Crash prone 'modern roundabouts'

[cl94]

One of the biggest reasons for the force-feeding of many roundabouts is to be due to federal HSIP funding structure. Many of those grants are tied specifically to the construction of roundabouts, due to the outdated IIHS information. Until a new study of similar size is done, good luck getting the feds to take off the rose-colored glasses. Unfortunately, no roundabout = no money for an improvement project, so many cash-strapped agencies will swallow the bitter roundabout pill in order to be seen as doing something to improve a safety problem.

This. It all boils down to money. Sometimes, roundabouts are a good solution (such as for a moderately-trafficked 5+ point intersection or low speeds). Other times, they aren't. Being on the research side, I can tell you that transportation research funding is very limited and, honestly, there are more important things to study right now where the money would have more impact.

[english si]

I'd love to post collision data from the several multi-lane roundabouts in my town. But as there haven't been any collisions at the roundabouts for over 20 years, there's no data to find.

[kphoger]

And I'm sure none of them have fixed objects in the island, all approaches have opaque fencing, all of them have exactly the same ideal traffic volumes, lines are never faded or obscured by snow......

I've really been waiting for a post from english si. His experience across the pond is strong evidence that roundabout crashes are a product of culture and education, not geometry and design.

[tradephoric]

However, I think you need to be willing to concede that roundabouts - even multilane roundabouts - do in general reduce the incidence of injury and fatality crashes, even if they don't prevent all of them (which has never been claimed by anyone, to my knowledge). In fact that's the key trade-off - with roundabouts, you get some more low-angle of incidence, lower-speed crashes in return for fewer 90-degree, high-speed crashes, which are the ones (particularly with today's safety technologies) are the most dangerous.

A roundabout with an ADT of 50,000 experienced 10% injury crashes last year.  A signalized intersection with an ADT of 50,000 experienced 50% injury crashes last year.  Which intersection had more injury crashes last year?  We can't answer that question without knowing the total crashes that occurred at both intersections.  You can argue that roundabouts on a percentage basis have fewer injury crashes, but it's a misleading statistics.  Maybe the roundabout had 100 crashes (10 leading to injury) and the traffic signal only had 12 crashes (6 leading to injury).  Under that scenario the traffic signal sounds much safer, even though 50% of the crashes at the signal led to injury. 

Let's start looking at injury crash rates, which account for total crashes, as opposed to the crash severity, which doesn't account for total crashes.  Like I have said before, if you have a big spike in total crashes at a roundabout - which is what we are seeing at these 2x2 and 2x3 configurations - there's no guarantee that there will be a reduction in injury crashes.

Does that mean I (and others) have callous disregard for the deaths and injuries that remain, as you have stated on several occasions? Of course not, and such an implication is frankly offensive. No intersection is perfect or (as DaBigE puts it) a "silver bullet," particularly when you have high traffic volumes. But unless you think that some human lives have more intrinsic value than other human lives, an option that results in fewer injuries and deaths overall is preferable to one that results in more of them, even if some deaths and injuries are going to happen either way. That doesn't bring back specific crash victims who might not have died in the counterfactual universe where there was a different intersection, but it does mean that more people didn't die in their place.

When someone suggests that only drunks and idiots are getting seriously injured and killed at roundabouts — which has been mentioned many times on this thread - then those comments do risk being construed as callous.  An argument made in ad nauseam is that it's not the fault of the roundabout when these crashes occur.  That line of thinking didn't stop engineers from replacing traffic signals with roundabouts for their supposed safety benefits.  How is it the fault of a traffic signal when a driver blows through a red light?  Maybe we should overlook red light running crashes when analyzing the safety benefits of traffic signals.  If that was the case there would be no need for roundabouts in the first place.

And I will say that if your beef boils down to (like kalvado's appears to boil down to) "transportation agencies are making false/exaggerated claims about total crash rates," this thread appears to have run its course and you need to address yourself elsewhere, because transportation agencies and their media representatives aren't reading random threads on Internet message boards to decide how they will communicate with the public.

