Crash prone 'modern roundabouts'

[english si]

What the fuck? No one drives 7 hours to work.

Indeed! And it's not like you don't get regular long trips to work in the UK. Add in that congestion is higher - even a thirty mile freeway trip on the M25 can take 90 minutes at rush hour and so this fatigue issue being higher in the US is nonsense.

And if you want large US-style countries with big distances, then Sweden has a lower fatality rate than the UK.

[kalvado]

With longer drives in UK - such as London-Glasgow being on par with what some people drive to work in US - that looks like a fairly reasonable thought.

What the fuck? No one drives 7 hours to work.

Just came across article about San Jose cops commuting from Reno NV. With real estate prices in Silicon valley, commute distance grows out of control...

[tradephoric]

The 161-Riverside Drive roundabout in Dublin, OH was opened on August 13rd and in 23 days has been the site of at least 10 crashes including at least one injury accident (Dublin Police twitter feed reported an injury motorcycle accident in the roundabout on September 5).

http://www.thisweeknews.com/content/stories/dublin/news/2016/09/06/route-161-riverside-roundabout-dublin-addressing-signs-after-accidents.html

This is what Dublin officials were saying before the roundabout was constructed:

"They improve safety.  We will reduce crashes.  It won't eliminate crashes.  The frequency of the crashes will be reduced and the severity will be reduced, of the crashes, meaning the number of injury related type crashes will be substantially reduced."

-Jeannie Willis, Engineering Manager, City of Dublin

Assuming the number of crashes continue at their current pace the roundabout would be the site of 159 crashes in the first year.  A roundabout in Lakeville, MN had a similar poor opening with reports of 7 crashes in the first week of operations.  Since that time, the Lakeville roundabout has been the site of 127 crashes in just 10 months (even 10 months out, that Lakeville roundabout is experiencing double digit crashes a month).  Is this Dublin triple-lane roundabout going to experience a similar fate as the Lakeville roundabout?  Maybe Jeannie Willis will understand that triple-lane roundabouts don't reduce total accidents (too bad she didn't make that clear to the public when promoting the roundabout before it was constructed).

[Rothman]

Assuming the number of crashes continue at their current pace the roundabout would be the site of 159 crashes in the first year.

You know what they say happens when you assume?

[tradephoric]

^^^
This is what i said when a roundabout in Lakeville MN had a lot of crashes shortly after opening...

After two months of operation, a double-lane roundabout in Lakeville, Minnesota has tallied up 44 crashes.  Traffic counts at the roundabout are about 30,000 vehicles per day.  This equates to a crash rate of 24.1 MEV.  The crash rate should drop as drivers become more familiar with the roundabout - but even still - this is a horrible initial result.  It will be interesting to track the crashes over time at this roundabout. 


http://sunthisweek.com/2015/11/05/number-of-lakeville-roundabout-accidents-drop/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+Thisweeklivecom+%28ThisweekLive.com%29

People responded saying it's too early to judge the effectiveness of the Lakeville roundabout based on just a few months of crash data.  That's a fair argument.  But now that the Lakeville roundabout has been around for a year, the roundabout is STILL experiencing double digit crashes a month.  So what started as a horrible initial result remains a very bad result even a year on.  Now we get to go through the same debate with this triple lane Dublin roundabout. 

Rothman, i have a question.  Can you cite a single example of a 2x2 or 2x3 roundabout that has a crash rate below 1.0 MEV?  That should be easy to do.  Besides, most signalized intersections - even extremely busy intersections - have a crash rate below 1.0.  Why is it so hard to find a multi-lane roundabout with a crash rate below 1.0 MEV?  So i'm waiting for your response.  You cite a 2x2 or 2x3 configuration roundabout that doesn't have vehicles crashing into each other on a near daily basis.  Jeannie Willis was trying to convince the public that total crashes will actually go down at this roundabout.  You probably still believe her.

