Crash prone 'modern roundabouts'

[kalvado]

I've not read the entire 99 pages so please forgive me if I am retreading old ground. I'll describe for driving on the right hand side of the road.

British style multi-lane roundabouts cause us no end of safety problems either due to high circulating speeds and low gap availability to merge into. This means entering traffic that has spent a while at the yield line tends to take a gamble and cut in front of others. The traditional two bridges over a freeway roundabouts with a straight section linked by curves are notoriously prone to this and as a result often end up under part time or complete signal control. This gets rid of the entry problems but at the cost of increasing delays on all arms.

The main driver of safety at any large roundabout is to ensure drivers are in the correct lane before entering, and that the lane markings direct drivers to their desired exit with no lane change necessary. This is not always achievable because you may need to move traffic towards the central island in a "spiral" layout but these moves to the left have fewer conflict potentials than moves across to the right.

DDIs and other advanced signal configurations are in my view much better than the traditional two bridge roundabouts. Minimise the conflict points and ensure speeds are at an appropriate level and that's how you reduce collisions.

The Michigan examples seem to have some design flaws but they could be partially fixed relatively cheaply by introducing visibility screens to prevent "reading through" and forcing traffic to yield as a consequence.

I think you're going into too fine details. Root cause is a serious lack of understanding of when roundabouts can work and where they cannot.
They are definitely oversold in US, with lots of wishful but uneducated publications. As a result, a lot of roundabouts are just not working anywhere close to expectations.
Unlike SPUI, where base concept of "more pavement for more throughput" works reasonably well and pretty intuitive, its not really the case with roundabouts.

[tradephoric]

The Michigan examples seem to have some design flaws but they could be partially fixed relatively cheaply by introducing visibility screens to prevent "reading through" and forcing traffic to yield as a consequence.

They added visibility screens along the splitter islands of the M-5 & Pontiac Trail roundabout several years ago but it didn't help reduce crashes.  In fact the 6 months before the screens were added had the exact same number of crashes as the 6 months after.  Even with it being nearly a decade old, there seems to be more crashes at the roundabout than ever before (with 161 crashes in 2019 which is 23 more crashes than the 138 crashes it had in 2018).   

[Sani]

Here are some pictures of the new lane controls at the Los Alamitos (Long Beach, CA) Traffic Circle.  We're going to go counterclockwise starting at the 3 o'clock (east) side of the Traffic Circle.

I'm not sure how I missed this post.

I was just thinking the other day that the Los Alamitos Circle had not been modified from its original configuration since it was updated in 1994. That update in 1994 made it match British standards, with lane-control arrows before the roundabout but without lane markings. But my impression was that it had not been modified since because crashes hadn't become a hysterical issue like at other roundabouts. Part of me was contributing this to the vagueness of the circulating roadway (no lane lines might mean that drivers are a bit less careless, although I don't know this to be fact). I guess it's a bit disappointing to see that it has been modified, since it was one of the few unmodified 90s-era modern roundabouts; could be that those along I-70 in the Rockies are the last of this dying breed.

Part of me genuinely thinks crashes will go up, since all they've done is paint it to look like the numerous other multi-lane modern roundabouts that almost all universally perform poorly in terms of overall crashes. But I guess only time will tell.

As an insurance adjuster who had to determine liability on multiple crashes in this traffic circle and, more often than not, went 50/50 on liability for lack of any way to determine who encroached on whose lane, I'm glad there are finally marked lanes inside the circle.

[jakeroot]

As an insurance adjuster who had to determine liability on multiple crashes in this traffic circle and, more often than not, went 50/50 on liability for lack of any way to determine who encroached on whose lane, I'm glad there are finally marked lanes inside the circle.

I don't see why it makes a difference to you. This entire thread is evidence that crashes are an issue at multi-lane roundabouts, so you're still going to be dealing with crashes.

It might make your job easier since liability is more obvious, but I think that misses the bigger issue at hand here which is that no amount of markings or signs seems to improve crash rates.

