Protected Left Turns - Beginning or End of Cycle?

Started by webny99, April 17, 2018, 09:17:10 AM

Previous topic - Next topic

webny99

Inspired by some comments in another thread, what are your thoughts on protected left turns?

When I say this, I mean when in the light cycle do you feel the protected phase should occur, at the beginning (before through traffic has their green) or the end (after through traffic has their green)? Whatever your opinion, explain your logic  :)

I can come up with advantages and disadvantages either way - and NYS uses both - but I'm not convinced yet as to which is preferred.


Rothman

In MA, people turn left at the start of the green light whether protected or not. :D
Please note: All comments here represent my own personal opinion and do not reflect the official position(s) of NYSDOT.

TheArkansasRoadgeek

As a passenger, I tend to like (and think) the turn (for traffic on my side of the intersection) to turn before the other's green.
Well, that's just like your opinion man...

tradephoric

I'm interested in this topic, but are we talking about protected only left turns or permissive-protected left turns?

SectorZ

Quote from: webny99 on April 17, 2018, 09:17:10 AM
Inspired by some comments in another thread, what are your thoughts on protected left turns?

When I say this, I mean when in the light cycle do you feel the protected phase should occur, at the beginning (before through traffic has their green) or the end (after through traffic has their green)? Whatever your opinion, explain your logic  :)

I can come up with advantages and disadvantages either way - and NYS uses both - but I'm not convinced yet as to which is preferred.

For some reason I prefer the left arrow at the beginning. Most intersections around me are like that, but there are still plenty that have it at the end. As long as the choice is the most practical for the traffic patterns, I honestly don't mind either way.

And I can confirm with Rothman as a Massachusetts resident that he is 100% correct.

webny99

Quote from: tradephoric on April 17, 2018, 10:02:52 AM
I'm interested in this topic, but are we talking about protected only left turns or permissive-protected left turns?

Well, both, although you could theoretically have a different answer for each type.

(At intersections where traffic flow is an issue, it's usually busy enough that permissive phases don't allow for much movement anyways...)

NE2

pre-1945 Florida route log

I accept and respect your identity as long as it's not dumb shit like "identifying as a vaccinated attack helicopter".

Brandon

IMHO, they should always be lagging lefts rather than leading lefts.  It allows vehicles to make their turns and then clear out the left turn lane afterward.
"If you think this has a happy ending, you haven't been paying attention." - Ramsay Bolton

"Symbolic of his struggle against reality." - Reg

jakeroot

I like when the signals are phased appropriately as part of a larger coordinated pattern. If this means lead-lead, lead-lag, or lag-lag, then so be it.

As a matter of practice, I prefer permissive lag-lag left turns. For one simple reason: leading pro/per lefts have two green-to-red phases (at the end of the initial protected phase, and at the end of the permissive phase). Lagging has one (at the end). This means less traffic has to come to a stop and then proceed again (if FYAs are being used), and all traffic can conceivably start moving at the beginning of the cycle (through traffic through completely, left turning traffic inching out into the middle), with the through phases ending appropriately to allow a protected left if necessary (rather than "as required" to allow through traffic to proceed after a leading left).

To quote Rick Perez, the head of traffic engineering in Federal Way, WA (a city with mostly lagging FYAs), "the best left‐turn phase is a skipped left‐turn phase".

UCFKnights

Quote from: jakeroot on April 17, 2018, 05:36:36 PM
I like when the signals are phased appropriately as part of a larger coordinated pattern. If this means lead-lead, lead-lag, or lag-lag, then so be it.

As a matter of practice, I prefer permissive lag-lag left turns. For one simple reason: leading pro/per lefts have two green-to-red phases (at the end of the initial protected phase, and at the end of the permissive phase). Lagging has one (at the end). This means less traffic has to come to a stop and then proceed again (if FYAs are being used), and all traffic can conceivably start moving at the beginning of the cycle (through traffic through completely, left turning traffic inching out into the middle), with the through phases ending appropriately to allow a protected left if necessary (rather than "as required" to allow through traffic to proceed after a leading left).