Transportation agencies may be making false/exaggerated claims about injury and fatal crash rates as well.  When I pointed out that 3 of the last 4 fatal crashes in Carmel occurred at roundabouts, people were critical saying that there wasn't enough data to make any meaningful determination.  Of course nobody had a problem when the IIHS study concluded that roundabouts reduce fatal crashes by 90%, even though the study only analyzed 24 roundabouts.

But we need to study all of the roundabouts in America before we come to a conclusion.

This. Unless you're studying the entire population of interest, you're going to arrive at bad conclusions. Otherwise you end up being the guy on Twitter claiming that Trump is winning the election because he got a few million votes on a Drudge Report web poll.

Here's a population of over 5,100 modern roundabouts in America.  If you want, query out and analyze the roundabouts of interest.  It's dangerous to mention the C-word and T-word on here, but based on your previous predictions I wouldn't rule out either candidate.  Remember Dublin?

http://www.mediafire.com/file/e9ch50iu2afkh4x/Modern+Roundabouts+%28May-2015%29.kmz

[english si]

I've really been waiting for a post from english si. His experience across the pond is strong evidence that roundabout crashes are a product of culture and education, not geometry and design.

I'd argue that the main problem with US roundabouts is culture and education, but in the UK it's geometry and design if a roundabout has lots of crashes.

In the US, when the issue seems to be design, it's actually due to education reasons - assuming that American drivers are stupid, lazy and unobservant, and therefore designing poor practice to be proscribed by road markings is a recipe for disaster.

If we educate US drivers well, design the roundabouts to match that education, and create a driving culture that's able to deal with roundabouts, and comes into contact with them regularly, then roundabouts will certainly meet the claims that Tradephoric is attacking.

[DaBigE]

If we educate US drivers well, design the roundabouts to match that education, and create a driving culture that's able to deal with roundabouts, and comes into contact with them regularly, then roundabouts will certainly meet the claims that Tradephoric is attacking.

Bingo. How many drivers had roundabouts as part of their driver's ed program? Currently, only a small percentage, but thankfully it's growing [slowly] daily. Since there's no mandatory retraining, older drivers don't get the same education. PI blitzes only go so far. Arguably, driver's ed is the only time you can force someone to learn about a roundabout...after one receives their license, keeping up on laws and designs is completely voluntary. I wonder how many members of this forum have bothered to reread their agency's drivers' handbook? And if so, how long ago?

[cl94]

I've been saying that people should be retested every 5-10 years for a while. Would do a lot to stop the lazy habits and force them to take a driver's ed course as part of it.

[kphoger]

Maybe we should overlook red light running crashes when analyzing the safety benefits of traffic signals.  If that was the case there would be no need for roundabouts in the first place.

Perhaps I'm quoting this out of context, or more likely ignoring the tongue-in-cheek you were attempting to employ. None the less, I think it betrays your true bias: roundabouts are unnecessary. You do not believe them to be 'one tool in the bag,' as we have recently taken to calling them in this thread, but rather a superfluous tool that no real craftsman would ever buy (if you'll allow me to extend the metaphor) because it's substandard, underperforming, overpriced crap. Real craftsmen buy stoplights.

While I appreciate your efforts at putting roundabouts and stoplights together on the same table when it comes to crunching the numbers, I still believe your goal to be discounting roundabouts en masse by highlighting the worst of the data at hand.

Traffic control is necessary (although quite a bit less necessary than we seem to think here in America), be it Yield signs, Stop signs, signals, roundabouts, what have you. It's necessary because intersections exist. But to say that roundabouts are only necessary if _______ is true compared to stoplights... That's bias. Turn it around and ask if it has the same ring of truth or not: Stoplights are only necessary if _______ is true compared to roundabouts.

[kalvado]

I'd love to post collision data from the several multi-lane roundabouts in my town. But as there haven't been any collisions at the roundabouts for over 20 years, there's no data to find.

Well, you may want to add traffic numbers to that statement. Maybe both drivers in your town have to share same car, hence there are no accidents?

