Are diverging diamonds a fad?

[tradephoric]

Trade, I don't think any of us are saying signal timing isn't important.  But at many service interchanges the through movement isn't the highest volume.  Traffic signals are always going to delay someone...but if the through movement is half the volume of those turning left onto the freeway, why give the through movement priority at the expense of the majority of drivers?

You just have more options with a signal that only stops one direction.  Assume the traffic volumes exiting one of the off-ramps during the PM rush is heavier than the thru traffic traveling on the arterial.  Go ahead and time the heavier off-ramp to coordinate with the main signal.  This will still only effect one direction of the corridor with a Parclo B4 (since the off-ramp signal being coordinated only stops one-direction).  If you wanted to coordinate a heavy Parclo A4 off-ramp, you would be effecting both directions along the corridor.  Closely spaced signals that stop both directions of travel leads to poor signal progression, regardless of what movement is being coordinated.

[kphoger]

why give the through movement priority at the expense of the majority of drivers?

So their eleven-light green streak doesn't get bummed, of course. It's just... such a downer!

[rickmastfan67]

Tradephoric, can you show me what a signal that only stops one direction of traffic looks like in one of your time/distance diagrams? I want to try to get the visual language right.

Here is an example of a Parclo B4 interchange near Miami, Florida with a corresponding time-distance diagram.


green circle icons = traffic signals that stop only one direction of travel
red circle icons = traffic signals that stop both directions of travel

Problem here tradephoric, is that you completely skipped a traffic light there.  You missed the light between the intersections of Dykes Road/Sheridan Street and the I-75 SB onramp light.
http://goo.gl/maps/LusIa

[tradephoric]

Rickmastfan67, I was responding to VTK's request....

Tradephoric, can you show me what a signal that only stops one direction of traffic looks like in one of your time/distance diagrams? I want to try to get the visual language right.

Included in the Parclo B4 time-distance are signals that stop both directions of traffic in addition to signals that stop only one direction of traffic.  You should be critical of me for including Dykes Road and 148th Lane since VTK only asked for signals that stop just one-direction of traffic.  For that, i apologize. 

[tradephoric]

why give the through movement priority at the expense of the majority of drivers?

So their eleven-light green streak doesn't get bummed, of course. It's just... such a downer!

Most drivers would appreciate cruising along a corridor without getting stopping at a red light.  Your mocking and cavalier attitude towards good signal progression would not be appreciated by the general public.  Why would you want to unnecessarily increase drivers delay and number of stops? 

Some ideologues want to make driving such a living hell that people will either decide to move closer to work or take other modes of transportation.  I'm not suggesting we build 10-lane roads everywhere, but the limited lane miles that are available should be efficient.  For all i know maybe you bike to work everyday and don't care about signal progression, but a lot of people do.  One may start to believe you're a wacked out ideologue based on some of your recent comments.

Maybe i am over zealot when it comes to designing corridors that achieve good signal progression.  Of course, poll 1,000 drivers and ask them if they enjoy getting stopped at red lights.

[jeffandnicole]

why give the through movement priority at the expense of the majority of drivers?

So their eleven-light green streak doesn't get bummed, of course. It's just... such a downer!

Most drivers would appreciate cruising along a corridor without getting stopping at a red light.  Your mocking and cavalier attitude towards good signal progression would not be appreciated by the general public.  Why would you want to unnecessarily increase drivers delay and number of stops? 

Some ideologues want to make driving such a living hell that people will either decide to move closer to work or take other modes of transportation.  I'm not suggesting we build 10-lane roads everywhere, but the limited lane miles that are available should be efficient.  For all i know maybe you bike to work everyday and don't care about signal progression, but a lot of people do.  One may start to believe you're a wacked out ideologue based on some of your recent comments.

Maybe i am over zealot when it comes to designing corridors that achieve good signal progression.  Of course, poll 1,000 drivers and ask them if they enjoy getting stopped at red lights.

It is extremely rare that an interchange ramp will produce more traffic than the cross street.  But even then, you can't account for a green corridor for traffic coming off limited-access highways where in theory they could go on an endless loop without hitting a traffic light. If they go up the ramp and have the green, great.  But if it's red, so be it.  After that the lights can be timed to hit as many green lights as possible.

