Crash prone 'modern roundabouts'

[kalvado]

Okay that was actually pretty funny.

But, you still seem to be taking this whole "force drivers to slow down" thing a little too seriously.

Try looking at in a bigger context.
If you know thee is an obstacle on a road, you're slowing down. Roundabout, speed bump, intersection, "object on road ahead".. Just knowing something may force you doing things.. Same thing as cop saying "make that music quieter" - no need to use actual force.
With that, those DUI (designed under influence) approaches to roundabout  send a good reminder to someone driving that roundabout on a regular basis - forcing them to prepare for safe pass. But at the same time DUI approaches hide an obstacle from a stranger. This is unlike traffic light, where approaches are mostly straight, and lights can be seen from a distance. And low visibility on approach can lead to higher accident rate etc. This is not unlike DUI roundabout approaches.
So yes - roundabouts try to force drivers to slow down.. but the message can easily be misunderstood.

[jakeroot]

So, basically, it's easier to run a red light than to run a roundabout?

[kalvado]

So, basically, it's easier to run a red light than to run a roundabout?

That depends on what you mean by "run".
Find yourself in a middle of intersection when you didn't expect that? Probably more difficult traffic light where visibility is not artificially engineered.
Enter intersection deliberately  disregarding interests and intentions of other drivers regardless of right of way? Fairly common for traffic lights as drivers accelerate on yellow, and basically design behavior for roundabouts... 

[jakeroot]

Find yourself in a middle of intersection when you didn't expect that? Probably more difficult traffic light where visibility is not artificially engineered.

What?

[JJBers]

So, basically, it's easier to run a red light than to run a roundabout?

That depends on what you mean by "run".
Find yourself in a middle of intersection when you didn't expect that? Probably more difficult traffic light where visibility is not artificially engineered.
Enter intersection deliberately  disregarding interests and intentions of other drivers regardless of right of way? Fairly common for traffic lights as drivers accelerate on yellow, and basically design behavior for roundabouts...

I don't understand half of this

LGL52VL

[kalvado]

Find yourself in a middle of intersection when you didn't expect that? Probably more difficult with traffic light where visibility is not artificially engineered.

What?

sorry, missed "with"

[jakeroot]

Find yourself in a middle of intersection when you didn't expect that? Probably more difficult with traffic light where visibility is not artificially engineered.

What?

sorry, missed "with"

Ahh okay.

While it is easier to "get away" with not stopping at a signal, not doing so can be catastrophic. Now, I'm not sure if that's something that actually happens a lot. We know at multi-lane roundabouts, crashes seem to go up. But are they the result of small things like failure to yield, or something larger like driving over the center circle? I don't think people driving over the center of the circle is very common.

It's possible to get away with running a red light. It's not possible to get away with failing to yield to the central island. I'll give you that. But how often is this an issue?

[kalvado]

Find yourself in a middle of intersection when you didn't expect that? Probably more difficult with traffic light where visibility is not artificially engineered.

What?

sorry, missed "with"

Ahh okay.

While it is easier to "get away" with not stopping at a signal, not doing so can be catastrophic. Now, I'm not sure if that's something that actually happens a lot. We know at multi-lane roundabouts, crashes seem to go up. But are they the result of small things like failure to yield, or something larger like driving over the center circle? I don't think people driving over the center of the circle is very common.

It's possible to get away with running a red light. It's not possible to get away with failing to yield to the central island. I'll give you that. But how often is this an issue?

Well, thing is this recent surge of activity in this never dying thread started with a video where this exact type of event is compiled.
And my point is that such accident is very unlikely to happen with commuter negotiating that roundabout 10 times a week - and I would buy  "force drivers to slow down" thing with respect to commuters. However for non-local drivers, roundabout with engineered limited visibility on approach and messy signs is quite conductive for the type of event.
And I would argue that such limiting visibility and often sign overload is pretty much today's trend. Need to know, access control, fake news,  all that crap.

