Crash prone 'modern roundabouts'

[kphoger]

Also, it's another new dual lane roundabout, which this thread has become a dumping ground for.

My one takeaway from this whole thread is that there are problems inherent in full dual-lane roundabouts:  specifically that there are non-glancing conflict points.  And that statistics for both single-lane roundabouts (with no such non-glancing conflict points) and dual-lane roundabouts should not be combined, lest the results be misleading.

If you will, I have a slightly wider version of this. What is the role of roundabouts in a grand scheme of things in terms of throughput? Where would you put roundabout in a list of intersection designs with increasing throughput - that is: 4-way stop, traffic light, grade separated?
I would say that roundabouts belong to busy 4-way stop/not-so-busy traffic light interval. We had some european guidelines here showing a 30K/daily limit for complex roundabout, which seems reasonable to slightly optimistic; and I know a (very busy) traffic light controlled intersection with 50k+ daily traffic.
Now you can increase traffic light throughput quite a bit by adding components, such as dedicated turn lanes or slip ramps - of course at a cost of footprint. And those upgrades can be fairly seamless, you don't have to bulldoze entire thing for an extra lane (although imminent domain of land is a pain..)
You can also upgrade stop signals to a traffic light.
Roundabout, however, is a solution with limited throughput which can NOT be easily upgraded to next tier - which is complex traffic light.
And that can easily become a problem: no upgrade short of increasing complexity beyond the reason OR total rebuild.... Or continue funneling traffic through intersection above design limit - and pretend problem doesn't exist.

This is a good point.  And I agree with your placement of roundabouts on the throughput continuum.

[tradephoric]

I visually attempted to make kalvado's excellent point in another post.   CJW2001 kept banging on about how much roundabouts have improved traffic flow in Carmel, ignoring the fact that traffic flow could have been improved by simply adding turn lanes to the signalized intersections that the roundabouts replaced.  And kalvado is right in pointing out that roundabouts have a limited throughput.... there reaches a point where roundabouts just don't work well.  So is it any surprise that traffic flow improved in Carmel after these intersections pictured below were upgraded?  An intersection is upgraded that probably hasn't been touched since the 60s... and we are surprised that traffic flow improves?  Let's just give all the credit to the roundabouts

To be honest, I'm sure there was an improvement in traffic flow in Carmel after the roundabouts were built.   But could that be because the intersections the roundabouts replaced were woefully inadequate to carry the capacity of a ballooning Carmel population?  Here are some before/after pictures of roundabout location in Carmel.  Not surprising traffic flow improved when you compare these pictures... not to mention the fact the two major surface streets in Carmel were converted to grade-separated freeways over the past decade.

[tradephoric]

But the most drastic improvement in traffic flow of any roundabout project I've seen was at Maple & Drake in West Bloomfield, Michigan.  Here are the before/after pictures.  The signalized intersection the roundabout replaced didn't even have dedicated right turn lanes.  Is it any wonder that traffic cleared up after this massive 3x2 roundabout was completed? 

[kphoger]

I get it that adding a right-turn lane to a stoplight approach can reduce the number of cars that have to stop at the light:  at least some right-turning traffic will be taken out of the queue.  But I imagine less of an improvement due to the addition of more left-turn storage.  Sure, dual turn lanes require a shorter arrow phase, but how much time savings are we really talking about there?  All that is to say, traffic might not have "cleared up" to the same extent with the mere addition of lanes as it did with the removal of red lights altogether.

[kalvado]

I get it that adding a right-turn lane to a stoplight approach can reduce the number of cars that have to stop at the light:  at least some right-turning traffic will be taken out of the queue.  But I imagine less of an improvement due to the addition of more left-turn storage.  Sure, dual turn lanes require a shorter arrow phase, but how much time savings are we really talking about there?  All that is to say, traffic might not have "cleared up" to the same extent with the mere addition of lanes as it did with the removal of red lights altogether.

Depends on traffic pattern and design. If left turn and straight through are squeezed into a single lane (very common on low-traffic roads), and there is enough of oncoming traffic - a single turning car can block a lot of flow. Or you need a very restrictive phasing of traffic light. Or extra footprint for a long left turn lane.
Widening roads to add extra lane - not just to intersection, but entire road - apparently adds throughput. And looks like that is the case for most pictures trade shows.
Eliminating red lights may help if there is significant dead time in a cycle, that is no sensors and a very short timing cycle ANd interestingly enough, longer cycle of light increases average wait time - but also increases peak throughput.
Actually one can take beloved "fewer conflict points in roundabout" picture and explain how roundabout peak throughput is lower if traffic doesn't come from one side of intersection, and why traffic light is better and safer..