You can't achieve an infinity length of green lights.  Even on the videos shown above, the corridors of green are shown for a certain point.  The camera operator makes it a point to make a U-turn just before a red light. 

[vtk]

Maybe i am over zealot when it comes to designing corridors that achieve good signal progression. 

You do seem just a bit more concerned with that one aspect of efficiency than the rest of us.  But, please & by all means, redesign some corridors in Columbus with good two-way signal progression.  Here are some that make the most sense as through corridors over significant distances:

• Cleveland Ave
• Hayden Run Rd, Bethel Rd
• Morse Rd
• Riverside Dr
• Fifth Ave
• Parsons Ave
• Frank Rd
• Refugee Rd
• Harrisburg Pk
• Broad St (particularly on the east side)
• Westerville Rd
• Powell Rd, Polaris Pkwy

I'm of the opinion that good two-way signal progression is essentially a lost cause on these corridors.  Prove me wrong.

[tradephoric]

Maybe i am over zealot when it comes to designing corridors that achieve good signal progression. 

You do seem just a bit more concerned with that one aspect of efficiency than the rest of us.  But, please & by all means, redesign some corridors in Columbus with good two-way signal progression.  Here are some that make the most sense as through corridors over significant distances:

• Cleveland Ave
• Hayden Run Rd, Bethel Rd
• Morse Rd
• Riverside Dr
• Fifth Ave
• Parsons Ave
• Frank Rd
• Refugee Rd
• Harrisburg Pk
• Broad St (particularly on the east side)
• Westerville Rd
• Powell Rd, Polaris Pkwy

I'm of the opinion that good two-way signal progression is essentially a lost cause on these corridors.  Prove me wrong.

There are 22 signals along a 5-mile section of Broad St (from Nelson Road to Boe Bixby Road).  All 22 signals stop both directions of traffic and the signals are irregularly spaced.  This is what an optimized time-distance diagram looks like on Broad Street:


There are 30 signals along a 5-mile section of Big Beaver Road (from Coolidge to Dequindre).  Only 5 signals stop both directions of traffic and the signals are evenly spaced.  This is what an optimized time-distance diagram looks like on Big Beaver:


https://www.youtube.com/watch?v=eB2H4bGp4Jc


Moral of the story, if you want a corridor to achieve good signal progression reduce the number of signals that stop both directions of traffic.  Traffic signals at DDIs, SPUIs, and Parclo A4s stop both direction of traffic, which is why they are so bad for progression. 

[kphoger]

why give the through movement priority at the expense of the majority of drivers?

So their eleven-light green streak doesn't get bummed, of course. It's just... such a downer!

Most drivers would appreciate cruising along a corridor without getting stopping at a red light.  Your mocking and cavalier attitude towards good signal progression would not be appreciated by the general public.  Why would you want to unnecessarily increase drivers delay and number of stops? 

My point is that I remain unconvinced that stopping SOME drivers means the interchange as a WHOLE is unacceptable. It's ingrained in our psyche to think that, if a traffic control design negatively affects one movement of traffic (generally the movement we happen to be a part of), then the design must be inefficient. This is a nearsighted view of things. Your thesis may very well be correct, that DDIs and certain parclo variants may be less efficient than other designs, but I cannot accept mere interruption of signal progression along one of the corridors as the basis for that claim. It's only one piece of a puzzle, and not every puzzle even looks the same.

The bare fact of the matter is that ALL traffic control "increases drivers' delay" and all stoplights increase "number of stops". The key word, however, is "unnecessarily". Unnecessary for what? for whom? for what reasons?

[Bickendan]

Again, you're not looking at why a SPUI, DDI or A4s are getting put in places of conventional interchanges: They're not for the through traffic on the crossroad. They're there to get traffic off and on the highway the crossroad's interchanging with. Yes, there will be through traffic going through a DDI/SPUI/A4; yes they will be delayed by less than 'optimal' signal timing because of these interchanges, but they're not the target of the interchange.

Furthermore, it is impossible to provide free-flow traffic signal progression on a two-way corridor. The direction that enjoys free-flow will result in poor timing in the other direction. Only one-way grids can avoid that, and even then it doesn't always work.

[tradephoric]

Furthermore, it is impossible to provide free-flow traffic signal progression on a two-way corridor. The direction that enjoys free-flow will result in poor timing in the other direction. Only one-way grids can avoid that, and even then it doesn't always work.