On a bigger scale, it is actually an interesting question of designing for commuter vs design for a stranger. First time that occurred to me when I got slightly lost in MTA ticket machine controls, and few days later came across an article describing that design as optimized for long-term resident. Average NYCer can go through the process in like 10 second, as next step is always in a right spot; stranger may struggle and have to start over as "back" button is not even implemented. But that is still net benefit in terms of machine throughput due to local/stranger ratio.
Once I started to pay attention, I realized it is not an uncommon scenario...
Can it be that roundabouts are working the same way - reducing commuter accident rate at a cost of stranger accident rate? I wouldn't be surprised, but collecting statistics is quite difficult...

[jakeroot]

Can it be that roundabouts are working the same way - reducing commuter accident rate at a cost of stranger accident rate? I wouldn't be surprised, but collecting statistics is quite difficult...

I have been convinced, pretty much since day one, that if we introduced more roundabouts, the average crash rate would drop when the data from all are compiled.

The only way I can think of figuring out who in a collision is a stranger or a commuter is to consider the address of the involved, and the time of the collision. If it's within, say 30 miles, and the collision occured during rush hours, there's a decent chance that it could be a "commuter". But there's no way to be sure. I don't think that's a statistic that you could collect, short of police directly inquiring. Of course, people wouldn't be required to answer the question.

[kalvado]

Can it be that roundabouts are working the same way - reducing commuter accident rate at a cost of stranger accident rate? I wouldn't be surprised, but collecting statistics is quite difficult...

I have been convinced, pretty much since day one, that if we introduced more roundabouts, the average crash rate would drop when the data from all are compiled.

The only way I can think of figuring out who in a collision is a stranger or a commuter is to consider the address of the involved, and the time of the collision. If it's within, say 30 miles, and the collision occured during rush hours, there's a decent chance that it could be a "commuter". But there's no way to be sure. I don't think that's a statistic that you could collect, short of police directly inquiring. Of course, people wouldn't be required to answer the question.

Small problem is that designs shouldn't be based on being convinced, personal beliefs, religion of parents and similar stuff. Data, understanding patterns, analyzing existing dependences and so on - that is a proper basis for the design.
As for assumption of stranger/commuter ratio.. First problem is figuring out ratio in every day traffic. I can see ways of doing that, but it is time and money..
But if (and that is just hypothetical if) assumption is correct, that should give a hint about some design criteria - like placement of roundabouts near  locations frequently visited by non-locals (tourist attractions, some government offices, hospitals..) should face additional scrutiny. Or considering changes in warning patterns. Or..
But I am wasting time - such details are probably beyond the level of competence  of most engineering workforce.

[US 89]

At small intersections, roundabouts work great. They often enable more cars to get through the intersection and avoid huge lines forming behind a stop sign while the cars in front wait for a gap, often darting out in front of cross traffic and causing crashes.

That said, when you start having more than one lane in roundabouts, then things start to get messy. At one particular roundabout in my area, it seems that nobody can stay in their lane. (The lanes have been reconfigured since this image was taken, but it is still an issue.

The idea of a roundabout is even worse if the roundabout replaced a traffic signal. That automatically decreases the potential volume, since signals are designed to accomodate much more traffic than roundabouts can. (Exception: there was very little traffic and therefore no need for a signal.)

And no matter what, there will always be a few people who don't understand that you're supposed to yield to cars that are already in the roundabout.

[tradephoric]

Many experts believe roundabouts should be safer than the intersections they are replacing.  And if a roundabout does have an excessive number of crashes it's easy to blame it on poor signage, pavement markings, or deflection angles.  But their expectations about roundabout safety are largely based on an IIHS study done in 2000 — at a time when complex roundabouts were nearly non-existent.  Now agencies are building these complex 2x2 or 3x2 roundabouts believing they will be safer simply because they are a roundabout.   That is a dangerous mindset.   

