Crash prone 'modern roundabouts'

[Rothman]

We're going in circles.

OH the irony humanity!

FTFY.

[iBallasticwolf2]

We're going in circles.

OH the irony humanity!

FTFY.

Well done.

[DaBigE]

We're going in circles.

OH the irony humanity!

FTFY.

[tradephoric]

No, you have not. What you have reiterated are more generalities for a small sample of intersections. When I say investigate, I mean take all of the crash-prone roundabouts and investigate each roundabouts features: fast path speeds, deflection, path overlap, lane configuration, signing, pavement marking, PHI-angles, in addition to size.

The study done by the IIHS in 2000 analyzed a small sample of intersections as well.  That study has been embraced by the engineering community and is routinely cited in DOT websites and a link to the study can be found on the FHWA website.  The study looked at the before/after crash data of 24 total roundabouts (15 single lane, and 9 multi-lane) using the empirical Bayes approach.  It didn't consider fast path speeds, deflection angles, path overlap, signing, pavement marking, PHI-angles, or size.  The conclusion of the study is that total crashes reduce by 39%, injury crashes reduce by 76%, and fatal crashes reduce by 90%. 

http://safety.fhwa.dot.gov/intersection/resources/fhwasa09027/resources/Crash%20Reductions%20Following%20Installation%20of%20Roundabouts.pdf

This thread, more than anything, highlights the fact that the IIHS study in 2000 is too limited in their analysis (and is heavily skewed towards single-lane roundabouts).  It should be painfully obvious that it is unlikely the high capacity 2x2 roundabouts being built today will see a 39% reduction in total crashes.  It should also be obvious that bigger roundabouts appear to perform better than smaller roundabouts.  Ultimately, the worst performing roundabouts in America have small central island diameters.  This is a simple fact that is difficult to dispute.

[tradephoric]

Based on MnDot's "critical crash rate index" , the 3rd most dangerous intersection in the Twin Cities metro for 2013 was at a double-lane roundabout at U.S. 61 & Broadway Avenue in Forest Lake.

http://kstp.com/article/stories/s3699209.shtml

The roundabout has a central island diameter of only 94 feet.  The fix?  They converted it to a single lane roundabout.  At least half-a-dozen examples have been cited of high crash rate roundabouts that have had circulating lanes removed.  Keep in mind, there are only about 40 high capacity 2x2 or 3x2 non-interchange roundabouts in America.  That means roughly 10% of the high capacity non-interchange roundabouts have had circulating lanes removed.

[tradephoric]

Here's an analysis of some multi-lane roundabouts in Oregon.  A multi-lane roundabout in Astoria, Oregon saw a 125% increase in injury crashes.  Another multi-lane roundabout in Springfield, Oregon had a crash rate of 2.91 MEV.  As a comparison, here is a list of major Springfield intersections with their corresponding crash rates.


http://alexforfg.com/more-troubling-statistics-on-oregons-dual-lane-roundabouts/

[Bickendan]

Coe Circle doesn't even make the list. I wonder why that is...

Edit: So this is at the top of Oregon's list:


One thing that would greatly help: swapping out the yield signs with stop signs.

[tradephoric]

Coe Circle doesn't even make the list. I wonder why that is...

Edit: So this is at the top of Oregon's list:


One thing that would greatly help: swapping out the yield signs with stop signs.

A few other potential ways to improve safety at this roundabout. 

#1.  Convert it to a 4-leg roundabout and remove access from Wayside Ln. (traffic can divert to Manor Drive to enter onto Hayden Bridge Way). 

#2.  Redesign the central island to be a true spiral roundabout (don't just rely on pavement markings to do the job).

[Bickendan]

Agreed, but I think the number one improvement is stop signs. Yield signs in these situations just promotes drivers risking their chances and trying to beat a car already in the roundabout. Coe Circle uses stop signs, and while drivers do 'California stops' if there aren't any cars already in the roundabout, they do keep an eye out and wait their turn. It also makes it more bicycle friendly.

[jeffandnicole]

It helps to know where/how most of the accidents are happening.  If there's one entry point that's causing the vast majority of crashes, then they should try to deal with the problem rather than masking a solution.

[english si]

It helps to know where/how most of the accidents are happening.  If there's one entry point that's causing the vast majority of crashes, then they should try to deal with the problem rather than masking a solution.