[tradephoric]

All that is to say, traffic might not have "cleared up" to the same extent with the mere addition of lanes as it did with the removal of red lights altogether.

I'll concede that point.  The congestion really has disappeared ever since the Maple & Drake roundabout was built (and it was built over a decade ago now).  By the way the 3x2 roundabout pictured above has since been converted to a 2x2 roundabout.  The roundabout had a crash rate of 4.43 MEV when it was a triple-lane roundabout, but my guess is it has dropped since the conversion to a 2x2. 

As much as I have focused on 2x2 roundabouts having high crash rates, I think they can work OK when there is a limited capacity upstream/downstream of the roundabout (ie. a two-lane road).  The limited roadway capacity can help create natural gaps in traffic, reducing the likelihood of drivers entering the roundabout side-by-side.  Drivers would theoretically only enter the roundabout side-by-side if a driver had to yield for a vehicle already inside the roundabout.  But if the upstream geometry is such that you are driving along a 4-lane boulevard approaching a roundabout, drivers can very easily enter a roundabout side-by-side (increasing the likelihood of sideswipe crashes and/or crashes when people don't follow the guide arrows).

[tradephoric]

They are planning to build a complex elevated roundabout in Lincoln.  Below the elevated roundabout there will be a standard signalized intersection with direct left turn movements (you know, the dangerous turning movements that roundabouts are designed to prevent in the first place).  Now Lincoln drivers will get to enjoy the crash prone multi-lane elevated roundabout with less frequent deadly crashes below it.   Great design!   Lincoln just doesn't have a good track record when it comes to roundabouts.  The triple-lane 14th Street and Superior lane roundabout was a mess from day one.  The only way to reduce the number of crashes was by downsizing it and reducing the number of circulating lanes.  Just recently the city built a retaining wall/sculpture in the middle of the 14th/Superior roundabout so any out of control driver will potentially crash dead on into a retaining wall.   Sounds safe.... of course what's going to happen when a driver blows through an elevated roundabout?

Elevated roundabout design for 14th, Old Cheney gets final public scrutiny
http://journalstar.com/news/local/govt-and-politics/elevated-roundabout-design-for-th-old-cheney-gets-final-public/article_1de51613-002b-51e7-ba08-d85ebaffc3eb.html

[tradephoric]

The intersection, which has garnered quite a reputation for being Australia's most hated over the years, is a particularly gnarly mix of heavy traffic, confusion and chaos.

The multi-lane nightmare even scored the dubious title of the state's most complained about roundabout in 2015, according to Transport for NSW records.

And just a few hundred metres away from the roundabout is a spot that's been named the worst merging location for accidents in NSW - the intersection of Homebush Bay Drive and the M4 Motorway. That's according to data from NSW insurance claims in 2016.

Is this Australia's worst roundabout?
https://www.centraltelegraph.com.au/news/is-this-australias-worst-roundabout/3423161/

[kalvado]

They are planning to build a complex elevated roundabout in Lincoln.  Below the elevated roundabout there will be a standard signalized intersection with direct left turn movements (you know, the dangerous turning movements that roundabouts are designed to prevent in the first place).  Now Lincoln drivers will get to enjoy the crash prone multi-lane elevated roundabout with less frequent deadly crashes below it.   Great design!   Lincoln just doesn't have a good track record when it comes to roundabouts.  The triple-lane 14th Street and Superior lane roundabout was a mess from day one.  The only way to reduce the number of crashes was by downsizing it and reducing the number of circulating lanes.  Just recently the city built a retaining wall/sculpture in the middle of the 14th/Superior roundabout so any out of control driver will potentially crash dead on into a retaining wall.   Sounds safe.... of course what's going to happen when a driver blows through an elevated roundabout?

Elevated roundabout design for 14th, Old Cheney gets final public scrutiny
http://journalstar.com/news/local/govt-and-politics/elevated-roundabout-design-for-th-old-cheney-gets-final-public/article_1de51613-002b-51e7-ba08-d85ebaffc3eb.html

Would be interesting to see what kind of traffic distribution prompted such layout. Must be something quite funny and pretty heavy...

[kphoger]

Lincoln drivers will get to enjoy the crash prone multi-lane elevated roundabout^{[citation needed]}

Because I'm sick of this thread being a dumping ground for every roundabout project you think might become crash-prone.  If all you want to do is be cynical and snarky about whatever roundabout project comes across your radar that you don't like, then just change the thread title to "tradephoric dumps on roundabouts" or something along those lines.