EDIT:  That Lakeville roundabout was designed based on the lessons learned in the Richfield study.  It was designed based on the latest best practices for roundabout design in Minnesota yet you still had 44 people crashing into each other in the first 2 months and 127 crashes in the first 10 months.  What a shock.     Using standard arrows vs. fishhook arrows really hasn't seemed to solve anything.

[lordsutch]

Just came across article about San Jose cops commuting from Reno NV. With real estate prices in Silicon valley, commute distance grows out of control...

San Jose-Reno is 4-5 hours except on days with Tahoe traffic (I just drove that exact route in June). And you could ride the train. Or live a hell of a lot closer for the same money or less - California's generally expensive, but there's plenty of areas closer to San Jose that have cheaper real estate than Reno. Maybe these cops don't want to live in Modesto or Merced but that's another issue.

[tradephoric]

http://dublinohiousa.gov/dev/dev/wp-content/uploads/2014/02/SR-161-Riverside-Drive-Intersection-Summary.pdf

As a signalized intersection, the 161-Riverside Drive intersection had a crash rate of 0.98 MEV, or about 1 crash per month.  After 23 days as a roundabout, the intersection already has double digit crashes.  What sounds like a higher crash frequency to everyone...1 crash a month or 10 crashes in 23 days?  Rothman, I'm still waiting for you to cite a 2x2 or 2x3 configuration roundabout with a crash rate below 1.0 MEV.  Based on the current crash numbers, it's obviously not going to be the 161-Riverside Drive roundabout.

[tradephoric]

That's exactly what's being argued though.  Listen to what Jeannie Willis has to say about the expected safety benefits of a 3-lane roundabout currently under construction in Dublin, Ohio (6A touched on this roundabout previously in this thread).  What are the residents of Dublin to think if her predictions don't come to pass?  At that point, who cares right?  It will already be constructed.

Well, I can't speak for the specific roundabout being built; it's entirely possible that the existing intersection is so dangerous that it will reduce both the absolute frequency and severity of crashes (the data suggest some roundabouts do lead to less crashes than the intersections they replaced, and without systematically distinguishing between what the previous intersection type was it's hard to predict which will reduce crashes and which will lead to more crashes).

Assuming this is the intersection in question, given the weird angle of the existing intersection I suspect the multi-lane roundabout will be an improvement on both scores.

There have already been 10 crashes at the 161-Riverside roundabout in the first 23 days.  Even if there are only 2 more crashes over the next 342 days (365-23) the roundabout will have the same crash rate as the pre-roundabout condition.  Lordsutch, do you really believe there will only be 2 crashes at the 161-Riverside Drive roundabout over the next 342 days?  Personally, I suspect there will be more than 2 crashes over the next 342 days.

[Rothman]

All I was pointing out was that it was an obviously gross assumption you were making about a crash rate continuing at a current pace.  Anyone with any background in data analysis knows that such assumptions are quite tentative with any phenomenon.

That all said, I'm no expert in finding location-specific accident data, but I'd wonder about the roundabouts on NY 85 at Maher Road and Blessing Road, especially in the recent couple of years.  Just going by my own limited experience, it seems those 2 x 2 roundabouts have been working pretty smoothly.  Don't know what their crash rates are, but I'd hypothesize that they're at least lower than the one at NY 140.

Where did you get the continuing crash rate for the Lakeville roundabout?

[kalvado]

All I was pointing out was that it was an obviously gross assumption you were making about a crash rate continuing at a current pace.  Anyone with any background in data analysis knows that such assumptions are quite tentative with any phenomenon.

That all said, I'm no expert in finding location-specific accident data, but I'd wonder about the roundabouts on NY 85 at Maher Road and Blessing Road, especially in the recent couple of years.  Just going by my own limited experience, it seems those 2 x 2 roundabouts have been working pretty smoothly.  Don't know what their crash rates are, but I'd hypothesize that they're at least lower than the one at NY 140.

Where did you get the continuing crash rate for the Lakeville roundabout?

