Crash prone 'modern roundabouts'

[kphoger]

I'm trying to explain to those who will listen (not you, apparently) why this failed design of a roundabout is begging for rampant crashes. I'm talking about what happens in practice, not what's painted on the road. You can't escape the inner lane of a circulating roadway if there is a car in the outer lane. If you're in the inner lane, and there's a car in the outer lane, your only way out of the roundabout is to risk crashing into it.

If you're in the left lane of an approach to the roundabout, it has to spiral you outward into a left turn. I don't see that happening here. See this example shown in the MUTCD. See those lanes getting generated at the center of the roundabout? The little yellow hatched triangles? Why are those there?

Ah, so you appear to be using the phrase "has to" to mean "should".  That makes more sense.  You'd prefer multilane roundabouts to spawn and peel off lanes, while still being completely aware that roundabouts all over the place don't actually do that IRL.

It reminds me of this post of mine from a few years ago:

Yes, it should be straightforward.  And yet, by my count, there are a total of nine conflict points that are not merely glancing blows:

[kalvado]

Outer lane must always exit. This has been the case at modern roundabouts since their invention in the 1960s by Frank Blackmore/TRL, and is only not the case when specifically painted or indicated otherwise.

Wrong. Northbound and southbound exits from this roundabout in this example shown in the MUTCD are shown as a straight/right option lane. Please learn about roundabout design before commenting on roundabout design.

[ img]
Screenshot 2025-06-08 144309 by Paul Across America, on Flickr

The inside lane of the circulating roadway is inescapable. You can't do a two-lane roundabout in all directions, unless you spawn lanes near the center and have them spiral outward. If a four-lane road intersects a two-lane road, you can get away with having two-lane approaches across from each other. But a configuration with two-lane approaches on all four sides is impossible without lanes that spiral outward.

1. Nope. Outer lane exits at every branch, you turn from inner to outer lane at that point. Look at the pavement marks
2. I was referring to design by those who have no clue as a standard practice.

I'm trying to explain to those who will listen (not you, apparently) why this failed design of a roundabout is begging for rampant crashes. I'm talking about what happens in practice, not what's painted on the road. You can't escape the inner lane of a circulating roadway if there is a car in the outer lane. If you're in the inner lane, and there's a car in the outer lane, your only way out of the roundabout is to risk crashing into it.

If you're in the left lane of an approach to the roundabout, it has to spiral you outward into a left turn. I don't see that happening here. See this example shown in the MUTCD. See those lanes getting generated at the center of the roundabout? The little yellow hatched triangles? Why are those there?

[ img]
Screenshot 2025-06-08 144309 by Paul Across America, on Flickr

And for me to say a roundabout generating tons of crashes is a bad design, and for you to disagree, does that mean you want more crashes? Whose side are you on?

OK, let's look closer. This is the intersection in question with all possible trajectories for cars coming from the west plotted. Pardon my crazy skills, orange yellow and blue should overlap on the bottom for the inner lane travel. Blue trajectory isn't very correct but may be OK. All 4 legs of this circle appear similar, so plotting only one set.
This is what is marked on the ground, this is what I see on the road every day.   So, where is the issue?

[paulthemapguy]

Ah, so you appear to be using the phrase "has to" to mean "should".  That makes more sense.  You'd prefer multilane roundabouts to spawn and peel off lanes, while still being completely aware that roundabouts all over the place don't actually do that IRL.

It reminds me of this post of mine from a few years ago:

Yes, it should be straightforward.  And yet, by my count, there are a total of nine conflict points that are not merely glancing blows:

Sure if you want to inject moral relativism into it. I struggle to see how a desire to prevent vehicle collisions and the injury or death that results is a negotiable or debatable thing, but sure. I prefer a setup that leads to fewer crashes.

Thank you for showing your previous post addressing conflict points, because that's exactly the type of analysis that's warranted here. If what's painted on the road evokes a configuration with more conflict points than that which are necessary, that is going to create added crashes.

