem of a wide swinging vehicle from tripping the detector.   

[Scott5114]

I wouldn't be surprised if you get stopped at McGee Drive often and that's the light you curse out the most.

Imhoff, Chautauqua, and Jenkins tend to be the worst offenders in my memory. I tend to be going westbound during periods of heavy traffic; when I come back eastbound it's late enough that the signals are all camera-actuated.

[mrsman]

I mentioned this in the "single ideas" thread, but if you can't time it for both directions, what about timing traffic lights in one direction, and then in the other direction for a parallel or near-parallel road?

The example I used was MA 114 and MA 62 between I-95 and MA 128, but I don't think most people are familiar with those roads.

I wish this would be implemented.  In Los Angeles, there was a plan about 10 years ago that was floated to make two wide streets, Olympic and Pico, each one-way in order to improve traffic.  One-way traffic could implement progression, utilize the existing left turn lane for driving, and be generally more efficient.  This was shot down because it was thought to lead to speeding, it would complicate transit along those routes, and the routes were too far apart from each other (typical one-way pairs are about 1 block apart, these distance between these two streets being about 1/4 -1/2 mile at times).

But if they instead kept both streets two-way but had the progression of each street differ, this would have made a great impact.  In this part of LA, traffic is about equal eastbound and westbound at rush hour.  In fact, traffic is usually worse heading away from Downtown in the mornings.  But this was never implememnted.

Another good candidate: Adams/Washington near the 10 freeway from Culver City to Downtown.

[mrsman]

I will agree that many major thoroughfares in Michigan have timing down pretty well (until traffic volumes increase so much that the roads can't handle it in general).  Many roads also have "smart" traffic lights that monitor the traffic flow and adjust the timing.

But in smaller towns and suburbs, sometimes they set the lights so you have to stop for most of them, as "calming".  Until you realize that if you go about 10 or 15 mph over the limit, you make the lights!

In Metro Detroit, I'm familiar with this phenomenon working on Woodward and on Telegraph, as I've seen videos for drivers along both streets cruising along.  Does it work on most other N/S streets?  The E/W streets?

[vdeane]

Plus some places just don't want coordinated signals.  Saratoga Springs bought an expensive signal timing system a few years ago and scrapped it after business complained about people not stopping in front of their stores any more.  And there's a light I pass by every single day that I swear only exists for "traffic calming" purposes.

[Rothman]

Plus some places just don't want coordinated signals.  Saratoga Springs bought an expensive signal timing system a few years ago and scrapped it after business complained about people not stopping in front of their stores any more.  And there's a light I pass by every single day that I swear only exists for "traffic calming" purposes.

I remember and am still waiting for the coordinated system on Central Ave that was touted by John Poorman at CDTC.  He used to boast that you would be able to drive between Albany and Schenectady without stopping and it would be the most innovative project in the country (used to joke that he loved The Jetsons way, way too much; pie-in-the-sky is an underestimated description of his "visions").

Every time the Waterman Ave light stops Central Ave traffic to let one car turn left, I think of him and laugh.

[tradephoric]

The mission statement of the National Association of City Transportation Officials (NACTO) is to build cities as places for people, with safe, sustainable, accessible and equitable transportation choices that support a strong economy and vibrant quality of life.  They put together an interesting website that talks all about signal cycle lengths and how they affect pedestrians.  Here is a snippet:

Though often invisible to the public, traffic signal cycle lengths have a significant impact on the quality of the urban realm and consequently, the opportunities for bicyclists, pedestrians, and transit vehicles to operate safely along a corridor.

Long signal cycles, compounded over multiple intersections, can make crossing a street or walking even a short distance prohibitive and frustrating. This discourages walking altogether, and makes streets into barriers that separate destinations, rather than arteries that stitch them together.
https://nacto.org/publication/urban-street-design-guide/intersection-design-elements/traffic-signals/signal-cycle-lengths/

NACTO is promoting short cycle lengths of 60-90 seconds so that pedestrians can cross the street without waiting too long.  But along major suburban corridors you see cycle lengths double that of 120-180 seconds, especially during rush hour.   Regardless of how cycle lengths affect pedestrians, I believe shorter cycles can push more traffic through a signal than longer cycles.  That goes against conventional wisdom, as short cycle lengths increased the % of lost time at the signal (ie. yellow and all red times soak up more of the pie).  But with longer cycles, the saturation flow rates reduces as the phase time increases.    Previous research that has been done on cycle lengths suggests a 45 second phase time optimizes the throughput of a saturated phase.  After 45 seconds, traffic begins to spread out and not utilize the green time anyways.  So for a 2-phase intersection, a 90 second cycle would theoretically be the optimal cycle length (which is what Woodward Avenue runs 24/7).   

Since Woodward runs a pedestrian friendly cycle length that optimizes throughput, could it be considered the "perfect" suburban corridor?  Well not so fast.  Here is another snippet from the NACTO website:

While short cycle lengths are desirable, ensure that cycle lengths are long enough for pedestrians to cross wide streets in a single leg without getting stuck in the median, unless the median is a destination in and of itself. 

The problem is pedestrians crossing Woodward get stuck in the median.  If you re-time the lights so they could make it all the way across, the cycle length would balloon to about 180 seconds.  But now pedestrians who don't want to wait long to get a Walk are waiting at a signal with a 180 second cycle length... Houston, we have a problem.