Listen to Mark McCulloch's remarks at the 5th International Conference on Roundabouts.  Mark is a project engineer for the Washtenaw County Road Commission and was the man behind the State & Ellsworth roundabout in Ann Arbor.  Before it was constructed he was cocky that the roundabout would work great.  Now reality has set in...

So my first roundabout that I put in was in 2007 and you know it's going really well — we're saving lives, we're doing capacity, things are just going great, and then 2013 my nemesis showed up.  I did my first two lane 2-by-2 multi-lane roundabout with high capacities — we're talking about 30,000 vehicles coming in and out of this thing.  And I remember sitting at a public forum and I'm telling 200 people that it's just going to work great.  I'm sitting up there and I'm very confident, maybe even a little bit too cocky, and I'm saying how great it's going to be, you'll see.  And then the first year that it opened, the first full year that it was opened to traffic, we had 170 fender bender crashes at this roundabout. 

We had fifteen - um typically we had 15 injury crashes at this intersection when it was a traffic signal.  You know from a standpoint of capacity, from a standpoint of efficiency, from a standpoint of safety we won.  But I and no one else saw the 170 crashes coming.  We thought to ourselves it was one hiccup, they are going to get use to it.  There was a road closure on a major freeway project that we did a mile down the road and we utilized this roundabout as a detour, so I conveniently blamed that for the reason of the 170, and the next year it was 171.  Oops.  Nice job Mark.  So it forced me to say alright what can we do better, what's going on, why is this happening, and what can we be doing?  And here it is, forever and ever in the state of Michigan in the calendar year 2015 - 2014 I believe it was - I have the most dangerous intersection in the state of Michigan.  So that's forever on my resume. 

Slide #5:  http://teachamerica.com/RAB17/RAB1710A_McCulloch/index.htm

[jakeroot]

^^^

Wow. That's a pretty telling confession. I kind of feel bad for him. He was a little misled by the early stats, and the double lane roundabout was a new thing, so there was no way to know how it was going to operate. Unfortunately for him, it ended up performing rather poorly.

THAT SAID, there have to be at least a few double lane roundabouts in the US that perform reasonably well. Tradephoric, any idea where the safest double lane roundabout is in the US (at least from what you can tell from the data you've compiled)?

[cjw2001]

An interesting presentation from the 2011 roundabout conference on crash rates in areas familiar with roundabouts (Carmel, IN) vs areas unfamiliar.   Some multilane stats on slide 48.  http://teachamerica.com/RAB11/RAB1107Barbier/player.html

[7/8]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

[US 89]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

[jakeroot]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

I don't think we've seen the end of gigantic, two and three lane roundabouts. Until we get a real research project going that involves several high-profile agencies, the only thing stopping more from popping up is an engineer's previous experience (e.g. Mark McCulloch probably won't be building any more multi-lane roundabouts).

[US 89]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

I don't think we've seen the end of gigantic, two and three lane roundabouts. Until we get a real research project going that involves several high-profile agencies, the only thing stopping more from popping up is an engineer's previous experience (e.g. Mark McCulloch probably won't be building any more multi-lane roundabouts).

If the new crash rate is 11x what it was before, like the one in Michigan, then those engineers probably won't install any more.

As more and more engineers make these mistakes, less and less of these roundabouts will be built.

[jakeroot]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

I don't think we've seen the end of gigantic, two and three lane roundabouts. Until we get a real research project going that involves several high-profile agencies, the only thing stopping more from popping up is an engineer's previous experience (e.g. Mark McCulloch probably won't be building any more multi-lane roundabouts).

If the new crash rate is 11x what it was before, like the one in Michigan, then those engineers probably won't install any more.

As more and more engineers make these mistakes, less and less of these roundabouts will be built.

There's two types of engineers (I'd reckon): the kind that do things once, and determine whether or not they'll do it again based on the feedback from that "thing", and the kind that do things over and over again until it's clear that something either works or doesn't.

