Crash prone 'modern roundabouts'

[kphoger]

A drunk driver speeding through a standard non-roundabout intersection and hitting another vehicle is also likely to result in a fatality.  But highlighting it in this thread implies that the fatality is somehow due to the intersection's being a roundabout.

[kalvado]

A drunk driver speeding through a standard non-roundabout intersection and hitting another vehicle is also likely to result in a fatality.  But highlighting it in this thread implies that the fatality is somehow due to the intersection's being a roundabout.

It really depends. If it is a one off event, then you have a point. if, as Trade said

Apart from the major increases in crashes ever since the roundabout was built there has now been a fatality.   

Then it's just another datapoint.

[kphoger]

Then it's just another datapoint.

I agree.  Just another datapoint.  Unnoteworthy, so far as the intersection design is concerned.

[kalvado]

Then it's just another datapoint.

I agree.  Just another datapoint.  Unnoteworthy, so far as the intersection design is concerned.

Fatal accident is generally a noteworthy datapoint for an intersection. Moreso if it occurs with high level of non-fatal crashes

[jakeroot]

I agree.  Just another datapoint.  Unnoteworthy, so far as the intersection design is concerned.

Fatal accident is generally a noteworthy datapoint for an intersection. Moreso if it occurs with high level of non-fatal crashes

They are now, to be sure. But fatality crashes did seem tolerated (to an extent) into the early 2000s. Crash rates had been dropping for decades and so too was the serious injury/fatality rate. But after fatality/serious injury rates began to rise around ten years ago, a lot more attention has been paid to them. Even those that seem to have nothing to do with engineering.

The problem with roundabouts is that they have a 100% crash rate among those drivers who fail to circumnavigate the roundabout. Outside of roundabouts, I cannot think of an intersection design (applicable to major intersections) where a geometric hazard is considered good design.

[kphoger]

Outside of roundabouts, I cannot think of an intersection design (applicable to major intersections) where a geometric hazard is considered good design.

Some skewed intersections have the side-road do a sinuous route in order to meet up at closer to 90°.  Not sure if that counts.

[kalvado]

Outside of roundabouts, I cannot think of an intersection design (applicable to major intersections) where a geometric hazard is considered good design.

Some skewed intersections have the side-road do a sinuous route in order to meet up at closer to 90°.  Not sure if that counts.

And any T intersection in general, where failure to stop on asymmetric leg can be anything from a guardrail damage to a flight off the cliff

[jakeroot]

Cross-posting from the Washington thread:

SR-9 to be reduced to one lane at SR-204 due to a high number of crashes in the roundabout (which are dropping, apparently...but not enough?)

It's amazing that we went from the initial plan of SR-9 grade separation down to a single southbound through lane. It's a shame they couldn't spring for a flyover, because now there's fewer southbound through lanes then 30 years ago.

https://wsdotblog.blogspot.com/2026/08/sr-9-sr-204-roundabout.html

[CoreySamson]

^^^ Why would you put the roundabout there with that side road?

Anyways, I came to this thread today because I discovered this near Cleveland, TX in the much beleaguered Colony Ridge subdivision:
https://maps.app.goo.gl/inaERaymSWMVmrJ56
https://maps.app.goo.gl/4aLR9CNvaAnRdzfG6

Is this the right way to mark a multilane roundabout? Because it doesn't look right to me.

[Scott5114]

Is this the right way to mark a multilane roundabout?

Hell no. Holy conflict points, Batman!

[jakeroot]

^^^ Why would you put the roundabout there with that side road?

The whole area is full of them. It's a totally botched setup on all counts, in my opinion: too many roundabouts too close together, too much traffic, too many chicanes, too many slip lanes...it's just a mess.

https://www.google.com/maps/@48.0038671,-122.1083555,434a,35y,119.86h,14.12t/data=!3m1!1e3?entry=ttu&g_ep=EgoyMDI2MDgxOS4wIKXMDSoASAFQAw%3D%3D

Anyways, I came to this thread today because I discovered this near Cleveland, TX in the much beleaguered Colony Ridge subdivision:
https://maps.app.goo.gl/inaERaymSWMVmrJ56
https://maps.app.goo.gl/4aLR9CNvaAnRdzfG6

Is this the right way to mark a multilane roundabout? Because it doesn't look right to me.

That whole area, including the design of the roundabouts, looks straight out of the 80s. I'm stunned this is a development from the last ten years. The roundabouts themselves are so...80s looking. Very European too, with straight-on entry/exit legs.

Hell no. Holy conflict points, Batman!

While I can see your concern, I think this is MUTCD compliant. Marking the section beyond that exit as two lanes probably isn't kosher, but I don't think it's disallowed either.

[Scott5114]

While I can see your concern, I think this is MUTCD compliant. Marking the section beyond that exit as two lanes probably isn't kosher, but I don't think it's disallowed either.

So if someone in the left lane wants to get out of the roundabout and the person in the right lane doesn't, and they each follow the lane markings there...What happens?

That whole area, including the design of the roundabouts, looks straight out of the 80s. I'm stunned this is a development from the last ten years. The roundabouts themselves are so...80s looking.

I think it's because the closely-spaced dashed lines and lack of edge lines give it dead-mall-parking-lot vibes.

[jakeroot]

While I can see your concern, I think this is MUTCD compliant. Marking the section beyond that exit as two lanes probably isn't kosher, but I don't think it's disallowed either.

So if someone in the left lane wants to get out of the roundabout and the person in the right lane doesn't, and they each follow the lane markings there...What happens?