To quote Rick Perez, the head of traffic engineering in Federal Way, WA (a city with mostly lagging FYAs), "the best left‐turn phase is a skipped left‐turn phase".
I definitely agree with that quote. I personally don't like the lag-lag lefts as almost invariably, whenever I've been to one, there has been a significant amount of left green time for one side of the street having many more cars turning left then the other. My preference is for lead-lag, with the direction receiving less turning cars being the lag segment, as this should give the most likelihood of the phase being able to be skipped. If one direction has very little through traffic in general, I don't mind the lag being on the opposite site as that again may allow the phase to be skipped, but I always prefer lead-lag.

jakeroot

Quote from: UCFKnights on April 17, 2018, 09:02:50 PM
Quote from: jakeroot on April 17, 2018, 05:36:36 PM
I like when the signals are phased appropriately as part of a larger coordinated pattern. If this means lead-lead, lead-lag, or lag-lag, then so be it.

As a matter of practice, I prefer permissive lag-lag left turns. For one simple reason: leading pro/per lefts have two green-to-red phases (at the end of the initial protected phase, and at the end of the permissive phase). Lagging has one (at the end). This means less traffic has to come to a stop and then proceed again (if FYAs are being used), and all traffic can conceivably start moving at the beginning of the cycle (through traffic through completely, left turning traffic inching out into the middle), with the through phases ending appropriately to allow a protected left if necessary (rather than "as required" to allow through traffic to proceed after a leading left).

To quote Rick Perez, the head of traffic engineering in Federal Way, WA (a city with mostly lagging FYAs), "the best left‐turn phase is a skipped left‐turn phase".
I definitely agree with that quote. I personally don't like the lag-lag lefts as almost invariably, whenever I've been to one, there has been a significant amount of left green time for one side of the street having many more cars turning left then the other. My preference is for lead-lag, with the direction receiving less turning cars being the lag segment, as this should give the most likelihood of the phase being able to be skipped. If one direction has very little through traffic in general, I don't mind the lag being on the opposite site as that again may allow the phase to be skipped, but I always prefer lead-lag.

Typical practice, at least in Federal Way, is for the busier approach to use lagging phasing, and the lighter approach to use leading phasing. I don't know why this is, as your thought process makes more sense to me. I think it really has to do with when the platoon of cars arrive, which phasing is used.

tradephoric

Operationally, leading-lefts should be more efficient at an isolated intersection.  A plant shift at 7AM may cause a signal to run 40 second left turn phases, only to clear out once the shift begins.  Then 30 minutes later a school lets in and the left turn phase gets heavy again.  Traffic signals are constantly reacting to these sudden changes in traffic patterns.  With leading lefts, you can just allocate 40 second left turns for the entire rush and allow it to gap out early if nobody is there.  Any extra time is given to the critical through movement.  With lagging lefts, how does the signal know when the left turn is going to be heavy or not?  During the previous phases, the left turns were heavy so the signal is allocating 40 seconds for the left turns, but suddenly there are only 2 cars waiting to turn left now.  You are going to have a tremendous amount of waisted green time that goes to the side-street through movement as opposed to the critical main-street through movement once that lagging left turn gaps out.  That's not efficient.

Back in the mid-80s several Arizona cities switched from leading-left turns to lagging-left operation.  Almost across the board, driver delay increased when the cities switched to lagging-lefts. 

https://apps.azdot.gov/ADOTLibrary/publications/project_reports/PDF/AZ321.pdf

tradephoric

Here's a visual representation of why leading-left turns are more efficient than lagging-lefts.   Each model has the exact same geometry, splits, and traffic volumes... except one has leading-lefts and one has lagging-lefts.  Now many will say that the WBLT in the lagging-left model doesn't have to run that long... it can run shorter and split phase with the EBLT movement.  But watch the video in the context that the WBLT was really heavy, but it suddenly got light.  The controller thinks the WBLT still needs the full 40 seconds to clear traffic based on the previous cycles, but when it suddenly doesn't get any traffic you have a tremendous amount of wasted green time for that WBLT (and once that WBLT arrow comes on, you can't gap it out early and allow the EB through green to run again).