[kalvado]

Ah, so you appear to be using the phrase "has to" to mean "should".  That makes more sense.  You'd prefer multilane roundabouts to spawn and peel off lanes, while still being completely aware that roundabouts all over the place don't actually do that IRL.

It reminds me of this post of mine from a few years ago:

Yes, it should be straightforward.  And yet, by my count, there are a total of nine conflict points that are not merely glancing blows:

Sure if you want to inject moral relativism into it. I struggle to see how a desire to prevent vehicle collisions and the injury or death that results is a negotiable or debatable thing, but sure. I prefer a setup that leads to fewer crashes.

Thank you for showing your previous post addressing conflict points, because that's exactly the type of analysis that's warranted here. If what's painted on the road evokes a configuration with more conflict points than that which are necessary, that is going to create added crashes.

I am totally with you, traffic lights are certainly a better option.

[jakeroot]

Outer lane must always exit. This has been the case at modern roundabouts since their invention in the 1960s by Frank Blackmore/TRL, and is only not the case when specifically painted or indicated otherwise.

Wrong. Northbound and southbound exits from this roundabout in this example shown in the MUTCD are shown as a straight/right option lane. Please learn about roundabout design before commenting on roundabout design.


Screenshot 2025-06-08 144309 by Paul Across America, on Flickr

I clearly said "...only not the case when specifically painted or indicated otherwise". If paint markings/arrows show a left or straight option, then the right lane can continue around. The northbound exit in your diagram is an example of that exception.

The other approaches clearly show the right lane as a forced exit at each crossover, so I'm not sure what you mean when you mention the southbound exit.

Generally speaking, roundabouts follow the exact same rules as regular intersections: you cannot go right from the left lane, and you cannot go left from the right lane. 2x2 roundabouts operate the same as a 2-lane approach to a signal: left from left, right from right, straight from both.

Some of the false understandings of roundabouts comes from the fact that many view the roundabout as its own road that you merge into and out of. With that in mind, it seems logical that only the outside lane could be used for exiting. Roundabouts, however, are really just a very large intersection with a huge concrete circle in the middle that you have to avoid by steering a bit to the right, then left, then right again. And just like any intersection, large or small, you use the lane corresponding with the direction you want to go: left for left, right for right, straight via both.

I have the honest-to-God view that roundabout markings are their own worst enemy. We go to such great lengths to create these outrageously complicated markings that I'm pretty sure people started to view roundabouts as their own circular roadway. I feel if we didn't use any markings, we might actually see improved crash rates.

[kalvado]

I would say that roundabout is a thing of it's own. It is not just an intersection, as you imply, since that rotary lane has right of way. It is not a road, as entry and exit rules are more involved - sometimes they want you to jump into the third lane from approach.
Pavement markings are a great help - if they last, if there is no snow/salt on the ground, if all lanes are plowed and you don't have to squeeze where you can...  Not that regular intersections are not prone to same problems, but at least lane reduction is handled more organically..

[kphoger]

I have the honest-to-God view that roundabout markings are their own worst enemy. We go to such great lengths to create these outrageously complicated markings

And, if someone doesn't already know what they're doing, then they're likely to ignore the markings entirely anyway.

I feel if we didn't use any markings, we might actually see improved crash rates.

They tried that at the elongated roundabout in Towson, MD, but then they eventually modified it.

[bwana39]

Roundabouts have two usages.

The first is to slow the traffic down. From a safety POV it is the primary one. They also minimize t-bone strikes, but may increase fender benders.

The reason cities like them is the exorbitant cost of placing and maintaining signal lights. The cost of redoing a dongle light control box is over $100K. A minimal new placement is around $250K and maintenance and utility costs is around $10K per year per intersection. This is in a MINIMAL placement.

[7/8]

OK, let's look closer. This is the intersection in question with all possible trajectories for cars coming from the west plotted. Pardon my crazy skills, orange yellow and blue should overlap on the bottom for the inner lane travel. Blue trajectory isn't very correct but may be OK. All 4 legs of this circle appear similar, so plotting only one set.
This is what is marked on the ground, this is what I see on the road every day.  So, where is the issue?