Indeed, and without knowing the severity of the crashes, and the statistics now, they are meaningless. We're looking at 05-09 data there - modern roundabouts were very rare in the US around that period - could the answer be found in education failures rather than the engineering ones that Tradephoric thinks exist?

Junctions like this work all the time in the UK with near-zero levels of crashes - the existence of a round central island, or Wayside Lane probably are red herrings.

It also makes it more bicycle friendly.

STOP signs make nothing more bicycle friendly unless they don't have to - starting from a stop is not only tiring, but leaves the rider more erratic and vunerable as they build up the inertia that they lost by stopping (a yield isn't so bad if they just have to slow down to a slow speed, rather than come to a complete stop). The issue isn't the type of signs used anyway, it's drivers ignoring them...

It doesn't help matters that the US uses STOP like the boy who cried wolf - so YIELD is basically ignored because conflicts that should be simple yields demand you stop, therefore if it's a yield, there won't be any danger.

[jakeroot]

It doesn't help matters that the US uses STOP like the boy who cried wolf - so YIELD is basically ignored because conflicts that should be simple yields demand you stop, therefore if it's a yield, there won't be any danger.

I can't help but think that wider use of the yield sign might help with roundabout education. I don't think enough drivers encounter "yield" situations enough on their daily drive enough for them to comprehend them at roundabouts.



Of course, does anything really need to be done? If we keep tackling intersections on top of "the list", we'll eventually go through every intersection in the city. I've said this before, and I'll say it again: roundabouts have more crashes because drivers aren't as familiar with them as typical four-way junctions (Coe Circle is a four way junction in my book). The two ways to tackle increased collisions at roundabouts is to do;

A) nothing, and drivers will eventually figure out what to do (this could take 10 to 20 years); and/or
B) build more roundabouts so drivers become more familiar with them.

[Bickendan]

Junctions like this work all the time in the UK with near-zero levels of crashes - the existence of a round central island, or Wayside Lane probably are red herrings.The issue isn't the type of signs used anyway, it's drivers ignoring them...

It doesn't help matters that the US uses STOP like the boy who cried wolf - so YIELD is basically ignored because conflicts that should be simple yields demand you stop, therefore if it's a yield, there won't be any danger.

These hit the nail on the head. As a biker myself, I don't regard having to come to a complete stop at a red light or stop sign an issue. Bikers that complain about it are the ones that will blow through a red light anyway, and I don't have much sympathy for them when something happens.

[tradephoric]

Indeed, and without knowing the severity of the crashes, and the statistics now, they are meaningless.

The link broke down the roundabouts by crash severity (K, A, B, C, O).  The detailed crash severity data isn't "˜meaningless'.

We're looking at 05-09 data there - modern roundabouts were very rare in the US around that period —

There were over 2,000 modern roundabouts in America by the end of 2005.  Of course "˜very rare' is a vague term that makes your statement difficult to dispute.

could the answer be found in education failures rather than the engineering ones that Tradephoric thinks exist?  Junctions like this work all the time in the UK with near-zero levels of crashes - the existence of a round central island, or Wayside Lane probably are red herrings.

A roundabout outside Farndon, UK has seen 150 crashes since it was redesigned in June, 2012.  This is on par with some of the worst performing roundabouts cited on this thread.  The British people have had 50+ years to perfect yield-at-entry roundabouts and should be well educated in their use.


http://newarkadvertiser.co.uk/articles/news/A46-Farndon-roundabout-layout-will-be-changed

[english si]

The link broke down the roundabouts by crash severity (K, A, B, C, O).

But doesn't have current data? What use is 5-year data from years ago good for unless you also have current data to see whether figures have decreased or not? Even if there's not a downward trend as more people understand how the junction works, then you are still arguing for rebuilding the junction because 6 years ago there were a lot of crashes.

The detailed crash severity data isn't "˜meaningless'.

You are interpreting words worse than you interpret statistics: I explicitly said that without severity data, the figures are meaningless - not that severity data is meaningless.

There were over 2,000 modern roundabouts in America by the end of 2005.  Of course "˜very rare' is a vague term that makes your statement difficult to dispute.

Keep in mind, there are only about 40 high capacity 2x2 or 3x2 non-interchange roundabouts in America.