[english si]

Well, you may want to add traffic numbers to that statement. Maybe both drivers in your town have to share same car, hence there are no accidents?

Only A Roads have traffic counts, and some of these legs are measured away from the roundabout, after some side turns (which add traffic to the roundabout).
Roundabout 1 - legs of 11363, 16818, 18470, 16569. 31610
Roundabout 2 - legs of 18804, 20388, 18470, 26135. 41898
Roundabout 3 - legs of 14997, 14917, 26135. 28024
Roundabout 4 - legs of 14917, 14917, unknown. At least 14917 vehicles use the roundabout daily
Roundabout 5 - legs of 20388, 20388, unknown, unknown. There probably is a lazy bit of counting here, but at least 20388 vehicles use the roundabout daily (as one leg certainly has that figure).

A mix of traffic figures, but none are small.

[lordsutch]

There's a valid statistic for measuring injury and death rates at intersections - incidence per MEV, the same rate that tradephoric has been using to measure total crash rates. You don't need total crashes to calculate deaths per MEV or injuries per MEV.

Alternatively you could use a weighting scheme, like that described here that weights fatality crashes 12x and injury crashes 3x as "severe" as a PDO crash - of course, other weighting schemes are possible.

[tradephoric]

While I appreciate your efforts at putting roundabouts and stoplights together on the same table when it comes to crunching the numbers, I still believe your goal to be discounting roundabouts en masse by highlighting the worst of the data at hand.

Here's a timeline of how roundabouts have evolved in America...

1990-1999:  Roundabouts introduced to America
The first modern roundabout in America was constructed in Summerlin, Nevada in 1990.  Over the next decade, roughly 300 roundabouts are built throughout the country.  These were mostly single-lane or multi-lane roundabouts with 1x2 configurations.  A lot of literature existed in other countries but nobody really knew how these "guinea pig" roundabouts would perform in America.

2000-2005:  The "simple"  roundabout era
In March 2000, the IIHS releases a study entitled "Crash Reductions Following Installation of Roundabouts in the United States" .  This is taken directly from the study - "Of the 24 intersections studied, 21 were previously controlled by stop signs, and 3 were controlled by traffic signals. Fifteen of the roundabouts were single-lane circulation designs, and 9, all in Colorado, were multilane."   This routinely cited study concluded that roundabouts reduce total crashes by 39%, injury crashes by 76%, and fatal crashes by 90%.  After the release of the study, roundabouts gain further acceptance.  Agencies build more and more roundabouts, mimicking the "simple"  configurations analyzed in the IIHS study (single-lane and 1x2 multi-lane).  By 2005, there are roughly 2000 modern roundabouts in America.

2006-current:  The "complex"  roundabout era
By the mid 2000s, most consider it "settled science"  that roundabouts are safer.  Agencies feel emboldened to design more "complex" roundabouts with 2x2 configurations and even some triple-lane roundabouts.  The problem is these complex configurations were seeing abysmally high crash rates that weren't jiving with the 2000 IIHS study.  We are talking 10X increases in crashes; suffice it to say that is not a 39% decrease!  Today, there are roughly 5000 roundabouts in America and about 100 of them have 2x2 or 2x3 configurations. 

So the focus of this thread has largely been on the "complex" roundabouts with 2x2 and 2x3 configurations.  By highlighting a small subset of roundabouts that aren't working well, does that mean I'm trying to discredit all roundabouts en masse?

[kphoger]

No, it does not necessarily mean that. And there are posts of yours that show an interest (perhaps even genuine) in determining what factors cause certain roundabouts to perform poorly.

But what I was drawing from is a general perception (possibly misguided), that you consider a stoplight to be superior and a roundabout a alternative novelty that must be proven. The history of roundabouts in America is only useful in understanding roundabouts as a relatively modern concept; this on paper (screen?) supports the assertion that more research needs to be done, but it also suggests (intentionally or unintentionally) that roundabouts in general are a newfangled fad that's worth discrediting.