[Ketchup99]

Roundabouts are good in certain situations, but not in others. A roundabout very effectively slows down traffic, which can be a blessing and a curse. On arterials, no. Roundabouts don't belong on arterials, in almost any situation. Timed traffic lights work lots better, and are also better at keeping traffic at a good speed - if the limit is 50, the only way to cruise down the road without stopping is at 50. That said, arterials should have as few intersections as possible anyway.

Roundabouts are good on residential collectors. For instance, this road in State College goes on for about three miles. Limit is 25 because it's a rather residential area, but when using it the way I do (I have to use the entire thing routinely), it's easy to watch the needle creep past 25, 30, 35. It just doesn't have the feel of the residential road that it largely is. Adding roundabouts at lots of intersections would definitely help keep speeds down.

[tradephoric]

https://www.youtube.com/watch?v=WNrz9_Fe-Us

Driving through a roundabout is like playing the arcade game "Frogger".  The problem is some people suck at Frogger and can't judge gaps in traffic.  When you have 50k people a day driving through a roundabout, that's like 50k individual games of Frogger taking place and some people won't make it past the first level.  Here is an arterial of State Street & Ellsworth roundabout in Ann Arbor which was the 5th most dangerous intersection in Michigan with 154 crashes last year.  Look at all those frogs (cars) having to judge gaps in traffic.

[BrynM65]

The entry angles also force you to have to look sharply over your shoulder to the left more than they should looking at that aerial again. This probably accounts for a lot of failure to see type collisions because that kind of posture is just uncomfortable so people don't do it.

The Ann Arbor example above has the problems I described that plague British roundabouts - including the two lane exits which, given the road reduces to single lane on three exit arms is not required - this is a hangover from the signals where you have the ahead lanes merge after the intersection because of the way platoons are sent out from a green.

I wonder if striping those exits down to a single lane and introducing a proper spiral lane system might help create better gaps.

[jakeroot]

The Ann Arbor example above has the problems I described that plague British roundabouts - including the two lane exits which, given the road reduces to single lane on three exit arms is not required - this is a hangover from the signals where you have the ahead lanes merge after the intersection because of the way platoons are sent out from a green.

To be clear, that practice exists specifically because of the British practice at the time. Many of our first roundabouts were designed according to British standards, or by actual British roundabout designers (such as Frank Blackmore from the TRB), including such things as widened entry and exit points relative to the approach roads, and minimal or no lane markings. We largely went more with the Australian roundabout guidelines when all was said and done, the guidelines became part of the MUTCD, but those traits are also seen in Australian designs.

The entry angles also force you to have to look sharply over your shoulder to the left more than they should looking at that aerial again. This probably accounts for a lot of failure to see type collisions because that kind of posture is just uncomfortable so people don't do it.

This is an issue that I don't believe is considered often enough. Roundabout design guidelines suggest that, to avoid path overlap, the crossover points should minimize the amount of "turning" input required by drivers; basically, the entry and exits should be straight on, with the only curving being once you're inside the roundabout, or the approach and exit (to "line up" with the crossover). This seems logical, but it does seem to have the negative effect of making the entries quite sharp, and multi-lane entries can be difficult if you're the outside car.

[Sani]

As an insurance adjuster who had to determine liability on multiple crashes in this traffic circle and, more often than not, went 50/50 on liability for lack of any way to determine who encroached on whose lane, I'm glad there are finally marked lanes inside the circle.

I don't see why it makes a difference to you. This entire thread is evidence that crashes are an issue at multi-lane roundabouts, so you're still going to be dealing with crashes.

It might make your job easier since liability is more obvious, but I think that misses the bigger issue at hand here which is that no amount of markings or signs seems to improve crash rates.

Right, my only point was that it makes liability easier to determine. I was being a bit glib.

I had at least one claim where two cars, side by side, entered the circle at the same time and tried to take the same multi-lane exit and still, somehow, managed to collide. At least lane markings will make it a bit easier for those folks to tell which lane they should be in, even if they don't know where the edges of their vehicles are.

[kphoger]

The entry angles also force you to have to look sharply over your shoulder to the left more than they should looking at that aerial again. This probably accounts for a lot of failure to see type collisions because that kind of posture is just uncomfortable so people don't do it.