The traffic engineer for a city near me replaced (roughly) ten protected lefts with flashing yellow arrows in 2006, and watched them for a year before he decided what he thought of them. They were a resounding success (reducing traffic congestion and collisions/injuries), and all remaining city-maintained single-lane protected lefts were replaced with FYAs over the next ten years.

He's the second kind of engineer, and that's a good thing: if he based his study on just one intersection, there'd be too many variables to be certain of any outcome. Installing many at once eliminates many of the location-based variables, such as sight distance, intersection size, TOD flow, etc. The first kind of engineer would have installed one or two, and based their future left turn signal designs entirely on the results of those two signals. I think that's just dumb. The sample size isn't big enough.

[jeffandnicole]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

I don't think we've seen the end of gigantic, two and three lane roundabouts. Until we get a real research project going that involves several high-profile agencies, the only thing stopping more from popping up is an engineer's previous experience (e.g. Mark McCulloch probably won't be building any more multi-lane roundabouts).

If the new crash rate is 11x what it was before, like the one in Michigan, then those engineers probably won't install any more.

As more and more engineers make these mistakes, less and less of these roundabouts will be built.

There's two types of engineers (I'd reckon): the kind that do things once, and determine whether or not they'll do it again based on the feedback from that "thing", and the kind that do things over and over again until it's clear that something either works or doesn't.

The traffic engineer for a city near me replaced (roughly) ten protected lefts with flashing yellow arrows in 2006, and watched them for a year before he decided what he thought of them. They were a resounding success (reducing traffic congestion and collisions/injuries), and all remaining city-maintained single-lane protected lefts were replaced with FYAs over the next ten years.

He's the second kind of engineer, and that's a good thing: if he based his study on just one intersection, there'd be too many variables to be certain of any outcome. Installing many at once eliminates many of the location-based variables, such as sight distance, intersection size, TOD flow, etc. The first kind of engineer would have installed one or two, and based their future left turn signal designs entirely on the results of those two signals. I think that's just dumb. The sample size isn't big enough.

Not to mention when you install just one of something, there's a steep learning curve for motorists at that one intersection, which is unusual and abnormal from all the rest.  Many will encounter it on rare occasion, and it's possible they'll be confused when they reach it.  Do it at 11 intersections, and more people see them more often.

[tradephoric]

THAT SAID, there have to be at least a few double lane roundabouts in the US that perform reasonably well. Tradephoric, any idea where the safest double lane roundabout is in the US (at least from what you can tell from the data you've compiled)?

The double-lane roundabout at County Road M & Valley View Road in Madison appears to be performing reasonably well.  But just a mile up the road there is a double-lane roundabout on Mineral Point Road with a relatively high crash rate (approaching 4.0 MEV).  I believe both roundabouts were built in 2010 so it could be interesting to compare the design of those two roundabouts.



The average crash rate of signalized intersections is roughly 0.7 MEV.  Double-lane roundabouts (with 2x2 or 3x2 configurations) have average crash rates approaching 4.0 MEV.  Nearly 1 in 3 roundabouts on that list have had circulating lanes removed or they are planning to do so (roundabouts highlighted in yellow).  Even if you don't believe the numbers, the fact that agencies are removing circulating lanes from so many of these complex roundabouts should indicate that there is indeed a problem.  And in all the research I've seen, when a circulating lane is removed the crash rate drops precipitously.  Conversely, when only "tweaks"  are made to a roundabout in attempt to reduce the crash rate (such as signage or lane marking changes), the reduction in crashes seems to only be temporary (the Bluffton Parkway roundabout in South Carolina comes to mind).