If I remember correctly, there was a table in this thread with one of those rt.85 circles making into top 10 nationwide... I would be more than happy to be proven wrong, though.

[UCFKnights]

Well.. tiny problem is that we're not talking about NASCAR track, this is all about general purpose road. And it has to accomodate not only men in their top shape, but everyone with a license. Including that grey-haired lady in her 70s...

In my experience, grey haired ladies in their 70s (and 80s) tend to be those who are most consistently good at signalling at roundabouts. And I'm thinking about ladies that are not spritely (though aren't infirm) and probably shouldn't be on the road due to other issues (vision in the dark, not caring about dints and stuff, etc).

If you can't react quickly enough to signal your exit, then - to put it bluntly - you are a car crash waiting to happen and shouldn't be on the road. If the car in front stops suddenly, you'd be going into the back of it!

The big issue with signalling is that 'expert' engineers want turbo roundabouts, but then stripe them like regular 2-lane roundabouts and wonder why people, after they go "don't signal your exit", get confused as to whether a car is exiting and so are unsure and have crashes. But lets go with this idea that drivers are incompetent morons who shouldn't be thought of as capable of doing simple driving tasks unless they are professionals, so why inform them how to use junctions safely. I mean that can't lead to trouble... 

Assume other drivers are morons and you won't be disappointed.

Ah, the American way, sadly you will be disappointed as your fatalities per billion km is nearly twice that of the UK (and the UK has stagnated in road safety since it started treating drivers as morons) where they (used to) assume drivers are able to drive cars in a way that doesn't create crashes if they were told how to do so.

Out of curiosity, how is a turn signal supposed to help change anyone's behavior in the roundabouts? I signal everywhere except roundabouts, because it isn't intuitive to do so (my car will put out maybe two blinks, probably one, and its surely not helping the person next to me who can't see it, the person behind me likely doesn't care since its not indicating I'm going to be slowing down or anything that will affect them, and the angle of the roundabouts doesn't allow the person at the next entry to see my right turn signal indicating I'm leaving the roundabout until its useless.

[jeffandnicole]

Well.. tiny problem is that we're not talking about NASCAR track, this is all about general purpose road. And it has to accomodate not only men in their top shape, but everyone with a license. Including that grey-haired lady in her 70s...

In my experience, grey haired ladies in their 70s (and 80s) tend to be those who are most consistently good at signalling at roundabouts. And I'm thinking about ladies that are not spritely (though aren't infirm) and probably shouldn't be on the road due to other issues (vision in the dark, not caring about dints and stuff, etc).

If you can't react quickly enough to signal your exit, then - to put it bluntly - you are a car crash waiting to happen and shouldn't be on the road. If the car in front stops suddenly, you'd be going into the back of it!

The big issue with signalling is that 'expert' engineers want turbo roundabouts, but then stripe them like regular 2-lane roundabouts and wonder why people, after they go "don't signal your exit", get confused as to whether a car is exiting and so are unsure and have crashes. But lets go with this idea that drivers are incompetent morons who shouldn't be thought of as capable of doing simple driving tasks unless they are professionals, so why inform them how to use junctions safely. I mean that can't lead to trouble... 

Assume other drivers are morons and you won't be disappointed.

Ah, the American way, sadly you will be disappointed as your fatalities per billion km is nearly twice that of the UK (and the UK has stagnated in road safety since it started treating drivers as morons) where they (used to) assume drivers are able to drive cars in a way that doesn't create crashes if they were told how to do so.

Out of curiosity, how is a turn signal supposed to help change anyone's behavior in the roundabouts? I signal everywhere except roundabouts, because it isn't intuitive to do so (my car will put out maybe two blinks, probably one, and its surely not helping the person next to me who can't see it, the person behind me likely doesn't care since its not indicating I'm going to be slowing down or anything that will affect them, and the angle of the roundabouts doesn't allow the person at the next entry to see my right turn signal indicating I'm leaving the roundabout until its useless.