https://www.youtube.com/watch?v=v4l8YDY8deM&feature=youtu.be

https://www.youtube.com/watch?v=oRJd5wIz2pw&feature=youtu.be

TheArkansasRoadgeek

Quote from: tradephoric on April 18, 2018, 02:16:33 PM
Here's a visual representation of why leading-left turns are more efficient than lagging-lefts.   Each model has the exact same geometry, splits, and traffic volumes... except one has leading-lefts and one has lagging-lefts.  Now many will say that the WBLT in the lagging-left model doesn't have to run that long... it can run shorter and split phase with the EBLT movement.  But watch the video in the context that the WBLT was really heavy, but it suddenly got light.  The controller thinks the WBLT still needs the full 40 seconds to clear traffic based on the previous cycles, but when it suddenly doesn't get any traffic you have a tremendous amount of wasted green time for that WBLT (and once that WBLT arrow comes on, you can't gap it out early and allow the EB through green to run again).

https://www.youtube.com/watch?v=v4l8YDY8deM&feature=youtu.be

https://www.youtube.com/watch?v=oRJd5wIz2pw&feature=youtu.be
Now that I have seen a visualization, US 71 has lagging left turn timings from Greenwood, AR to Fort Smith. It works for the traffic volume.
Well, that's just like your opinion man...

jakeroot

Quote from: tradephoric on April 18, 2018, 12:22:15 PM
Operationally, leading-lefts should be more efficient at an isolated intersection.  A plant shift at 7AM may cause a signal to run 40 second left turn phases, only to clear out once the shift begins.  Then 30 minutes later a school lets in and the left turn phase gets heavy again.  Traffic signals are constantly reacting to these sudden changes in traffic patterns.  With leading lefts, you can just allocate 40 second left turns for the entire rush and allow it to gap out early if nobody is there.  Any extra time is given to the critical through movement.  With lagging lefts, how does the signal know when the left turn is going to be heavy or not?  During the previous phases, the left turns were heavy so the signal is allocating 40 seconds for the left turns, but suddenly there are only 2 cars waiting to turn left now.  You are going to have a tremendous amount of waisted green time that goes to the side-street through movement as opposed to the critical main-street through movement once that lagging left turn gaps out.  That's not efficient.

Back in the mid-80s several Arizona cities switched from leading-left turns to lagging-left operation.  Almost across the board, driver delay increased when the cities switched to lagging-lefts. 

https://apps.azdot.gov/ADOTLibrary/publications/project_reports/PDF/AZ321.pdf

I don't know much about lagging protected lefts. My experience with lag/lag lefts has only been at locations with permissive phasing. The only lagging protected lefts are double left turns with lead/lag phasing, which is usually part of a larger coordinated system.

Nonetheless, I can see your point. As far as I know, the local cities near me that use lag/lag operations (with permissive phasing) extend the through green time if there is no through traffic detected. This seems ass-backwards, but traffic turning left doesn't need a green arrow if there is no one coming. Basically, the time that most cities would dedicate to a left turn green arrow, these cities dedicate it to through traffic. Through traffic isn't coming, but the left turning traffic can see this, and is still able to turn. If another platoon of through traffic arrives, the light is still green and they can proceed as usual. If, after a very lengthy time, the controller still detects traffic waiting to turn left, the green arrow would be called up.

You may want to do some further research on Tucson. That city continues to use permissive lefts with lagging phasing (5-section signals too), unlike the metro Phoenix cities.

tradephoric

Quote from: TheArkansasRoadgeek on April 18, 2018, 03:51:07 PM
Now that I have seen a visualization, US 71 has lagging left turn timings from Greenwood, AR to Fort Smith. It works for the traffic volume.