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

[7/8]

The Region of Waterloo (in Ontario) will be starting a pedestrian safety pilot project testing RRFB's, raised crosswalks, and both combined at three separate roundabouts. I think the raised crosswalks if done well will be the most helpful by forcing drivers to slow down. I find we design roundabout exits too straight, which encourages drivers to quickly speed off, making for a dangerous pedestrian crossing.
https://www.cambridgetoday.ca/local-news/flashing-lights-coming-to-franklin-and-bishop-roundabout-this-summer-10774657

[kalvado]

OK, let's look closer. This is the intersection in question with all possible trajectories for cars coming from the west plotted. Pardon my crazy skills, orange yellow and blue should overlap on the bottom for the inner lane travel. Blue trajectory isn't very correct but may be OK. All 4 legs of this circle appear similar, so plotting only one set.
This is what is marked on the ground, this is what I see on the road every day.  So, where is the issue?

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

Blue cannot be the correct one as you have no guarantee of two lane exit northbound, and  disallowing green makes  that turn impossible for a single-lane exit.
But what you're describing is basically a difference between a traffic circle and a roundabout.

[kphoger]

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

Blue cannot be the correct one as you have no guarantee of two lane exit northbound, and  disallowing green makes  that turn impossible for a single-lane exit.

I don't think I've ever seen a true two-lane roundabout where the blue path is impossible.  Two lanes on approach, two lanes of circulating roadway, not a spiraling/turbo roundabout, left turn intended, but only one exit lane available.  Can you provide an example?

[kalvado]

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

Blue cannot be the correct one as you have no guarantee of two lane exit northbound, and  disallowing green makes  that turn impossible for a single-lane exit.

I don't think I've ever seen a true two-lane roundabout where the blue path is impossible.  Two lanes on approach, two lanes of circulating roadway, not a spiraling/turbo roundabout, left turn intended, but only one exit lane available.  Can you provide an example?

https://www.google.com/maps/@42.9485247,-73.7896294,113m/data=!3m1!1e3!5m1!1e1
6 o'clock to 9 o'clock.

UPD:
second one, just north:
https://www.google.com/maps/@42.9630374,-73.7913518,113m/data=!3m1!1e3

[kphoger]

https://www.google.com/maps/@42.9485247,-73.7896294,113m/data=!3m1!1e3!5m1!1e1
6 o'clock to 9 o'clock.

UPD:
second one, just north:
https://www.google.com/maps/@42.9630374,-73.7913518,113m/data=!3m1!1e3

What these two roundabouts illustrate for me is that the distinction between a spiral roundabout and a regular roundabout can be fuzzy.  The pavement markings violate my "not a spiraling/turbo roundabout" criterion but, with only some very slight modification to the striping, it would meet the criterion.  *ugh*

[7/8]

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

Blue cannot be the correct one as you have no guarantee of two lane exit northbound, and  disallowing green makes  that turn impossible for a single-lane exit.

I don't think I've ever seen a true two-lane roundabout where the blue path is impossible.  Two lanes on approach, two lanes of circulating roadway, not a spiraling/turbo roundabout, left turn intended, but only one exit lane available.  Can you provide an example?

https://www.google.com/maps/@42.9485247,-73.7896294,113m/data=!3m1!1e3!5m1!1e1
6 o'clock to 9 o'clock.

UPD:
second one, just north:
https://www.google.com/maps/@42.9630374,-73.7913518,113m/data=!3m1!1e3

These examples remind me of Fountain/Fairway roundabout in Cambridge, ON.


Scenario A (EB-NB): Blue is correct and red is incorrect. Note the white dashed line within the lime green circle which indicates the creation of a new inner lane. Therefore, the blue path is not changing lanes within the roundabout. The red line is incorrect since you can't exit from the inner lane into a single-lane exit (also note the left-turn only arrow just above the lime green circle, another indicator that you can't exit from the inner lane).