When I have seen turn signals in use, especially the right signal, just that one blink is enough to let me know what that driver should be doing.  I'll still wait though to make sure they commit to the turn...just in case!

[Rothman]

All I was pointing out was that it was an obviously gross assumption you were making about a crash rate continuing at a current pace.  Anyone with any background in data analysis knows that such assumptions are quite tentative with any phenomenon.

That all said, I'm no expert in finding location-specific accident data, but I'd wonder about the roundabouts on NY 85 at Maher Road and Blessing Road, especially in the recent couple of years.  Just going by my own limited experience, it seems those 2 x 2 roundabouts have been working pretty smoothly.  Don't know what their crash rates are, but I'd hypothesize that they're at least lower than the one at NY 140.

Where did you get the continuing crash rate for the Lakeville roundabout?

If I remember correctly, there was a table in this thread with one of those rt.85 circles making into top 10 nationwide... I would be more than happy to be proven wrong, though.

That was one at NY 140 that I mentioned and, although accidents were indeed frequent in the months after it opened, I'm betting the crash rate has come down even at that one.  That said, it's still a white-knuckler, mainly due to drivers who don't expect you to go "straight" from the inner lane as they try to enter the roundabout.  Have to say it's been a long while since some turkey tried to go around the roundabout from the outer lane, though (i.e., and almost hit me as I proceed "straight" from the inner lane).

[tradephoric]

That all said, I'm no expert in finding location-specific accident data, but I'd wonder about the roundabouts on NY 85 at Maher Road and Blessing Road, especially in the recent couple of years.  Just going by my own limited experience, it seems those 2 x 2 roundabouts have been working pretty smoothly.  Don't know what their crash rates are, but I'd hypothesize that they're at least lower than the one at NY 140.

The roundabouts at NY 85 / Maher Road and NY 85 / Blessing Road are 1x2 roundabouts.  They aren't true 2x2 roundabouts.  The "failure to yield"  misconception prevalent at 2x2 roundabouts is practically non-existent at 1x2 roundabouts.  Drivers seem to manage pretty well at these simplified 1x2 multi-lane roundabouts.

The 85/140 roundabout is a different animal.  It has two-entry lanes at all 4-legs of the roundabout (making it a true 2x2 roundabout).   It was built in 2007 and went from averaging 9.6 crashes a year before the roundabout to 38.3 crashes after.  The new Slingerlands Bypass was part of the roundabout construction so it's expected that crashes would go up.  That's why it's important to look at crash rates which accounts for any increases in traffic volumes.  Based on the most current crash data, the 85/140 roundabout has a crash rate of 2.62 (assuming 38.3 average crashes and an ADT of 40,000).  Keep in mind that a typical crash rate for a signalized intersection is only 0.8.   Maybe things have improved as of late at this roundabout, but I'll wait for actual crash data before making the determination.

http://www.timesunion.com/local/article/The-roundabout-Is-it-a-vicious-circle-1440833.php

Where did you get the continuing crash rate for the Lakeville roundabout?

From here:  http://sunthisweek.com/2016/07/15/some-drivers-still-adjusting-to-lakeville-roundabout/

There were 44 crashes in the first 2 months after the Lakeville roundabout opened on August 31, 2015.  OK fine.  Big spikes in crashes are to be expected as people get use to the new configuration.  But in May of 2016 there were 19 reported accidents and 12 crashes in the first 22 days of June.  Drivers had nearly a year to get use to the operations and there are still almost 20 crashes a month.  That's crazy.  It's concerning that the Lakeville roundabout is experiencing so many crashes since it was designed based on the lessons learned in the Richfield study.  If the Lakeville roundabout is a failure, what 2x2 roundabout is going to be a success?

[Rothman]

That all said, I'm no expert in finding location-specific accident data, but I'd wonder about the roundabouts on NY 85 at Maher Road and Blessing Road, especially in the recent couple of years.

The roundabouts at NY 85 / Maher Road and NY 85 / Blessing Road are 1x2 roundabouts. 