This is an issue that I don't believe is considered often enough. Roundabout design guidelines suggest that, to avoid path overlap, the crossover points should minimize the amount of "turning" input required by drivers; basically, the entry and exits should be straight on, with the only curving being once you're inside the roundabout, or the approach and exit (to "line up" with the crossover). This seems logical, but it does seem to have the negative effect of making the entries quite sharp, and multi-lane entries can be difficult if you're the outside car.

The deflected entry is also intended to slow drivers down.  They're more likely to enter a roundabout at too-high speeds if the entry is more of a straight line.  The exit is more of a straight shot because that's not a factor upon exiting.

[jakeroot]

The entry angles also force you to have to look sharply over your shoulder to the left more than they should looking at that aerial again. This probably accounts for a lot of failure to see type collisions because that kind of posture is just uncomfortable so people don't do it.

This is an issue that I don't believe is considered often enough. Roundabout design guidelines suggest that, to avoid path overlap, the crossover points should minimize the amount of "turning" input required by drivers; basically, the entry and exits should be straight on, with the only curving being once you're inside the roundabout, or the approach and exit (to "line up" with the crossover). This seems logical, but it does seem to have the negative effect of making the entries quite sharp, and multi-lane entries can be difficult if you're the outside car.

The deflected entry is also intended to slow drivers down.  They're more likely to enter a roundabout at too-high speeds if the entry is more of a straight line.  The exit is more of a straight shot because that's not a factor upon exiting.

I know what the purpose of the curvature is. I'm say it's ill-advised, and/or had unintended consequences. The strong curvature and desired entry angle can, unintentionally, makes it difficult to see to the left (mostly for the outside lane, but sometimes inside too).

[kphoger]

it's ill-advised, and/or had unintended consequences.

I suspect a combination of the two.  It's all but impossible for me to believe nobody considered that it's harder to look to the right when the angle is thus deflected.  But it's easy for me to believe it was determined that such wasn't a big deal.

[BrynM65]

it's ill-advised, and/or had unintended consequences.

I suspect a combination of the two.  It's all but impossible for me to believe nobody considered that it's harder to look to the right when the angle is thus deflected.  But it's easy for me to believe it was determined that such wasn't a big deal.

One of my roles is road safety auditing and it's frightening how we often as engineers ignore the basics, because to us it's 'obvious'. Average drivers don't think like engineers, which is why there's often inexplicable resistance to sensible engineering concepts like DDIs because they look like a complex monster that makes you drive on the wrong side despite being actually ten times more efficient than a standard diamond.

A lot of 'fear' of roundabouts is the same thing I think - only an engineer would make you turn right to turn left (see also the humble cloverleaf interchange).

[kphoger]

DDIs ... actually ten times more efficient than a standard diamond.

Uh oh.  Now you've asked for it.

[BrynM65]

DDIs ... actually ten times more efficient than a standard diamond.

Uh oh.  Now you've asked for it.

They're not infallible; intersection choice is always location dependent - another thing us engineers screw up. No two intersections are the same after all.

[kphoger]

It's just that, in tradephoric's darling thread, you brought up another topic he's passionate about:  the comparative performance of DDIs.  Hopefully this won't now devolve into a discussion about the ParClo B4 and/or signal progression in Detroit.

[kalvado]

it's ill-advised, and/or had unintended consequences.

I suspect a combination of the two.  It's all but impossible for me to believe nobody considered that it's harder to look to the right when the angle is thus deflected.  But it's easy for me to believe it was determined that such wasn't a big deal.

One of my roles is road safety auditing and it's frightening how we often as engineers ignore the basics, because to us it's 'obvious'. Average drivers don't think like engineers, which is why there's often inexplicable resistance to sensible engineering concepts like DDIs because they look like a complex monster that makes you drive on the wrong side despite being actually ten times more efficient than a standard diamond.

A lot of 'fear' of roundabouts is the same thing I think - only an engineer would make you turn right to turn left (see also the humble cloverleaf interchange).

I would say that as long as this remains the selling picture for roundabouts, I would still assume that engineers do not simply ignore, but plainly don't understand basics:

[kphoger]

basics

Such as drawing to scale.

A whole lot of those conflict points in the second picture could be collapsed into just a few dots.