You spent the first who knows how many pages of this thread tenaciously convincing us that there is actually a subset of roundabouts that are inherently more crash-prone than others and discussing possible reasons for that.  You've hung onto this thread for around two years, slowly but surely gaining the respect and changing the minds of many of us–even if only to a small degree.  But now you've devolved into blasting roundabout projects as "crash-prone" before they've even been constructed.  And because of that, frankly, you're losing my respect in the process.

[tradephoric]

But now you've devolved into blasting roundabout projects as "crash-prone" before they've even been constructed.  And because of that, frankly, you're losing my respect in the process.

Making predictions about crash prone multi-lane roundabouts is almost as easy as predicting that the sun will rise in the east tomorrow.  I've stated this before, but there hasn't been a documented 2x2 multi-lane roundabout with a crash rate below 2.0 MEV.  Many agencies are required to undergo safety audits when a signalized intersection has a crash rate over 2.0 MEV.  Constructing a 2x2 modern roundabout will almost ensure that the crash rate will exceed this "critical"  crash rate.  I have made predictions about the number of roundabout crashes in the past, and you can even bash me for betting the over on my prediction and I won't take offense.

Prediction:

Here's a traffic simulation of a roundabout currently under construction in Farmington Hills, Michigan at 14 Mile & Orchard Lake.  Over the last 5 years, the intersection has averaged 46 crashes per year.  This roundabout could experience a significant number of crashes once completed.  Does anybody want to wager how many crashes this roundabout will experience in FY2016?  The over/under is 180 crashes (I'll bet the over). 

https://www.youtube.com/watch?v=_88_Z01RA1g

Reality:

14-Orchard roundabout is tops for accidents, but is it really dangerous?
http://www.freep.com/story/news/local/michigan/oakland/2017/04/04/orchard-roundabout-tops-accidents-really-dangerous/100046220/

Turns out there were 163 crashes at the Orchard Lake/M-5 roundabout in 2016.  That's a 232% increase in total crashes compared to the pre-roundabout condition when it averaged 49 crashes per year (look at the chart below for pre-roundabout crash data).  Worse yet, there were 27 injury crashes at the roundabout compared to an average of 9.5 injury crashes/year before the roundabout - a 184% increase in injury crashes.


http://semcog.org/Data-and-Maps/High-Frequency-Crash-Locations/Point_Id/63052513/view/RoadIntersectionCrashDetail

[tradephoric]

Carmel's Police Department recently released it's 2017 annual report.  In previous reports the Police Department would always show a map of the top 20 accident locations in Carmel during the year and would list the total number of crashes that occurred.  That all changed in 2014, when the top accident location map was suddenly omitted from their annual report.  There was a sudden lack of transparency, and now there is no easy way for the public to see what locations in Carmel are experiencing the most crashes (put another way the mayor doesn't want any bad PR for his precious roundabouts).

Carmel's non-transparent 2017 annual report:
http://www.carmel.in.gov/home/showdocument?id=10651

[kphoger]

there hasn't been a documented 2x2 multi-lane roundabout with a crash rate below 2.0 MEV. 

In all countries of the world?

[jakeroot]

there hasn't been a documented 2x2 multi-lane roundabout with a crash rate below 2.0 MEV. 

In all countries of the world?

I can't speak for some of the European countries, but I believe some of the worst intersections (in terms of pure collisions) in Australia are also roundabouts. There was a news article posted somewhere back on this thread about a very bad intersection in Australia that was basically like a double roundabout. To my knowledge, roundabouts have been built in Australia for a long time, and many people in Australia are also British, so, really, everyone should be used to them by now. Oddly enough, Australia's roundabouts are also very reminiscent of American roundabouts. To my knowledge, American roundabouts were originally based on British designs, just reflected, but then switched to the Australian style, which is how they are still designed today. The big giveaway is the chicanes that are used approaching roundabouts, which is a chiefly Australian design (at least originally), as well as the large truck apron. Many European countries do not use truck aprons, or at least large ones like those in the US, or at least large ones like those in the US.

tl;dr -- I'm not 100% sure, but there are definitely other crash-prone roundabouts outside of the US.

[jeffandnicole]

there hasn’t been a documented 2x2 multi-lane roundabout with a crash rate below 2.0 MEV. 

In all countries of the world?

The key is "documented".