Oh, duh.  My bad.  That's me trying to think off-the-top of my head on a Monday morning.

Where did you get the continuing crash rate for the Lakeville roundabout?

From here:  http://sunthisweek.com/2016/07/15/some-drivers-still-adjusting-to-lakeville-roundabout/

Thanks.

Also, I'll take the time savings from the building of the NY 85/NY 140 roundabout.  The traffic signals were horrible.

[cl94]

Also, I'll take the time savings from the building of the NY 85/NY 140 roundabout.  The traffic signals were horrible.

You can say that about most of the roundabouts in this area. The Glens Falls roundabout used to be a 5-way signalized intersection. Split phasing. Miss the light and you could be sitting there for a few minutes.

[tradephoric]

Here's an ABC6 news story that indicates the new Dublin roundabout has experienced 15 crashes in the first 20 days.  More than first thought. 

"While it is always the City's goal that no crashes occur, it is worth nothing that 99.999% of vehicles have traveled through the roundabout without a crash report."
http://abc6onyourside.com/news/local/drivers-still-getting-used-to-dublin-roundabout-causing-some-traffic-confusion

That statement is shameful.  Instead of acknowledging the high number of crashes that have occurred at the roundabout, the city spins some BS 99.999% statistic.  Not that the 99.999% statistic isn't truthful, but it's incredibly misleading to the public.  They don't want to say that the current crash rate of the roundabout is 10MEV... but just do some simple arithmetic:

1 million — (99.999% of 1 million) = 10 MEV

Crash rate before roundabout = 0.98
Crash rate 20 days after roundabout = 10.0

The average crash rate for major 2x2 roundabouts in America is roughly 4.0 MEV.  If i had to guess, this Dublin roundabout will settle down around there.  When you have 4x the number of crashes, there's no guarantees you will see a reduction in injury accidents.  Is this Dublin roundabout safer than the signalized intersection it replaced?  I'm not at all convinced.

[lordsutch]

The average crash rate for major 2x2 roundabouts in America is roughly 4.0 MEV.  If i had to guess, this Dublin roundabout will settle down around there.  When you have 4x the number of crashes, there's no guarantees you will see a reduction in injury accidents.  Is this Dublin roundabout safer than the signalized intersection it replaced?  I'm not at all convinced.

Your one-man crusade against public relations officers' failure to unduly alarm the public on the basis of 20 days of data is noted. What that has to do with the likely injury or death rate at this particular intersection remains unclear.

[tradephoric]

Well, I can't speak for the specific roundabout being built; it's entirely possible that the existing intersection is so dangerous that it will reduce both the absolute frequency and severity of crashes (the data suggest some roundabouts do lead to less crashes than the intersections they replaced, and without systematically distinguishing between what the previous intersection type was it's hard to predict which will reduce crashes and which will lead to more crashes).

Assuming this is the intersection in question, given the weird angle of the existing intersection I suspect the multilane roundabout will be an improvement on both scores.

^^^
Remember that lordsutch?  A year ago you suspected the Dublin roundabout would reduce the absolute frequency of crashes.  There have been 15 crashes in just 20 days of the new roundabout opening.  Before the roundabout, the intersection averaged 12 crashes a year.  What ever happened to this reduction in crash frequency?   

Your one-man crusade against public relations officers' failure to unduly alarm the public on the basis of 20 days of data is noted. What that has to do with the likely injury or death rate at this particular intersection remains unclear.

Since your one prediction was wrong let's shift focus to injury crashes and death.  It's almost as if in your own mind you never suggested that the Dublin roundabout would reduce crash frequency, even though you did.  Anyways, on September 5th the Dublin Police department reported an injury motorcycle crash at the new roundabout.  It wouldn't shock me if the Dublin roundabout averaged an injury crash per month over the next year.  As I recall the Jeffersonville, Indiana roundabout had 9 injury crashes in the first 10 months of operation.  Both the Dublin and Jeffersonville roundabouts are confusing triple-lane clusters of fun and could end up with similar injury crashes.  I wouldn't bet against it.  Before you take that bet lodrsutch, just remember it only took you 20 days to be wrong about the frequency of crashes.