As english si has illustrated, however, there are plenty of places in the world where roundabouts have a longer and more established history, and the sort of problems we see here don't happen nearly to the same extent there. And all these ideas that get thrown around on here (median fencing, elimination of objects, specific striping and signage, large or small diameter, elimination of circulating lanes) are definitely not ubiquitous in these other places. Elsewhere in the world, roundabouts of all sorts function much better than here.

I think, in order to have a straight conversation on the topic, we need to consider those multi-lane roundabouts that perform well. You asked for an example, and english si provided you with a whole town. Now that you have the data you were seeking, can we assume you'll be using it for fair analysis?

[kalvado]

Well, you may want to add traffic numbers to that statement. Maybe both drivers in your town have to share same car, hence there are no accidents?

Only A Roads have traffic counts, and some of these legs are measured away from the roundabout, after some side turns (which add traffic to the roundabout).
Roundabout 1 - legs of 11363, 16818, 18470, 16569. 31610
Roundabout 2 - legs of 18804, 20388, 18470, 26135. 41898
Roundabout 3 - legs of 14997, 14917, 26135. 28024
Roundabout 4 - legs of 14917, 14917, unknown. At least 14917 vehicles use the roundabout daily
Roundabout 5 - legs of 20388, 20388, unknown, unknown. There probably is a lazy bit of counting here, but at least 20388 vehicles use the roundabout daily (as one leg certainly has that figure).

A mix of traffic figures, but none are small.

OK, to put things in perspective:
An number often mentioned in this thread is 0.7 crashes per 1 million vehicles. For the numbers you have, that means 1 crash a month to 4 crashes a year.
And crash doesn't mean that police had to use power tools to remove trapped people; it is something serious enough to call insurance; e.g. dent on a door...
Are you sure you can find data on events like that?

As a matter of fact, I can find a few accidents in Buckinghamshire  roundabouts attended by firefighters just within one month of 2014: Wednesday 23 July, 12.34am; Sunday 20 July, 2.41am; Tuesday 8 July, 11.23am.

I am not sure which roundabouts you had in mind; but if these collisions were on circles you mentioned, then accident rate for a given month is higher than on traffic light controlled intersections in US (and US drivers are indeed statistically less safe than UK)

[cl94]

Very few US intersections have leg traffic counts as high as some of those, either.

[english si]

An number often mentioned in this thread is 0.7 crashes per 1 million vehicles. For the numbers you have, that means 1 crash a month to 4 crashes a year.

Which is very high.

And crash doesn't mean that police had to use power tools to remove trapped people; it is something serious enough to call insurance; e.g. dent on a door...

Duh!

Are you sure you can find data on events like that?

I can, the problem is a lack of datums to create data. It's hard to find data when there's almost zero to find.

I found 2011 data that covers the busiest 4 roundabouts. 3 slight collisions, all at the busiest roundabout, none of which have anything to do with the junction being a roundabout, but have everything to do with the junction being a junction, the road they took place on being a slope, and the road being slippery at the time. Only 3 of the 6 cars have any damage (one collision was entirely damage free) and there were no injuries. But that's one year - and a big spike - and still fewer (and far less severe) than a mile long stretch of straight road with no junctions and only a couple of driveways up the top of the hill - we don't see a 889 post thread here on "crash prone straight roads" as there no vendetta against them.

3 collisions in a year at that roundabout is a collision rate of 0.196. Let's call it 0.2.

As a matter of fact, I can find a few accidents in Buckinghamshire roundabouts attended by firefighters just within one month of 2014: Wednesday 23 July, 12.34am; Sunday 20 July, 2.41am; Tuesday 8 July, 11.23am.

I've googled them and the two early hours (which suggests that the issue is less likely to be the existence of a roundabout rather than any other form of at-grade junction) ones aren't in Buckinghamshire. The other one was near that second busiest roundabout I mentioned, but wasn't actually at the junction itself.