He's prone to find information regarding roundabouts where a serious or fatal crash has occurred. He recently brought up a news story about a crash at a roundabout in The Villages, FL.  The Villages have at least 31 roundabouts, most 2x2, on their two main roads (Morse Blvd & Buena Vista Blvd). He's yet to reveal a story involving any of the other 30.  I would be absolutely shocked that every single one of them had MEVs above 2. 

[johndoe]

As I saw some "failure to yield" mistakes today I thought about this thread... I know I've seen studies where different pavement markings, signs, and fewer lanes are used to encourage proper yielding (and as I recall most don't have optimistic conclusions).  I'm assuming there is a correlation between higher speeds and yielding violations (just a guess...has a study looked at this?)

I was wondering about two other options: speed tables and stop signs.  Speed tables have been discussed as a way to control speeds at entry/ped crossings, are DOTs starting to build these now?  Regarding stop signs, I remember someone earlier in the thread talking about replacing yield signs with stop signs.  Although it would drastically lower the capacity, does anyone know if DOTs have done this on particularly problematic approaches?

[kalvado]

As I saw some "failure to yield" mistakes today I thought about this thread... I know I've seen studies where different pavement markings, signs, and fewer lanes are used to encourage proper yielding (and as I recall most don't have optimistic conclusions).  I'm assuming there is a correlation between higher speeds and yielding violations (just a guess...has a study looked at this?)

I was wondering about two other options: speed tables and stop signs.  Speed tables have been discussed as a way to control speeds at entry/ped crossings, are DOTs starting to build these now?  Regarding stop signs, I remember someone earlier in the thread talking about replacing yield signs with stop signs.  Although it would drastically lower the capacity, does anyone know if DOTs have done this on particularly problematic approaches?

This all would ultimately defeat the purpose of roundabout as a higher throughput traffic control device. Is there a reason even to consider stop signs at the entering lanes when that degrades entire system to a glorified 4-way stop?

[jakeroot]

As I saw some "failure to yield" mistakes today I thought about this thread... I know I've seen studies where different pavement markings, signs, and fewer lanes are used to encourage proper yielding (and as I recall most don't have optimistic conclusions).  I'm assuming there is a correlation between higher speeds and yielding violations (just a guess...has a study looked at this?)

I was wondering about two other options: speed tables and stop signs.  Speed tables have been discussed as a way to control speeds at entry/ped crossings, are DOTs starting to build these now?  Regarding stop signs, I remember someone earlier in the thread talking about replacing yield signs with stop signs.  Although it would drastically lower the capacity, does anyone know if DOTs have done this on particularly problematic approaches?

This all would ultimately defeat the purpose of roundabout as a higher throughput traffic control device. Is there a reason even to consider stop signs at the entering lanes when that degrades entire system to a glorified 4-way stop?

I would be extremely hesitant to install stop signs, but its capacity could still be higher than a 4-way stop because multiple vehicles could occupy the circle at once.

Still, it should be a last resort. I would love to see speed tables used. They are used frequently in the Netherlands, and I don't see why they wouldn't work here, especially at intersections where vehicles must yield to pedestrians, such as at roundabouts or slip lanes. The latter situation is a frequent use-case in Boulder, CO: https://goo.gl/Q9SV8V & https://goo.gl/B9ybvs.

[kalvado]

I would be extremely hesitant to install stop signs, but its capacity could still be higher than a 4-way stop because multiple vehicles could occupy the circle at once.

Still, it should be a last resort. I would love to see speed tables used. They are used frequently in the Netherlands, and I don't see why they wouldn't work here, especially at intersections where vehicles must yield to pedestrians, such as at roundabouts or slip lanes. The latter situation is a frequent use-case in Boulder, CO: https://goo.gl/Q9SV8V & https://goo.gl/B9ybvs.

Multiple vehicles occupying circle for capacity boost is a... khm ....  interesting argument.
Pretty much as saying that you can enjoy your beverage faster if you use a longer straw which would holds more liquid. In both situations, however, it is the bottleneck - in case of roundabout being the entry rate - which defines total throughput.
Slowing down to 20 mph means average car takes about 0.5 second to travel its length, and that directly cuts into 2 second interval. So same story - decreasing throughput and killing suspension.
You see, this is why  traffic lights are so great - they move 2D space conflict into a time domain separation. Resolving the conflict in third dimension is still more efficient, but much more expensive.