[cl94]

Well, I can't speak for the specific roundabout being built; it's entirely possible that the existing intersection is so dangerous that it will reduce both the absolute frequency and severity of crashes (the data suggest some roundabouts do lead to less crashes than the intersections they replaced, and without systematically distinguishing between what the previous intersection type was it's hard to predict which will reduce crashes and which will lead to more crashes).

Assuming this is the intersection in question, given the weird angle of the existing intersection I suspect the multilane roundabout will be an improvement on both scores.

^^^
Remember that lordsutch?  A year ago you suspected the Dublin roundabout would reduce the absolute frequency of crashes.  There have been 15 crashes in just 20 days of the new roundabout opening.  Before the roundabout, the intersection averaged 12 crashes a year.  What ever happened to this reduction in crash frequency?   

Your one-man crusade against public relations officers' failure to unduly alarm the public on the basis of 20 days of data is noted. What that has to do with the likely injury or death rate at this particular intersection remains unclear.

So before you suggested the new Dublin roundabout would reduce total crash frequency; but since it's becoming obvious that prediction wont hold water let's shift the focus to injury crashes and death.  It's almost in your own mind you never suggested that the Dublin roundabout would reduce crash frequency, even though you did.  Anyways, on September 5th the Dublin Police department reported an injury motorcycle crash at the new roundabout.  It wouldn't shock me if the Dublin roundabout averaged an injury crash per month over the next year.  As I recall the Jeffersonville, Indiana roundabout had 9 injury crashes in the first 10 months of operation.  Both the Dublin and Jeffersonville roundabouts are confusing triple-lane clusters of fun and could end up with similar injury crashes.  I wouldn't bet against it.  Before you take that bet lodrsutch, just remember it only took you 20 days to be wrong about the frequency of crashes.

Crashes typically spike with a change in traffic patterns, especially a significant change. We need to wait a couple years for things to stabilize. The SPUI nearby took 5-7 years to stabilize. Also won't be nearly as bad when the southbound bypass reopens: SB through traffic typically bypasses the intersection altogether, so traffic counts are much higher than normal.

[UCFKnights]

I was reading today that as part of a major redevelopment project, this 6 way intersection  in Winter Park, FL https://www.google.com/maps/@28.5860923,-81.3649353,3a,75y,145.46h,70.31t/data=!3m6!1e1!3m4!1sDsivZ_qYf5EMmiP8uSUI5g!2e0!7i13312!8i6656 is proposed to be converted to a presumably multilane roundabout, likely with 6 exits. Since people can't figure out the lane usage in 4 way 2x2 roundabouts, I can't imagine how they'd figure it out in a 6 way one. Is there  examples of modern roundabouts with more then 4 exits/entrances?

This project is still just a proposal from a developer who bought up a lot of the land around the intersection, I haven't seen any plans yet. I haven't found data for it, but I believe it is currently already a bit of a higher crash intersection.

[cjw2001]

You won't have to worry about comparing roundabouts to the traffic signals in Carmel much longer.   Most of the remaining traffic signals are on their way out.

[tradephoric]

Crashes typically spike with a change in traffic patterns, especially a significant change. We need to wait a couple years for things to stabilize. The SPUI nearby took 5-7 years to stabilize. Also won't be nearly as bad when the southbound bypass reopens: SB through traffic typically bypasses the intersection altogether, so traffic counts are much higher than normal.