Lets say we count that anyway and pretend that the crash was not only at the roundabout, but a monthly occurance that year (rather than the one off it was), just to skew things your way, and add that to the 0 crashes recorded in 2011 (despite the surfaces of roads in the area being so potholed as to nearly cause accidents, getting the council in trouble that year), and you get a value of 0.52 per million vehicles.

then accident rate for a given month

A skewed month designed to make your case, but one that can only find just 2 serious crashes at roundabouts in a very large town despite that town having about 100 roundabouts.

circles

If you don't even understand the terminology, what's the point in bothering to discuss it?

[kalvado]

An number often mentioned in this thread is 0.7 crashes per 1 million vehicles. For the numbers you have, that means 1 crash a month to 4 crashes a year.

Which is very high.

And crash doesn't mean that police had to use power tools to remove trapped people; it is something serious enough to call insurance; e.g. dent on a door...

Duh!

Are you sure you can find data on events like that?

I can, the problem is a lack of datums to create data. It's hard to find data when there's almost zero to find.

I found 2011 data that covers the busiest 4 roundabouts. 3 slight collisions, all at the busiest roundabout, none of which have anything to do with the junction being a roundabout, but have everything to do with the junction being a junction, the road they took place on being a slope, and the road being slippery at the time. Only 3 of the 6 cars have any damage (one collision was entirely damage free) and there were no injuries. But that's one year - and a big spike - and still fewer (and far less severe) than a mile long stretch of straight road with no junctions and only a couple of driveways up the top of the hill - we don't see a 889 post thread here on "crash prone straight roads" as there no vendetta against them.

3 collisions in a year at that roundabout is a collision rate of 0.196. Let's call it 0.2.

As a matter of fact, I can find a few accidents in Buckinghamshire roundabouts attended by firefighters just within one month of 2014: Wednesday 23 July, 12.34am; Sunday 20 July, 2.41am; Tuesday 8 July, 11.23am.

I've googled them and the two early hours (which suggests that the issue is less likely to be the existence of a roundabout rather than any other form of at-grade junction) ones aren't in Buckinghamshire. The other one was near that second busiest roundabout I mentioned, but wasn't actually at the junction itself.

Lets say we count that anyway and pretend that the crash was not only at the roundabout, but a monthly occurance that year (rather than the one off it was), just to skew things your way, and add that to the 0 crashes recorded in 2011 (despite the surfaces of roads in the area being so potholed as to nearly cause accidents, getting the council in trouble that year), and you get a value of 0.52 per million vehicles.

then accident rate for a given month

A skewed month designed to make your case, but one that can only find just 2 serious crashes at roundabouts in a very large town despite that town having about 100 roundabouts.

circles

If you don't even understand the terminology, what's the point in bothering to discuss it?

Just a few posts ago you were claiming

I'd love to post collision data from the several multi-lane roundabouts in my town. But as there haven't been any collisions at the roundabouts for over 20 years, there's no data to find.

Now we're talking numbers. See the change?

I did not look for any bad or good month. I opened first page I found, believe it or not.
Assembling full statistics out of these accidents reports.. You know, I am not paid for that.  My point was not to show that Buckinghamshire is oh-so-dangerous place. Point is, you perfectly demonstrated that safety issues may be less obvious than a casual observer would believe.

[tradephoric]

A serious accident occurred at a roundabout in Walker, Michigan. 
http://woodtv.com/2016/11/04/walker-pd-driver-may-have-suffered-medical-condition-before-crash/

Accidents are up since the roundabout opened in 2015:
http://www.mlive.com/walker/index.ssf/2016/07/roundabout_use_in_walker_bring.html

Here was the pre-roundabout condition.  Adding a protected left turn could have improved the safety of this intersection without building a costly roundabout (just my opinion):
https://www.youtube.com/watch?v=ZHD-ADZt7xs

[tradephoric]

Roundabout at Auburn and Main Still Causing Car Carnage
http://www.mystateline.com/news/roundabout-at-auburn-and-main-still-causing-car-carnage

https://www.youtube.com/watch?v=bvfS2fPTjPM

Going from single digit crashes to nearly triple digit crashes a year.  This roundabout is just so much safer for everyone.  Great job.  Just fantastic.