I am keeping that mind, and then you attack me for doing so! You harp on about multi-lane (and specifically those ones) roundabouts being dangerous death traps, but as well as the (inevitable) small samples, you aren't accounting for unfamiliarity.

A figure of 40 today suggests 'very rare' is highly unlikely to be false 10 years ago.

Of course it's not precise - I couldn't be bothered to do a detailed and time-consuming analysis of how many junctions in the US there were, what proportion of them were roundabouts and factor in regional data - if there was 20000 of them in Florida, then a handful in Oregon will still mean that they aren't commonly occurred by drivers in the area. However, 'very rare' is a qualitative phrase not a quantitative one and therefore more meaningful - your figure of over 2000 roundabouts is a fairly precise quantitative one, but it is meaningless without knowing whether in this context 2000 is big or small.

A roundabout outside Farndon, UK has seen 150 crashes since it was redesigned in June, 2012.  This is on par with some of the worst performing roundabouts cited on this thread.

I'm a roadgeek, and I went "where the fuck is that" (it's a tiny hamlet): about as descriptive as 'Bob's Farm, USA'. Thankfully you gave an article, so I could know that it is the Farndon Roundabout near Newark, Notts.

That roundabout sees 29900 Vehicles/average weekday on the new A46 leg with high levels of trucks (26200 a year after opening on the A46 north of the roundabout, 13200 on the B6166, 2950 on the old route), making this junction 13.2MEV in that first year and roughly 3 crashes/MEV. That is truly awful (though this, like the Oregon example above seems to be, is a clear outlier).

Looking, the most obvious safety improvement would be to put in the yellow rumble strips that exist almost every time a long (and many not-very long) grade-separated route ends at a roundabout. I can't help but think that 70mph+ traffic on a well-aligned freeway-quality trunk road coming arriving at the roundabout without realising it was that close is going to be a factor in why there is a high level of crashes there.

There's narrow lanes on, and an atypical (I think - it might be common to spiral unsignalised roundabouts in that neck of the woods, but I cannot think of any around here) design of, the circulatory carriageway.

There's also the merging issue just north of the roundabout, where two lanes of trunk road traffic (there was only one before, save for at the roundabout) are forced into one lane. And the article says that the surface was poorly done and anti-skid lost.

So lots of possible factors here as to why this roundabout has become a lot more dangerous - some of which you recommend as safety features, others of which you totally and completely ignore.

If you look at it on google's satellite imagery then you'll see that the old roundabout had a much smaller circulatory carriageway and 3 lanes on it - both of which have been your go-to "these are problems for safety".

And, of course, it really needed to be a GSJ (and was certainly seen in the planning as a fairly temporary junction, though no plans have been sorted on finishing the corridor)

[tradephoric]

A reoccurring argument in this thread is that crashes at American roundabouts are due to driver unfamiliarity.   It's the "dumb American"  argument.  Yet when a roundabout with a high crash rate is cited from the UK, the focus shifts entirely to design:

Looking, the most obvious safety improvement would be to put in the yellow rumble strips that exist almost every time a long (and many not-very long) grade-separated route ends at a roundabout. I can't help but think that 70mph+ traffic on a well-aligned freeway-quality trunk road coming arriving at the roundabout without realising it was that close is going to be a factor in why there is a high level of crashes there.

There's narrow lanes on, and an atypical (I think - it might be common to spiral unsignalised roundabouts in that neck of the woods, but I cannot think of any around here) design of, the circulatory carriageway.

There's also the merging issue just north of the roundabout, where two lanes of trunk road traffic (there was only one before, save for at the roundabout) are forced into one lane. And the article says that the surface was poorly done and anti-skid lost.

So lots of possible factors here as to why this roundabout has become a lot more dangerous - some of which you recommend as safety features, others of which you totally and completely ignore.

[tradephoric]

If you look at it on google's satellite imagery then you'll see that the old roundabout had a much smaller circulatory carriageway and 3 lanes on it - both of which have been your go-to "these are problems for safety".

Just to be clear, this is the roundabout in question right? 
https://www.google.com/maps/@53.06409,-0.83613,131m/data=!3m1!1e3

The old roundabout did not have 3-cirulating lanes of traffic.  In addition, the central island diameter of the old roundabout was 135 feet (much larger diameter than 100 feet).  IMO, the size of this roundabout looks pretty decent.