Lagging-lefts aren't so bad when left-turn volumes are light.  At intersections with short left-turn pockets, lagging-lefts may be preferred as the few drivers waiting to turn left could potentially turn in gaps during the permissive phase allowing the lagging-left to skip entirely.  But for intersections with significant left-turn volumes, leading-lefts should push more traffic through the intersection. 

Quote from: jakeroot on April 18, 2018, 04:26:09 PM
I don't know much about lagging protected lefts. My experience with lag/lag lefts has only been at locations with permissive phasing. The only lagging protected lefts are double left turns with lead/lag phasing, which is usually part of a larger coordinated system.
Nonetheless, I can see your point.

Yeah the models were set up as protected only, but the same points can be made with permissive-protected lefts.  As webny99 said earlier, "˜where traffic flow becomes an issue, it's usually busy enough that permissive phases don't allow for much movement anyways'.

jakeroot

Quote from: tradephoric on April 18, 2018, 04:44:58 PM
Quote from: jakeroot on April 18, 2018, 04:26:09 PM
I don't know much about lagging protected lefts. My experience with lag/lag lefts has only been at locations with permissive phasing. The only lagging protected lefts are double left turns with lead/lag phasing, which is usually part of a larger coordinated system.
Nonetheless, I can see your point.

Yeah the models were set up as protected only, but the same points can be made with permissive-protected lefts.  As webny99 said earlier, "˜where traffic flow becomes an issue, it's usually busy enough that permissive phases don't allow for much movement anyways'.

Did you read the rest of my post? I addressed the issue, and how it could be solved with permissive phasing (beyond what you stated to TheArkansasRoadgeek).

Even when traffic flow is heavy, there's still the occasional gap, plus the clearance/all-red phase at the end.

Scott5114

#17
I prefer leading lefts, as that's what I personally tend to expect, but I'm in favor of whatever makes that particular intersection function most effectively.

What grinds my gears is when there is a street with heavy oncoming traffic and the signal is timed for an extensive permissive phase that does no good because there's no safe time to turn. You may as well make it a protected left, because a FYA with oncoming traffic is exactly as useful as a red arrow.

There's an intersection near me that often gives a FYA on a street with an AADT of 29,300. Thanks, I hate it.
uncontrollable freak sardine salad chef

tradephoric

Quote from: jakeroot on April 18, 2018, 04:53:14 PM
Did you read the rest of my post? I addressed the issue, and how it could be solved with permissive phasing (beyond what you stated to TheArkansasRoadgeek).

Even when traffic flow is heavy, there's still the occasional gap, plus the clearance/all-red phase at the end.

So i reduced the lagging-left to a minimum green in the model and gave all the extra time to the thru movement and sure enough all movements cleared with no problems.  I definitely favor more green time for the thru phase as opposed to the left turns when possible.  What good is it if the left turn pockets clear each cycle if it takes drivers 20 minutes to get to the intersection because the thru movement is backed up to the next county? 

As i said about that lagging-left model, watch it in the context that the WBLT was heavy for several cycles, but suddenly got light.  If that WBLT was really heavy, more green time would have been needed for the left turns because there isn't as many gaps in traffic for WBLT  (since the EB thru movement has much higher volumes than the WB thru in the model).  So you would need more than a minimum green for the left-turns in order to clear out the WBLT movement... but then suddenly the next cycle when it is light... all that green time dedicated for the WBLT is wasted that could have otherwise been given to the heavy EB thru movement.  This is a little long winded, but there are just so many scenarios where having leading-lefts are going to improve traffic flow compared to lagging-lefts (whether it be protected-only or permissive-protected). 

tradephoric

Quote from: webny99 on April 17, 2018, 09:17:10 AM
I can come up with advantages and disadvantages either way - and NYS uses both - but I'm not convinced yet as to which is preferred.