Let me know if you don't understand how to read a time-distance diagram. The dash cam video demonstrates good two-way progression along a two-way corridor. 


https://www.youtube.com/watch?v=eB2H4bGp4Jc

[vtk]

There are 22 signals along a 5-mile section of Broad St (from Nelson Road to Boe Bixby Road).  All 22 signals stop both directions of traffic and the signals are irregularly spaced.  This is what an optimized time-distance diagram looks like on Broad Street:

Can it get any better than that on Broad St?  What if some of the signals were changed so they don't stop both directions of traffic at the same time?  There's no room for median U-turns so left turns have to be accommodated somehow; I'm thinking lead/lag operation, for example the following phase sequence: cross traffic left turns, cross traffic thru, eastbound thru + left, eastbound & westbound, westbound thru+left.  The signals at minor streets can be set up that way (or with eastbound & westbound reversed) to try to work around the waves of traffic set up between the more significant intersections, which will probably have eastbound and westbound stopped at the same time, and let's be honest, a greater proportion of red time because there's more cross traffic.  Anyway, does it get much better using that approach?

[tradephoric]

Here is Broad St with lead/lag optimization.  It's a bit disjointed but lead/lag will definitely help with progression.  Some signals will lead, some will lag, and some will lead/lag based on what SYNCHRO calculates to be the lowest delay.

[vtk]

Here is Broad St with lead/lag optimization.  It's a bet disjointed but lead/lag will definitely help with progression.  Some signals will lead, some will lag, and some will lead/lag based on what SYNCHRO calculates to be the lowest delay.

Hey that's a lot better than I expected, though the eastbound wave gets collectively screwed at Cassady through Roosevelt and the westbound wave gets collectively screwed at Napoleon through James and potentially again at Drexel.  And Columbus drivers aren't used to so many lagging left operations (most typical is leading left for both directions, which doesn't appear to be the case at any signals in your diagram).  If the corridor runs this well, then signal progression is indeed a valid (if not overriding) concern in choosing an interchange design.  However, the other half of my argument is that the Parclo B4 is not the only design that can be friendly to signal progression.  I'll be posting more on that shortly.

[vtk]

Imagine an idealized Detroitoid grid of arterials spaced one mile apart, with Michigan lefts.  With signal periods of 144s, traffic can cruise along at 50 MPH indefinitely.  It's ideal two-way progression.  A place like this stands to lose the most from poor interchange design, at least in the context of signal progression.

Now let's throw in some interchanges that aren't parclo B4s.

Parclo A4


This interchange has a "wide stance", that is, its exit ramp termini are quite far (3168ft) apart.  The signals at this interchange don't interrupt the green waves on the arterial at all.  The green time for the exit ramps is nearly 40 seconds (actually about 43s if you also count the yellow) out of the 144s cycle, which should be plenty if there exiting traffic is small compared to the arterial through traffic.

SPUI

This one requires the freeway to be a bit off-center in the mile-grid.  The signal operates in a lead-lag manner: westbound thru + left onto the freeway (plus northbound exit right to eastbound), eastbound & westbound thru, eastbound thru + left onto the freeway (plus southbound exit right to westbound), then left turns off of the freeway ramps.  In the timing configuration shown, the thru green waves on the arterial aren't interrupted at all; left turns onto the freeway and right turns off the freeway get nearly 40 seconds of green; left turns off the freeway get nearly 30 seconds of green.  The timing can be tweaked to allow more time for turning movements, at some expense to the thru green waves on the arterial.

Diverging Diamond
Example 1

The freeway is back to the middle of the mile again.  This example is designed to preserve the thru green waves, but about 20% of the wave gets delayed for a few seconds at the DDI.  Traffic turning off the freeway potentially gets delayed for a whole signal cycle between the DDI itself and the first major intersection on the surface arterial.  Traffic turning onto the freeway experiences no delay.