[Tarkus]

The link broke down the roundabouts by crash severity (K, A, B, C, O).

But doesn't have current data? What use is 5-year data from years ago good for unless you also have current data to see whether figures have decreased or not? Even if there's not a downward trend as more people understand how the junction works, then you are still arguing for rebuilding the junction because 6 years ago there were a lot of crashes.ugh no plans have been sorted on finishing the corridor)

The bit tradephoric showed in the post was the initial stats from when the roundabout first opened, from the City of Springfield's Transportation System Plan.  If you follow the link, you'll find that there's further stats from the period of 1/1/2009 to 12/31/2013 (the most recent that were available then).  The actual PDFs from the ODOT Crash Reporting system are also linked for the various legs.  The KABCO stats in the main post are an aggregate of all the legs during that latter reporting period.  Pioneer Pkwy W/Hayden Bridge had 46 total, Pioneer Pkwy E/Hayden Bridge had 78 total, and there's 57 more crashes associated with MLK/Hayden Bridge (whether this is at the NW corner or NE corner is unclear, due to how ODOT compiled things).

[jakeroot]

A reoccurring argument in this thread is that crashes at American roundabouts are due to driver unfamiliarity.   It's the "dumb American"  argument.  Yet when a roundabout with a high crash rate is cited from the UK, the focus shifts entirely to design:

Are you suggesting that UK drivers are unfamiliar with roundabouts? Or that US drivers understand them perfectly and that design is entirely to blame?

[tradephoric]

Are you suggesting that UK drivers are unfamiliar with roundabouts? Or that US drivers understand them perfectly and that design is entirely to blame?

I'm suggesting that the "˜driver unfamiliarity' argument is masking poor design.  Take another look at the aerial footage of the DePere roundabout.  At 4 seconds in the video, a yellow semi pulls up and waits for a gap in traffic to enter the roundabout.  The semi enters the roundabout at the 12 second mark and forces circulating vehicles to nearly come to a complete stop.  If one of the circulating drivers couldn't stop in time and struck the semi, the semi driver would be cited for "˜failure to yield.  Did the yellow semi really do anything wrong here?  During rush hours, do we expect the yellow semi to wait 15 minutes for traffic to clear all legs of the roundabout before pulling out?   It's a design that forces some drivers (IE. larger vehicles during rush hour) to play Russian Roulette (the bullet being a circulating driver not paying attention).

https://www.youtube.com/watch?v=xxkWMtAr4o8

The theory is if the roundabout was bigger (and there was greater distance between legs), circulating traffic will have more time to react when a semi tries to squeeze into the roundabout (and if properly designed, a larger roundabout doesn't mean circulating speeds are faster... they could actually be slower since entry deflection angles can be more pronounced at large diameter roundabouts).  Sure, some drivers will still enter the roundabout without even looking and run straight into a circulating vehicle, but some of the failure to yield crashes might be prevented with a larger central island diameter.  There has to be reasons why so many multi-lane roundabouts with central-island diameters of 100 feet or less have performed so poorly.  The idea that we just ignore failure to yield crashes and chalk it up to "˜driver unfamiliarity' doesn't sit well with me. 

[jeffandnicole]

Are you suggesting that UK drivers are unfamiliar with roundabouts? Or that US drivers understand them perfectly and that design is entirely to blame?

I’m suggesting that the ‘driver unfamiliarity’ argument is masking poor design.  Take another look at the aerial footage of the DePere roundabout.  At 4 seconds in the video, a yellow semi pulls up and waits for a gap in traffic to enter the roundabout.  The semi enters the roundabout at the 12 second mark and forces circulating vehicles to nearly come to a complete stop.  If one of the circulating drivers couldn’t stop in time and struck the semi, the semi driver would be cited for ‘failure to yield.  Did the yellow semi really do anything wrong here?  During rush hours, do we expect the yellow semi to wait 15 minutes for traffic to clear all legs of the roundabout before pulling out?   It’s a design that forces some drivers (IE. larger vehicles during rush hour) to play Russian Roulette (the bullet being a circulating driver not paying attention).