Webny99, i'm curious on your take on the advantages/disadvantages for each.  Are you considering mostly operational advantages or safety advantages?  So far it looks like most of the comments here have related to operational advantages. 

jakeroot

Quote from: Scott5114 on April 18, 2018, 06:52:15 PM
What grinds my gears is when there is a street with heavy oncoming traffic and the signal is timed for an extensive permissive phase that does no good because there's no safe time to turn. You may as well make it a protected left, because a FYA with oncoming traffic is exactly as useful as a red arrow.

Many jurisdictions account for at least two left turning vehicles during the all red clearance cycle. Some don't, which is why time of day phasing (what you seem to prefer) exists. I think they are termed "sneakers" (not that I agree with that term).

My issue with time of day phasing is that, even when a certain time of day is considered fully saturated, there's still the occasional cycle with a large gap. Maybe during certain cycles, you aren't able to turn. But then just go ahead and wait for the next green arrow. You don't need a red arrow to stop you from turning.

Quote from: tradephoric on April 18, 2018, 07:43:23 PM
Quote from: webny99 on April 17, 2018, 09:17:10 AM
I can come up with advantages and disadvantages either way - and NYS uses both - but I'm not convinced yet as to which is preferred.

Webny99, i'm curious on your take on the advantages/disadvantages for each.  Are you considering mostly operational advantages or safety advantages?  So far it looks like most of the comments here have related to operational advantages. 

I've read that lagging left turns in Tucson improved intersection safety. No idea why.

webny99

Quote from: tradephoric on April 18, 2018, 07:43:23 PM
Quote from: webny99 on April 17, 2018, 09:17:10 AM
I can come up with advantages and disadvantages either way - and NYS uses both - but I'm not convinced yet as to which is preferred.

Webny99, i'm curious on your take on the advantages/disadvantages for each.  Are you considering mostly operational advantages or safety advantages?  So far it looks like most of the comments here have related to operational advantages.

Part of the reason for this thread was so I could learn a bit more while developing my take. In fact, I wasn't even familiar with "lead/lag" terminology until now (hence the awkward thread title), so take my take with at least a few grains of salt:

Like Scott, leading lefts are what I tend to expect by default, and the busier the road, the more this holds true. Leading lefts will occur automatically if there's anyone waiting, and I see that as a good thing for traffic flow, because it reduces the chance of traffic backing up and disrupting through traffic - turning traffic can always clear at a dependable interval (ie, flow is not as dependent on opposing traffic volume). This becomes a problem, though, when volumes exceed what the protected phase can accomadate, especially when it's protected-only.

Lagging lefts are used most often around here at freeway on-ramps, and I think that's where they're most effective, since only one direction has an extended red phase (and the off-ramp is irrelevant - same aggregate wait time for them regardless). The interesting thing about lagging lefts is that the queue is actually going to be longest during the green phase - and then (hopefully) clear out and leave an empty intersection. So they can be more efficient from that perspective, as the protected phase can be extended as needed (in most cases). I see a clear intersection when the light turns red as optimal, and lagging lefts can be helpful to achieve this during peak times. Lagging lefts also reduce the chance that cars will have to stop for a red light. With good timing, all cars should approach and clear the intersection in the same cycle. This doesn't usually happen with leading lefts - you're almost guaranteed to wait through a red cycle if there's a queue of any description, but then you can proceed  first after that red cycle.

In a word, leading is more dependable, lagging is more adaptable. In the middle of the night or at an off-peak time, I'll take leading all the way. But if it's rush hour and a wait is unavoidable, lagging seems like the better bet for reducing net wait time (and getting through within one cycle).




The safety aspect isn't one I've given much thought to. I thought it was an interesting comment that leading has two stop-go cycles whereas lagging only has one. That seems to have positive implications for lagging - since drivers know they'll have a chance later, they aren't going to run the red like they might at the end of the leading-protected phase. But if said leading left is long enough, that eliminates the issue anyways.