Example 2

Another way to time the DDI that is less focused on the thru arterial traffic, because DDIs are usually chosen when the exiting and entering traffic is significantly greater than the thru traffic staying on the surface arterial.  Here I've also shown cars that just exited the freeway as blue lines, and the lines turn red where they represent queuing vehicles.  Only about 40% of the thru green wave gets by unscathed, and it's probably less than that due to delays due to slowing down for the DDI's unconventional geometry.  Traffic turning right onto the freeway still experiences no delay, and traffic turning left onto the freeway experiences the same amount of delay as thru traffic.  Traffic exiting the freeway has slightly reduced delay compared to Example 1, and has a slightly increased chance of transitioning to the arterial with no delay at all at the DDI or at subsequent intersections.  To be honest, this doesn't really look that much better than my first DDI example; maybe I could come up with a better timing if I tried again, but I don't really feel like it.  I'm willing to call Example 1 "good enough".

[thenetwork]

This may be a silly question, but I used to drive in and around the Western Detroit area, and knew even when I was a kid that setting the cruise at 45 MPH could get you past miles of green lights, on the majority of North-South and East-West surface streets.   But when it came to the "spoke" roads which came out of Downtown Detroit (Michigan, Woodward, Gratiot, Grand River,...)  and were not true N-S/E-W roads, which of the two roads (grid vs. spoke) had the green light advantage?

You didn't see it as much on the west side since it was only Grand River and Michigan Avenues that would be the sore thumbs that could throw off the flow, and I mostly cruised Telegraph and the other grid roads.


BTW, Toledo had a pretty decent signal system with many of their major intersections on a lead/lag cycle for opposing traffic.  Though not a true Detroit flow due to some oddball streets and improper spacing of traffic lights, it did help in many cases and the lights at some of the major intersections were programmed to flip the direction of the lead/lag depending on the AM or PM rush hour.

[tradephoric]

This may be a silly question, but I used to drive in and around the Western Detroit area, and knew even when I was a kid that setting the cruise at 45 MPH could get you past miles of green lights, on the majority of North-South and East-West surface streets.   But when it came to the "spoke" roads which came out of Downtown Detroit (Michigan, Woodward, Gratiot, Grand River,...)  and were not true N-S/E-W roads, which of the two roads (grid vs. spoke) had the green light advantage?

The signals are evenly spaced along the spoke roads in Detroit which is key to achieve good two-way signal progression.  The signals are much closer though (0.6 miles as opposed to 1.0 miles) so a shorter cycle is needed to achieve good dual progression.  Instead of 144 second cycles along the mile roads, the diagonal spoke roads will run roughly 90 second cycles to achieve dual progression (assuming 50 mph).  So the mile roads might run a 70 second low cycle at night and a 140 second high cycle during the rushes to achieve dual.  On the other hand, a spoke road will run a 90 second cycle length 24/7.   Both setups are capable of achieving good dual and I'd just say each setup has its own set of pros and cons.

[tradephoric]

I like those time-distance diagrams you came up with VTK.   The Parclo B4 solves the progression problem but probably isn't very practical because of the amount of space it would take up.  The lead/lag SPUI is most interesting, but it requires the interchange to be offset to maximize the effects of lead/lag optimization.  In regards to the DDI example, I wonder if a 50 mph speed limit could be safely maintained at the crossing conflict points.  If the speed limit along a 50 mph corridor was reduced to 35 mph surrounding the DDI, progression would probably suffer.  What is a typical speed limit surrounding a DDI interchange?

I must say I am quite impressed with what you showed in that SPUI example and I see no reason why lead/lag SPUIs couldn't work.  Does anyone know if any lead/lag SPUIs currently exist? 

[NE2]

Does anyone know if any lead/lag SPUIs currently exist? 

http://books.google.com/books?id=t0jfX9fK5oUC&pg=PA19
I can't find which one it is.

[tradephoric]

My point is that I remain unconvinced that stopping SOME drivers means the interchange as a WHOLE is unacceptable. It's ingrained in our psyche to think that, if a traffic control design negatively affects one movement of traffic (generally the movement we happen to be a part of), then the design must be inefficient. This is a nearsighted view of things. Your thesis may very well be correct, that DDIs and certain parclo variants may be less efficient than other designs, but I cannot accept mere interruption of signal progression along one of the corridors as the basis for that claim. It's only one piece of a puzzle, and not every puzzle even looks the same.