That's pretty normal, though, no?

https://maps.app.goo.gl/CdXU3LBnnpEMPzV28

Traffic in the "outer" lane of a roundabout rarely is permitted to continue around a roundabout unless it's expressly marked as permitted.

That whole area, including the design of the roundabouts, looks straight out of the 80s. I'm stunned this is a development from the last ten years. The roundabouts themselves are so...80s looking.

I think it's because the closely-spaced dashed lines and lack of edge lines give it dead-mall-parking-lot vibes.

For sure. And the lack of truck apron, lack of crosswalks or sidewalks...just devoid of all normal features!

[CoreySamson]

That whole area, including the design of the roundabouts, looks straight out of the 80s. I'm stunned this is a development from the last ten years. The roundabouts themselves are so...80s looking.

I think it's because the closely-spaced dashed lines and lack of edge lines give it dead-mall-parking-lot vibes.

For sure. And the lack of truck apron, lack of crosswalks or sidewalks...just devoid of all normal features!

The road infrastructure with the geometry and lane lines specifically reminds me of something you'd see in a newer middle-class development in Mexico:

Reynosa: https://maps.app.goo.gl/hBHyT7MohkrBWpSi9
Cancun: https://maps.app.goo.gl/TG5TSQ1MNgUFZ8FY8
Outside Heroica Veracruz: https://maps.app.goo.gl/uyjMqpXHHvAxGrL87

[kalvado]

Traffic in the "outer" lane of a roundabout rarely is permitted to continue around a roundabout unless it's expressly marked as permitted.

I would say that you should have not more than 1 option lane.
As constructed, they can designate either lane as option, with some excessive pavement to support that designation. Could be that they wanted both variants to be available?

[kphoger]

Unless the lanes are otherwise marked, I approach a multi-lane roundabout with this assumption:

right lane = right turn or straight ahead
left lane = left turn or straight ahead

[kalvado]

Unless the lanes are otherwise marked, I approach a multi-lane roundabout with this assumption:

right lane = right turn or straight ahead
left lane = left turn or straight ahead

This implies low fraction of left turning traffic..

[tradephoric]

Lake County roundabout in the works after nearly 100 crashes at intersection
https://www.clickorlando.com/news/local/2026/09/07/lake-county-building-roundabout-after-nearly-100-crashes-at-eudora-road-intersection/

Here's an article about a planned roundabout at a Lake County, Florida intersection after experiencing "nearly 100 crashes at intersection".  Going deeper into the crash stats in the body of the article, there were 85 crashes over the past decade (2016-2025). 

85 crashes the past decade is high?  In 2019, there were 229 reported crashes at the 18 1/2 Mile @ Van Dyke roundabout in Sterling Heights, Michigan. Crashes have remained in the triple digits each year since even during the Covid era.

2019: 229 crashes
2020: 131 crashes
2021: 168 crashes
2022: 118 crashes
2023: 135 crashes
2024: 162 crashes

943 crashes reported in just 6 years of data. 
85 crashes over the course of a decade is child's play.

[Rothman]

*makes a pie with the cherries picked*

[tradephoric]

*makes a pie with the cherries picked*

I don't exactly know what that means, but I'm hungry now.

EDIT:  Something about cherry picking data?  The article references 85 crashes over the course of a decade as "high" but if that's high what's an intersection with 943 crashes in 6 years considered?  Astronomically high roundabout crashes would be a fitting description.

[tradephoric]

It begs the question though, has a traffic signal ever had 943 crashes over any 6 year period in the history of the United States?  The Van Dyke @ 18 1/2 Mile roundabout is quite possibly the most crash prone intersection in the entire country.

[Scott5114]

It begs the question though, has a traffic signal ever had 943 crashes over any 6 year period in the history of the United States?

In classical rhetoric and logic, begging the question or assuming the conclusion (Latin: petitio principii) is an informal fallacy that occurs when an argument's premises assume the truth of the conclusion. Historically, begging the question refers to a fault in a dialectical argument in which the speaker assumes some premise that has not been demonstrated to be true. In modern philosophical usage, it has come to refer to an argument in which the premises assume the conclusion without supporting it. This makes it an example of circular reasoning.

[tradephoric]

I'm open to other intersections in America being more crash prone than the Van Dyke @ 18 1/2 Mile roundabout, but what intersection would it be?  I asked Chat GPT (because why not)... it listed 79th St & Stony Island Ave in Chicago as having the most total crashes with 687 crashes from 2020–2024. That Chicago intersection comes out to 137.4 crashes per year, still less than the 157.1 crashes per year at the Van Dyke roundabout.

 

[kphoger]

It begs the question though, has a traffic signal ever had 943 crashes over any 6 year period in the history of the United States?

In classical rhetoric and logic, begging the question or assuming the conclusion (Latin: petitio principii) is an informal fallacy that occurs when an argument's premises assume the truth of the conclusion. Historically, begging the question refers to a fault in a dialectical argument in which the speaker assumes some premise that has not been demonstrated to be true. In modern philosophical usage, it has come to refer to an argument in which the premises assume the conclusion without supporting it. This makes it an example of circular reasoning.

I'm glad someone finally pointed out an erroneous use of the phrase "begs the question".  Thank you.

Also, wouldn't it be ironic (don't worry, nobody knows what irony actually is and what it isn't) if we caught |tradephoric| using circular reasoning?

[Scott5114]

I asked Chat GPT (because why not)

Because it's usually full of shit?