An interesting presentation from the 2011 roundabout conference on crash rates in areas familiar with roundabouts (Carmel, IN) vs areas unfamiliar.   Some multilane stats on slide 48.  http://teachamerica.com/RAB11/RAB1107Barbier/player.html

Communities with a high concentration of roundabouts aren't immune to high crash rates.  While Carmel has an abundance of roundabouts, relatively few have complex geometries.  In fact, of the 100 roundabouts in Carmel not a single one has a 3x2 geometry. And the few 2x2 roundabouts in Carmel that exist have high crash rates.  The roundabout at 116th and Keystone is one of the worst performing intersections in the city with a crash rate over 5.0 MEV.  To put that into perspective, the roundabout at M-5 & Pontiac trail has the same crash rate as the 116th & Keystone roundabout, and that roundabout had the most crashes in Michigan in 2015.  The only reason 116th and Keystone Parkway isn't winning "most crashes in the State"  awards is because it carries 20,000 fewer vehicles than the M-5 roundabout.
 
Over the past few years more complex roundabouts have been built in Carmel but it's too early to see how well they are performing.  Unfortunately, the Carmel Police department no longer lists the highest crash intersections in their annual reports, so it's harder to pin down numbers on recently constructed roundabouts.  Too bad the city of Carmel isn't being as transparent as they once were.

[kalvado]

Communities with a high concentration of roundabouts aren't immune to high crash rates.  While Carmel has an abundance of roundabouts, relatively few have complex geometries.  In fact, of the 100 roundabouts in Carmel not a single one has a 3x2 geometry. And the few 2x2 roundabouts in Carmel that exist have high crash rates.  The roundabout at 116th and Keystone is one of the worst performing intersections in the city with a crash rate over 5.0 MEV.  To put that into perspective, the roundabout at M-5 & Pontiac trail has the same crash rate as the 116th & Keystone roundabout, and that roundabout had the most crashes in Michigan in 2015.  The only reason 116th and Keystone Parkway isn't winning "most crashes in the State"  awards is because it carries 20,000 fewer vehicles than the M-5 roundabout.

An interesting safety metrics for a given community is average insurance rate - which is published online. Interestingly enough, Carmel is lower than neighbors; not much - but noticeable.

[kalvado]

Speaking of multi-lane roundabouts, construction has started on two side-by-side roundabouts on Ottawa Street in Kitchener (one at Homer Watson Blvd, and one at Alpine Road). This is not far from the infamous Homer Watson/Block Line Roundabout. They're aiming to be finished by the fall. We'll have to wait and see if it ends up improving the Region's second worst intersection or not.

The one at Homer Watson looks like a really busy intersection. Based on what has been posted, that sounds like it's going to be a bad move.

I don't think we've seen the end of gigantic, two and three lane roundabouts. Until we get a real research project going that involves several high-profile agencies, the only thing stopping more from popping up is an engineer's previous experience (e.g. Mark McCulloch probably won't be building any more multi-lane roundabouts).

If the new crash rate is 11x what it was before, like the one in Michigan, then those engineers probably won't install any more.

As more and more engineers make these mistakes, less and less of these roundabouts will be built.

A smart person is the one who can learn from their mistakes
A wise person is the one who can learn from mistakes others made
Some people don't learn from neither their own mistakes nor mistakes others made...

[jakeroot]

THAT SAID, there have to be at least a few double lane roundabouts in the US that perform reasonably well. Tradephoric, any idea where the safest double lane roundabout is in the US (at least from what you can tell from the data you've compiled)?

The double-lane roundabout at County Road M & Valley View Road in Madison appears to be performing reasonably well.  But just a mile up the road there is a double-lane roundabout on Mineral Point Road with a relatively high crash rate (approaching 4.0 MEV).  I believe both roundabouts were built in 2010 so it could be interesting to compare the design of those two roundabouts.

The thing that sticks out to me are the slip lanes at the CR-M/Valley View roundabout.

Out of curiosity, do AADT numbers count those that turn right via slip lanes?

[tradephoric]

^^^
The AADT numbers would include the right turn counts via the slip lanes.