Scenario B (WB-SB): Blue is correct and red is incorrect. Note the lack of a white dashed line in the lime green circle. Following the red line is changing lanes in the roundabout which is against the rules.

Unfortunately scenario A confuses a lot of drivers and it's common to see people take the red line. Thankfully, drivers in our area have no issues with scenario B.

[kalvado]

Blue is the correct way to make a left. Green is incorrect, since you can't change lanes within the roundabout (at least this is how it works where I'm from, and we have lots of roundabouts in my area).

Blue cannot be the correct one as you have no guarantee of two lane exit northbound, and  disallowing green makes  that turn impossible for a single-lane exit.

I don't think I've ever seen a true two-lane roundabout where the blue path is impossible.  Two lanes on approach, two lanes of circulating roadway, not a spiraling/turbo roundabout, left turn intended, but only one exit lane available.  Can you provide an example?

https://www.google.com/maps/@42.9485247,-73.7896294,113m/data=!3m1!1e3!5m1!1e1
6 o'clock to 9 o'clock.

UPD:
second one, just north:
https://www.google.com/maps/@42.9630374,-73.7913518,113m/data=!3m1!1e3

These examples remind me of Fountain/Fairway roundabout in Cambridge, ON.
[ img]

Scenario A (EB-NB): Blue is correct and red is incorrect. Note the white dashed line within the lime green circle which indicates the creation of a new inner lane. Therefore, the blue path is not changing lanes within the roundabout. The red line is incorrect since you can't exit from the inner lane into a single-lane exit (also note the left-turn only arrow just above the lime green circle, another indicator that you can't exit from the inner lane).

Scenario B (WB-SB): Blue is correct and red is incorrect. Note the lack of a white dashed line in the lime green circle. Following the red line is changing lanes in the roundabout which is against the rules.

Unfortunately scenario A confuses a lot of drivers and it's common to see people take the red line. Thankfully, drivers in our area have no issues with scenario B.

My take of it is that some things are implied, rather than explicitly painted - especially if the paint wears out.
In your example, adding inner lane at one point implies turbo design across the loop - hence I would assume each exit is an add lane and drop lane. Then staying in lane means going into an outer one. 

[jakeroot]

I feel if we didn't use any markings, we might actually see improved crash rates.

They tried that at the elongated roundabout in Towson, MD, but then they eventually modified it.

Roundabout markings really didn't become a standardized feature until the 2009 MUTCD. Prior to this, each state had their own standards. Some came from the UK (like Colorado and California), and some from Australia (like Washington State in some places) -- I think that might be the only two sources, since they were the only English-language manuals available in the 1990s.

Part of these design standards dictated either very simple markings, or absolutely no markings at all. The Towson roundabout was merely one of many roundabouts built in that timeframe that used no markings.

In Washington State, the Marvin Road roundabouts, built around 2001 by the City of Lacey, had Australian style markings, which merely divided the "circulatory roadway" using an extremely simple solid white line. This is a 2x2 roundabout, and it follows the same rules as a regular intersection (left from left, right from right, straight from both), and there was (and still is) absolutely no prohibition on turning into either the near or far lane from the left lane when going left:


Marvin Road at Britton Pkwy, 2003 by Jacob Root, on Flickr

You can see it in action in the below video from WSDOT, around the same time-frame as the satellite image above:

Two other roundabouts in WA, built around the same time-frame, also had no markings at all, although these were built by the state using UK standards:

I spoke to Brian Walsh years ago, an engineer with Washington State DOT, about his experience building them at the time. Essentially, they had no American standards to go by, so they experimented with different options. Mostly they used UK style, but they also tried concentric markings too. From what he told me, concentric markings made it much worse because everyone thought you had to be in the "outside lane" to exit. They eventually switched from concentric to spiral markings, which is now the US standard.