In a recent study, the Parclo B4 had the lowest delay and fewest number of stops when compared to the DDI and Parclo A4:

COMPARATIVE ANALYSIS BETWEEN THE DIVERGING DIAMOND INTERCHANGE AND PARTIAL CLOVERLEAF INTERCHANGE USING MICRO SIMULATION MODELING

http://fau.digital.flvc.org/islandora/object/fau%3A3779/datastream/OBJ/view/Comparative_analysis_between_the_diverging_diamond_interchange_and_partial_cloverleaf_interchange_using_microsimulation_modeling.pdf

Two simple reasons why the Parclo B4 minimizes delays:

#1.  Parclo B4 on/off ramp signals only stops one direction of traffic making good signal progression possible.  This reduced delay for drivers on the corridor.  With DDIs, SPUIs, and Parclo A4s, traffic signals stop both directions of travel killing good signal progression and increasing delays.

#2.  Drivers exiting the freeway at a Parclo B4 make a simple right turn to enter onto the arterial.  Drivers exiting the freeway just keep moving if there are gaps in traffic along the arterial.  This reduces delay for drivers exiting the freeway.  With DDIs, SPUIs, and Parclo A4s, drivers exiting the freeway making a left onto the arterial must wait for a green light before proceeding increasing delays.  How annoying is it to wait at a red light at 2AM when no traffic is coming?  That scenario happens routinely for drivers exiting the freeway at DDIs, SPUIs, and Parclo A4s.

[roadfro]

Does anyone know if any lead/lag SPUIs currently exist? 

Plumb Lane & US 395/I-580 in Reno, NV.  It is lead-lag due in part to geometry--a former tight diamond converted to modified SPUI (with partial frontage roads) without reconstructing the existing freeway. Left turns can't quite proceed together in a couple directions, and it is an unusually wide SPUI intersection that takes some getting used to. It is helpful in this case that Plumb Lane ends just east of here at the Reno airport entrance, so through progression is not a factor.


I feel like some of the SPUIs in the Las Vegas area may run lead/lag turns. I can't think of any specific examples, but feel I have seen such operation before. Vegas' coordination system extensively utilizes lead/lag phasing in order to maximize two-way through progression on major arterials, so it stands to reason that lead/lag would be used at the SPUI signals as well.

[jeffandnicole]

How annoying is it to wait at a red light at 2AM when no traffic is coming?  That scenario happens routinely for drivers exiting the freeway at DDIs, SPUIs, and Parclo A4s.

That scenario exists for drivers approaching nearly every traffic light everywhere; more so for those not on the main thru street.

[tradephoric]

How annoying is it to wait at a red light at 2AM when no traffic is coming?  That scenario happens routinely for drivers exiting the freeway at DDIs, SPUIs, and Parclo A4s.

That scenario exists for drivers approaching nearly every traffic light everywhere; more so for those not on the main thru street.

The scenario doesn't exist for drivers exiting the freeway at a Parclo B4.  All drivers exiting the freeway at a Parclo B4 make a simple right turn onto the arterial and experience minimal delays.  There are other advantages to the Parclo B4 interchange, specifically maintaining good signal progression along the corridor (which also helps reduce delays).   

[tradephoric]

MDOT is in the process of redesigning a full cloverleaf interchange at I-75 & Holland Road near Saginaw.  Here's a link to the project report:

https://www.michigan.gov/documents/mdot/MDOT_I-75_Widening_in_Saginaw_County_Environmental_Assessment_422371_7.pdf

Here are models of the interchange running 2035 peak hour volumes as a Parclo B4 and a Parclo AB4 (MDOT's preferred alternative).  Which interchange design has the lowest delay? 

https://www.youtube.com/watch?v=9y-8okb2VOA
https://www.youtube.com/watch?v=7C_1qd34e4Y

[jeffandnicole]

How annoying is it to wait at a red light at 2AM when no traffic is coming?  That scenario happens routinely for drivers exiting the freeway at DDIs, SPUIs, and Parclo A4s.

That scenario exists for drivers approaching nearly every traffic light everywhere; more so for those not on the main thru street.

The scenario doesn't exist for drivers exiting the freeway at a Parclo B4.  All drivers exiting the freeway at a Parclo B4 make a simple right turn onto the arterial and experience minimal delays.  There are other advantages to the Parclo B4 interchange, specifically maintaining good signal progression along the corridor (which also helps reduce delays).   

If it exists in the first place. 

Besides, there are a host of things to look when engineering an intersection.  You're in bed with the B4, but a real engineer is going to look at a whole lot of scenarios to determine if it's the best option.