[Rothman]

I'll take the roundabouts on NY 85 south of Albany over the former traffic lights any day.  I'd imagine a poll would result in most citizens if Bethlehem, NY saying they are more efficient that the lights that preceded them.

[kalvado]

I'll take the roundabouts on NY 85 south of Albany over the former traffic lights any day.  I'd imagine a poll would result in most citizens if Bethlehem, NY saying they are more efficient that the lights that preceded them.

Yep, those roundabouts actually helped. Probably building up from 2-lane undivided to 4-lane divided also added some minor improvements in traffic, though. Bypassing a chunk of New Scotland road could also contribute a bit.
But the last one out is a father, you know...

[tradephoric]

A Canadian report discussing some potential solutions to driver errors at multi-lane roundabouts...

SOLUTIONS TO DRIVER ERRORS AT MULTI-LANE ROUNDABOUTS
http://www.tac-atc.ca/sites/default/files/conf_papers/weber_p_-_solutions_to_driver_errors_at_multi-lane_roundabouts.pdf

The case study that interested me was the St. Joseph Boulevard and Jeanne D'Arc Boulevard roundabout in Ottawa.  The design is similar to the crash-prone 2x2 State and Ellsworth roundabout in Ann Arbor, Michigan (the very first roundabout discussed in this thread).  Each roundabout has roughly the same traffic volumes, roughly the same ICD, both are about 2000 feet from a freeway, both have commercial developments along all 4-corners, the link speeds approaching each roundabout are roughly the same... yet the Ann Arbor roundabout has historically double the number of crashes as the Ottawa roundabout.  Maybe the reporting requirements between American and Canadian agencies are different, and that can explain the discrepancy between the crashes, but I thought the comparison was interesting nonetheless.






Agencies seem too quick to attribute reductions in accidents to whatever new "fix"  they are testing out at the roundabout.  Time-and-time again there is an initial spike in crashes at these new roundabouts only for the numbers to level off over the next few years.  Apart from whatever "fix"  they are pushing now, I'm more interested in why the initial crash numbers at the Ottawa roundabout were so much lower than the initial crash numbers at the Ann Arbor roundabout.

[webny99]

I'm more interested in why the initial crash numbers at the Ottawa roundabout were so much lower than the initial crash numbers at the Ann Arbor roundabout.

Much better markings.
Also, the one approach to the Ann Arbor one (bottom left in the image) comes in at an angle, meaning traffic entering from that approach has a harder turn into the roundabout and probably sub-optimal visibility.

[tradephoric]

Much better markings.
Also, the one approach to the Ann Arbor one (bottom left in the image) comes in at an angle, meaning traffic entering from that approach has a harder turn into the roundabout and probably sub-optimal visibility.

Another noticeable difference is the plantings (or lack thereof) in the central islands.  The Ann Arbor roundabout just has a patch of grass in the central island while the Ottawa roundabout has tall plantings and 3 pine trees. Shrubs in the central island are cool...  they help the driver realize that they better start turning soon because it's no longer a paved road infront of them.  I just don't like plantings in the central island that will kill you if you strike them at high speeds.



If you notice there is also a lot more side-friction at the Ottawa roundabout.  There are several poles with flags on them along the splitter island and the sidewalks are right up against the travel lane.  OTOH, in the Ann Arbor roundabout the splitter islands are pretty bare and the sidewalks are set back several feet from the travel lane.  You could make the argument that drivers would be more likely to drive cautiously through the Ottawa roundabout since there is so much more side-friction.

[kalvado]

Much better markings.
Also, the one approach to the Ann Arbor one (bottom left in the image) comes in at an angle, meaning traffic entering from that approach has a harder turn into the roundabout and probably sub-optimal visibility.

Another noticeable difference is the plantings (or lack thereof) in the central islands.  The Ann Arbor roundabout just has a patch of grass in the central island while the Ottawa roundabout has tall plantings and 3 pine trees. Shrubs in the central island are cool...  they help the driver realize that they better start turning soon because it's no longer a paved road infront of them.  I just don't like plantings in the central island that will kill you if you strike them at high speeds.

If you notice there is also a lot more side-friction at the Ottawa roundabout.  There are several poles with flags on them along the splitter island and the sidewalks are right up against the travel lane.  OTOH, in the Ann Arbor roundabout the splitter islands are pretty bare and the sidewalks are set back several feet from the travel lane.  You could make the argument that drivers would be more likely to drive cautiously through the Ottawa roundabout since there is so much more side-friction.

another big factor in roundabout performance is number of cars doing left turns. Hard to say without knowing local traffic patterns