This roundabout below in Gig Harbor opened with no markings, then they added the concentric markings (second image). Notice in the second image that the phantom tire tracks shift from both lanes in the first image, to almost exclusively the outside lane in the second image, as drivers give up using the inside lane:


WA-16 at Borgen Blvd, 2002 by Jacob Root, on Flickr


WA-16 at Borgen Blvd, 2005 by Jacob Root, on Flickr

This roundabout below, also built by WSDOT (and Brian Walsh, the aforementioned engineer) opened in 2001 with no markings in the city of Monroe. It had basic spiral markings added around 2005:


WA-522 at Main St / Tester Rd, 2002 by Jacob Root, on Flickr


Over in Colorado, their roundabouts built in the 1990s originally had no markings at all. And even today, the markings are pretty sparse. In this image from 2004, you can see a very large roundabout with three-lane entries and no markings:


I-70 at Vail Road, Colorado, 2004 by Jacob Root, on Flickr

And in Long Beach, California, the grand-daddy of UK roundabouts in America was rebuilt into the below configuration in the 1990s, I think by Frank Blackmore himself (from the UK's TRL). Four lane entries with no markings at all. It now has spiral markings and substantially smaller entries. This roundabout basically took every single UK standard and flipped it, right down to the massively flared entries and wide open "circulatory roadway" with no markings.


Long Beach Circle by Jacob Root, on Flickr

And that is my history lesson on roundabout markings.

[kphoger]

And that is my history lesson on roundabout markings.

And no mention of 'Alberta'?? 

For the record, I'm in favor of vanilla Alberta striping for roundabouts.

[kphoger]

For the sake of the uninitiated, I suppose I ought to say what Albert striping is.

It's when the dashed line of the circulating roadway follows the exit path at each quadrant.  Illustration from Alberta's 2013 design bulletin below:

[kalvado]

For the sake of the uninitiated, I suppose I ought to say what Albert striping is.

It's when the dashed line of the circulating roadway follows the exit path at each quadrant.  Illustration from Alberta's 2013 design bulletin below:

[ img]

Turbo traffic pattern in a circular build. And turn into the outer lane is clearly an option for  circulating traffic...

[kphoger]

And turn into the outer lane is clearly an option for  circulating traffic...

Nope.  You seem to be under the impression that Alberta striping intends to allow traffic to take the bright red path shown below:



But this is not the case.

Drivers using the left lane to enter the circle must do so when it is safe and stay in the left lane while in the circle/roundabout. They must exit using the left lane. Drivers planning to travel past the first exit should use the left lane to enter and exit.



https://open.alberta.ca/dataset/ddca813d-5463-4daa-afc9-093807a1bb6a/resource/e72fcd84-c5e9-4241-b907-4b1ef00dbce7/download/trans-drivers-guide-cars-light-trucks-2021-01.pdf

[7/8]

For the sake of the uninitiated, I suppose I ought to say what Albert striping is.

It's when the dashed line of the circulating roadway follows the exit path at each quadrant.  Illustration from Alberta's 2013 design bulletin below:

This might be common throughout Canada, the Region of Waterloo (ON) has similar striping (see OneDrive link to a PDF of their standard).

Turbo traffic pattern in a circular build. And turn into the outer lane is clearly an option for  circulating traffic...

Maybe this is a differs by locale, but despite the striping, our Region's guidelines clearly state that switching to the outer lane when turning left is not allowed.

"Exit directly from the inner lane":


"Do not change lanes within the roundabout":

[kalvado]

For the sake of the uninitiated, I suppose I ought to say what Albert striping is.

It's when the dashed line of the circulating roadway follows the exit path at each quadrant.  Illustration from Alberta's 2013 design bulletin below:

This might be common throughout Canada, the Region of Waterloo (ON) has similar striping (see OneDrive link to a PDF of their standard).

Turbo traffic pattern in a circular build. And turn into the outer lane is clearly an option for  circulating traffic...

Maybe this is a differs by locale, but despite the striping, our Region's guidelines clearly state that switching to the outer lane when turning left is not allowed.

"Exit directly from the inner lane":


"Do not change lanes within the roundabout":

That's exactly why I don't think roundabouts are a bad idea, but I just think DOTs - or their equivalents - should not be allowed to build those until they have at least 50 years of experience in building roundabouts.