I get what you are saying but the design life of a roundabout is only so long.  Consider the State & Ellsworth roundabout in Ann Arbor.  Before the roundabout the intersection was averaging about 20 crashes per year.  In 2013 the roundabout was built and from 2013-2015 the intersection had 351 crashes.  If nothing had been done to the intersection it would have taken roughly 18 years to rack up the same number of crashes that the roundabout racked up in just 3 years.  In addition, the number of injury crashes in 2014 was higher than the injury crashes that occurred in the 3 years before the roundabout was built.  It's just a bad result all around.  It's unlikely the State & Ellsworth roundabout will ever get back down to 20 crashes per year.  The design life of the roundabout will expire before that happens. 


http://semcog.org/Data-and-Maps/High-Frequency-Crash-Locations/Point_Id/81016689/view/RoadIntersectionCrashDetail

*I didn't update the chart, but there were 3 injury crashes at the roundabout in 2015.  Maybe nobody else finds this chart problematic but i do.  There was a huge increase in total crashes and no improvement in injury crashes.   A lose-lose from a safety perspective. 

[cl94]

Of course, another big variable is area population and its effect on traffic volumes. The effect of volume on congestion and accidents is nonlinear and correlation does not always imply causation. In Ann Arbor, there was a pretty significant population growth between the 2010 census and the most recent estimate. Take the Bethlehem example on NY 85/140 that was cited earlier. Not only does that involve a new roadway, but the area has grown quite significantly over the past decade.

I do agree that a higher number of accidents is troubling, especially if it is higher than would be expected if the prior geometry remains. But, at the same time, if we can expect a roundabout to last 20-30 years, it could stabilize at 30% lower than projections for the prior design and still be a net benefit. Typically, the number of total accidents is relatively constant from year to year if corrected for population change. I'm really curious about where the roundabouts will stabilize.

[tradephoric]

I was reading today that as part of a major redevelopment project, this 6 way intersection  in Winter Park, FL https://www.google.com/maps/@28.5860923,-81.3649353,3a,75y,145.46h,70.31t/data=!3m6!1e1!3m4!1sDsivZ_qYf5EMmiP8uSUI5g!2e0!7i13312!8i6656 is proposed to be converted to a presumably multilane roundabout, likely with 6 exits. Since people can't figure out the lane usage in 4 way 2x2 roundabouts, I can't imagine how they'd figure it out in a 6 way one. Is there  examples of modern roundabouts with more then 4 exits/entrances?

The one that comes to mind is the multi-lane Five-Points roundabout in Valparaiso, Indiana.  According to the city, in the 3 years before the roundabout the intersection had 61 crashes involving 11 personal injuries.  The roundabout was constructed in 2013 and had 25 crashes that year.  In 2014, the first full year of operation, the roundabout had 98 crashes.  That's obviously a big increase in total crashes but I don't have any injury crash data.  If only 5% of those 98 crashes resulted in an injury, it would exceed the injury crash rate before the roundabout was built.  Roundabouts aren't void of injury crashes and assuming only 5% of the crashes would lead to an injury is a pretty reasonable assumption.

Valparaiso Engineering Director Tim Burkman said while it's not pleasing to see that amount of accidents, it's important to note the severity of those 98 collisions were for the most part quite minor. Many fender benders involving rear ends or side swipes were reported, with more than half of the incidents involving reporting a damage estimate of less than $2,500 and 90 percent having a damage estimate of less than $5,000 -- estimate accounts for damage sustained by both vehicles.
The average daily traffic at the roundabout is 34,112 vehicles per day.
"Over the course of the entire year, that's an estimated total of 1.24 million vehicles that pass through five points," Burkman said. "This equates to an accident rate of one per 127,000 vehicles that traveled through the intersection, which is still considered to be quite low."

http://www.nwitimes.com/news/local/porter/nearly-accidents-at-five-points-roundabout-in/article_3aa6e8e3-5fc3-5a0c-9884-bc7c2dedc36a.html

Seriously Tim?  One crash per 127,000 vehicles equates to a crash rate of 7.87 MEV.  That isn't considered "quite low"  in the engineering community.  In fact, a crash rate of 7.87 MEV is nearly 10x higher than the crash rate of a typical signalized intersection.  Either Tim Burkman is a complete moron who got his degree from the back of a cracker-jack box or he's lying to the people of Valparaiso.