So, as to what's preferred, I guess I'd really have to make a case for a specific intersection - as I said above, traffic volumes play a key role and can easily make or break it. And I'm skeptical of committing to full-time defense of one vs. the other :-P

jakeroot

Quote from: webny99 on April 18, 2018, 08:51:54 PM
The safety aspect isn't one I've given much thought to. I thought it was an interesting comment that leading has two stop-go cycles whereas lagging only has one. That seems to have positive implications for lagging - since drivers know they'll have a chance later, they aren't going to run the red like they might at the end of the leading-protected phase. But if said leading left is long enough, that eliminates the issue anyways.

I actually brought that up earlier. With a leading left, there are two points where drivers have to make a decision to stop or go due to a yellow arrow. Lagging lefts only have one of these, so in theory, then there's less decision-making and, again theoretically, less risk.

I seem to recall a thread a while back about an issue with flashing yellow arrows' solid yellow arrow phase. With a leading left, driver's individually face two solid yellow arrows, but they actually mean different things. If it's at the end of a protected phase, drivers are not approaching you, and your decision to stop is based solely on whether or not you want to. But, if the solid yellow is following the permissive phase, traffic is actually coming at you full speed. The decision is then whether or not you can go, not whether or not you want to. In practice, this is pretty hard to screw up. You would have to have not been paying any attention. But it's an oddity that I'm not sure that many people have considered.

johndoe

Interesting discussion.  Thanks for the links.  I guess my take is that there is no global "right" or "wrong" answer.  The signal timing I've done has all been as part of a coordinated system, so lefts will be led or lagged in order to provide two-way progression for the throughs.  As cycle lengths change (for busy and lighter times of day) the same movement may flip from a leading to lagging movement.  If you globally mandated lefts led or lagged, you'd be punishing the throughs, increasing the times they stopped along the corridor.

For permissive lefts I do like the "third car actuation" mentioned in that Arizona report as it can cut down on the number of phases, and therefore lost time.

In the case of one lone signal, engineers don't spend as much time optimizing the timing.  If it's running free you're not going to drastically change the operations by using alt sequences anyway (assuming the max times are reasonable).

Quote from: tradephoric on April 18, 2018, 07:39:07 PM
I definitely favor more green time for the thru phase as opposed to the left turns when possible.  What good is it if the left turn pockets clear each cycle if it takes drivers 20 minutes to get to the intersection because the thru movement is backed up to the next county? 
To play devil's advocate: what good is it if the through lanes get plenty of green if left-turns experience queue spillback?  The throughs would eventually experience "starvation" if the lefts spillback.  It goes both ways.  In my experience it's much more likely that a left-turn lane blocks a through lane than vice versa. 

Quote from: tradephoric on April 18, 2018, 07:39:07 PM
but then suddenly the next cycle when it (WBLT) is light... all that green time dedicated for the WBLT is wasted that could have otherwise been given to the heavy EB thru movement.
True, and this is a danger of lagging lefts...but there are a few things to combat that.  First, the engineer could allow phase reservicing.  This would allow the EBT to come back on after the WBL gapped out.  Second, keep in mind the time wasted (on WBL) is due to the other ring (EBL).  In your example the steady stream of EBL traffic is holding the WBL green.  In real life, that EBL storage is probably shorter, and the EBT queue would eventually block EBL.  Once EBL and WBL both gapped out the barrier would be jumped and the sidestreet gets served.

Jet380

In my area, it's almost all leading turns. The benefit I see in leading turns, is that it can avoid issues that might arise if the turning lane is too short. If there are so many cars waiting to turn that they spill out into the through lane, a leading turn phase gets them through the intersection and clears the way for through traffic.

Also, if FYA isn't a thing in the area (and it definitely isn't around here :-/), leading turns are the only option if you want the turn phases to differ in length based on demand without causing a yellow trap.



Opinions expressed here on belong solely to the poster and do not represent or reflect the opinions or beliefs of AARoads, its creators and/or associates.