CRASH DATA:
2012 - 3 crashes (Jan-Jun)
2013 - roundabout constructed
2014 - 86 crashes
2015 - 91 crashes
2016 - 71 crashes (Jan-Oct)

[doorknob60]

Maybe doesn't add a lot to this discussion we haven't already heard, but ODOT just uploaded this today:

[UCFKnights]

A new "double" 2 lane roundabout was constructed right after a new bridge over I-75 in Gainesville, FL. Here's Google Street View during contruction:
https://www.google.com/maps/@29.6258591,-82.3883651,3a,75y,210.13h,68.7t/data=!3m6!1e1!3m4!1sPF5UUK8WoIjdeW7QewCnrw!2e0!7i13312!8i6656

I didn't take pictures as I first saw it tonight and it was dark out, but few cars are using it so far as its leading to a new development with only one store open and few people seemingly aware of the new route. Once this get some traffic, I can't imagine it being safe and having good flow.

I'll try to go back and get some pictures, but the bridge drops immediately into the roundabout, which has its first exit into another roundabout that exits to 2 roads and back into the roundabout, and the next exit into a stop sign facing the road I pictured (where the left turn lane enters the roundabout under construction). I'll have to get pictures, but this is probably the worst roundabout I've seen yet

[silverback1065]

Fun facts on Carmel roundabouts: http://www.indystar.com/story/news/local/hamilton-county/2016/11/17/carmels-opens-100th-roundabout/93966544/

[tradephoric]

http://www.gooddaycarmel-bepartofthepositive.com/single-post/2016/11/18/Carmel-celebrates-historic-opening-of-100th-roundabout

As mentioned, Carmel opened its 100th roundabout at Rangeline Road and Carmel Drive on November 17th, 2016.  But to be honest, I have my concerns about this one.  This 2x2 geometry reminds me of Rockford's "wreck-it"  roundabout or Ann Arbor's Ellsworth roundabout.  This quite possibly could be the first roundabout in Carmel to average 100 crashes a year (would be fitting being the 100th roundabout and all!).  According to the Carmel Police Department Annual Reports, this busy intersection averaged 27 crashes a year from 2012-2014.  We will have to wait and see how many crashes occur moving forward.

[silverback1065]

http://www.gooddaycarmel-bepartofthepositive.com/single-post/2016/11/18/Carmel-celebrates-historic-opening-of-100th-roundabout

As mentioned, Carmel opened its 100th roundabout at Rangeline Road and Carmel Drive on November 17th, 2016.  But to be honest, I have my concerns about this one.  This 2x2 geometry reminds me of Rockford's "wreck-it"  roundabout or Ann Arbor's Ellsworth roundabout.  This quite possibly could be the first roundabout in Carmel to average 100 crashes a year (would be fitting being the 100th roundabout and all!).  According to the Carmel Police Department Annual Reports, this busy intersection averaged 27 crashes a year from 2012-2014.  We will have to wait and see how many crashes occur moving forward.

ya i wouldn't be making grand pronouncements on things like this, we'll see if crashes go up. this intersection is the busiest in carmel at around 22,000 cars a day.  I was against this roundabout because it simply wasn't necessary. There are several multi-lane roundabouts in the city and 2 more will be going in on rangeline next year.  there will be 2 more multi-lane ones on towne road on the west side of town.

[tradephoric]

Armdale Roundabout safety probed
http://thechronicleherald.ca/metro/1195561-armdale-roundabout-safety-probed

The Armdale rotary in Halifax, Nova Scotia was completed in 1956 and was converted to modern roundabout standards in 2007 (although I'll note there is no truck apron at the roundabout).  A few years after the conversion, municipal staff noted a "sharp and sustained"  increase in accidents (jumping from the low 60s to the high 90s).  Here are a few recent news reports of injury accidents that have occurred at the roundabout:

http://www.cbc.ca/news/canada/nova-scotia/armdale-roundabout-crash-sends-3-to-hospital-with-serious-injuries-1.3116629
http://www.cbc.ca/news/canada/nova-scotia/armdale-roundabout-motorcyclist-crash-1.3300385