And that's not even considering the fact that it considers points at which vehicle paths diverge to be points of conflict.  I estimate that the number of red dots could realistically be reduced by half, and the yellow squares by one third.

[jeffandnicole]

Yeah, I don't think so.

If I'm guessing, that graphic is comparing a 1 lane-per-direction roundabout with a 3 lane-per direction intersection.

Even within a roundabout, I'd say there's really two conflict points between each leg - where a vehicle enters the roundabout and where the vehicle exits the roundabout.  The point between the two areas would actually be the safest point.

[kalvado]

Yeah, I don't think so.

If I'm guessing, that graphic is comparing a 1 lane-per-direction roundabout with a 3 lane-per direction intersection.

Even within a roundabout, I'd say there's really two conflict points between each leg - where a vehicle enters the roundabout and where the vehicle exits the roundabout.  The point between the two areas would actually be the safest point.

Thing is, even then that is irrelevant. First question to ask is:  why number of conflict points is a figure of merit to begin with?

[kphoger]

The FHWA is certainly under the impression that reducing conflict points is a good thing.  What I can't find, however, is why they think so.

The success of innovative intersection designs begins with reducing the number and severity of conflict points — locations where user paths coincide.

The common element in each treatment is the reduction in conflict points at the intersection, which provides safety and operational benefits by reducing the number of phases and conflicting volume at a single location.

It is desirable to minimize the number of conflict points created with existing and future driveways since more conflict points increase the risk of a crash occurring.

[kalvado]

The FHWA is certainly under the impression that reducing conflict points is a good thing.  What I can't find, however, is why they think so.

The success of innovative intersection designs begins with reducing the number and severity of conflict points — locations where user paths coincide.

The common element in each treatment is the reduction in conflict points at the intersection, which provides safety and operational benefits by reducing the number of phases and conflicting volume at a single location.

It is desirable to minimize the number of conflict points created with existing and future driveways since more conflict points increase the risk of a crash occurring.

Because.. khmmm it is called education, you know... Imagine a grid with multiple parallel streets. Would compressing multiple parallel into one make things safer? Fewer conflict points, thats for sure

[sparker]

Good old AASHTO Weekly; they're reprinting a bit of roundabout flack:

https://aashtojournal.org/2020/07/02/trb-analysis-illustrates-benefits-of-roundabout-intersections/

One of the primary points raised is that there is an inherent reduction of overall speed when roundabouts are deployed.  All well and good in dense urban areas where general speed reduction might be beneficial -- but that concept isn't terribly portable to rural highways.  But it seems that some planners -- particularly at the state level -- seem to have posited a "zero-sum" approach to such things; that to provide certain benefits to the overall populace the methods employed to do so must be at the detriment of the driving portion of the public.  A "one size fits all" curriculum doesn't really address underlying problems; as I've averred previously, this is what's known as the "garbage can" policy methodology -- apply one's favored "solution" to as many problems as possible regardless of appropriateness.  Too much of that sort of BS can backfire against transportation agencies in terms of both operation and appropriations, particularly if repeated complaints are lodged by legislative constituents. 

[tradephoric]

I'm a fan of the "partial" Parclo A4 roundabout.  It has only one vehicle-to-vehicle conflict point compared to four at a more standard interchange roundabout (assume each example below is a "single lane" roundabout when calculating the number of conflict points).  I like the fact that all traffic entering the freeway doesn't have to travel through a roundabout and it's a simple design that limits the number of vehicle-to-vehicle conflicts.

https://www.youtube.com/watch?v=W8k1acLeK7E

^Through traffic along the arterial should never have to yield meaning that platooning can largely stay in tact in regards to good signal progression (arterial traffic will have to slow down to 25 mph to navigate the "partial" roundabout but only the off ramp traffic actually yields). 

[jakeroot]

^^^
So what benefits are there exactly, as compared to a standard left-hand "add lane" merge?

I suppose the curve makes it so that traffic slows down, allowing the off-ramp more chances to go, although that slowing could cause cars to "pile up" approaching the curve, making it just as hard to find gaps (especially if the approach road is a relatively straight, wide, high-speed arterial with heavy traffic).

There is a left-hand merge A4 in Surrey, BC seems to work pretty well.