https://www.youtube.com/watch?v=xxkWMtAr4o8

The theory is if the roundabout was bigger (and there was greater distance between legs), circulating traffic will have more time to react when a semi tries to squeeze into the roundabout (and if properly designed, a larger roundabout doesn’t mean circulating speeds are faster… they could actually be slower since entry deflection angles can be more pronounced at large diameter roundabouts).  Sure, some drivers will still enter the roundabout without even looking and run straight into a circulating vehicle, but some of the failure to yield crashes might be prevented with a larger central island diameter.  There has to be reasons why so many multi-lane roundabouts with central-island diameters of 100 feet or less have performed so poorly.  The idea that we just ignore failure to yield crashes and chalk it up to ‘driver unfamiliarity’ doesn’t sit well with me. 

Looking at this video, the "circulating" vehicles were just coming out of a yield at the same time.  They paused for a few seconds, and everyone went on their merry way.  It's really not a huge deal.

You are always going to have examples where certain vehicles - such as trucks - are going to have it a little hard.  At a traditional 4 way intersection, trucks occasionally have to swing out from the right lane in order to make their right turn without going up and hitting anything on the curb.  If they hit something on the curb, or a vehicle in another lane, they'll be in the wrong then also.  Yes, intersections are designed when possible so that trucks don't have to do this.  But in tight street environments such as cities, there's no way around this happening.

You simply can't compensate for every possible scenario out there.

[tradephoric]

Looking at this video, the "circulating" vehicles were just coming out of a yield at the same time.  They paused for a few seconds, and everyone went on their merry way.  It's really not a huge deal.

It's not a huge deal to stop in the middle of a roundabout to let in a semi, but it's an annoyance that increases the likelihood of a crash.  Consider the following scenario. The car and semi pull into the roundabout at the same time.  Would u want the car to take 3.3 seconds or 5.8 seconds to travel the blue path?  The additional 2.5 seconds of reaction time at the larger diameter roundabout may allow the car to simply let off the gas to let the semi in (as opposed to coming to a stop in the middle of the roundabout). 

Would an additional 2.5 seconds of reaction time potentially prevent a crash?  I think so.

115 FT central island diameter:


160 FT central island diameter:

[jeffandnicole]

Looking at this video, the "circulating" vehicles were just coming out of a yield at the same time.  They paused for a few seconds, and everyone went on their merry way.  It's really not a huge deal.

It’s not a huge deal to stop in the middle of a roundabout to let in a semi, but it’s an annoyance that increases the likelihood of a crash.  Consider the following scenario. The car and semi pull into the roundabout at the same time.  Would u want the car to take 3.3 seconds or 5.8 seconds to travel the blue path?  The additional 2.5 seconds of reaction time at the larger diameter roundabout may allow the car to simply let off the gas to let the semi in (as opposed to coming to a stop in the middle of the roundabout). 

Would an additional 2.5 seconds of reaction time potentially prevent a crash?  I think so.

The f'ing cars didn't crash into the f'ing truck.  There was no potential crash that could've potentially been avoided.  What's your point?

[jakeroot]

I don't think roundabouts are the only place where trucks tend to try and squeeze into gaps that aren't big enough for them. But they are much larger than cars so they'll win. That, and sometimes with roundabouts, the throughput rate is so high that a reasonable gap for a truck may never form, even with a large roundabout (though perhaps a roundabout the size of the UK one above may permit gaps large enough...not sure).

Basically, I don't think this is a problem that needs solving.

[tradephoric]

The f'ing cars didn't crash into the f'ing truck.  There was no potential crash that could've potentially been avoided.  What's your point?

I'm trying to lay out an explanation why small diameter roundabouts have performed so poorly.  The truck pulling out into the roundabout forced circulating vehicles to take evasive action to avoid a crash (IE. applying the brakes and nearly coming to a complete stop in the roundabout).  If the roundabout had a larger diameter (and it took an additional 2.5 seconds for circulating vehicles to encounter the semi), evasive action could be avoided.  Relying on vehicles to take evasive action to avoid a crash isn't a great scenario and increases the potential of a crash.  In the end, 2x2 roundabouts with central island diameters of 100 feet or less have seen crashes increase by 500%, 800%, up to 1400%.  That's not a good track record.  The fact that the cars didn't hit the truck in the video doesn't invalidate the point.