Random Dutch junction ramblings

4 posts / 0 new
Last post
PaulJames
PaulJames's picture
Random Dutch junction ramblings

Yesterday I had a quick email conversation with Shaun McDonald (@smsm1) about a deck produced by Rik Andrew for discussion at the LCC junction review group. Below is my quick feedback for Shaun on the document and Dutch junction design in general, I thought it might be interesting for other people and/or discussion.

Hi Paul,

Please find attached a presentation that Rik Andrew has created of his idea of how to solve the left hook and traffic light phasing problem.

https://drive.google.com/file/d/0B5sPUQkfJ6WNTnBvQVZaYnk4ekh4UVRBNHE5Vjd...

What are your thoughts on Rik’s presentation?

Shaun

Hi Shaun,

Thanks for forwarding this and bringing it to my attention.

Overall Rik's deck is very good, there are some excellent ideas clearly delivered.

From a Dutch/Sustainable Safety perspective there are a few things I'd add:

The SS approach to road safety is a holistic approach. It says that you can not look at a single element but must look at the whole network as a single whole. Thus there is no one size fits all solution. Rik's deck alludes to this idea.

So before looking at junction design, the Dutch would first look at the use of the road, what is it's purpose. In general, through routes require a lot of space while access roads require little (and distributors are somewhere in between). So the designation of a road as "through", "distributor" or "access" is a balance between the purpose of the road (what the city wants the road to be used for) and the amount of space available for the road.

If for example, we have a road that is being used as a through route and the prime concern is to keep it as a through route but there is not enough space for everyone who needs to use that road, then something has to give. This ultimately leads to traffic removal of some kind, usually by it being bypassed in some way. This may just be a parallel route for bicycles/pedestrians, or removing motor vehicles in one direction (making it one-way for motors) and introducing a parallel route for the other direction, etc. etc. There are many solutions depending on the circumstances.

One of the key things this leads to when it comes to the junctions between roads is that the safest and most efficient junction is no junction. The Dutch will remove any un-needed junctions from through routes turning the road network into a cell-like structure where each "block" between through roads has only a few entry/exit points off of the through roads. This reduces the conflict points along the though roads and means you need less heavy engineering to figure out junctions (since you have less junctions). Turning cross roads into T-junctions by closing one arm is a very effective way of making space for a fully separated (time/space) junction. Obviously how much of this you can do depends on your network.

At the individual junction level, Rik makes some good points.

The Dutch in general do 1 of 3 things with junctions depending on the type of road:

On an access road, they eschew explicit priority all together and go with the default give way to the traffic from the right. This means no markings and vehicles have to travel slowly so as to keep an eye out for turning traffic.

On distributor roads where there is no always enough room for full segregation, bikes (and peds) and motors will get a simultaneous green for going straight on, but also allow left/right turns to be made at any point from a single motor traffic lane (no dedicated turning lanes). Motors crossing the cycleway/footway crossing are expected to give way and this is often enforced by markings (sharks teeth/elephants feet, zebra). The cycleway/footway is either bent away from the junction to physically separate it from the roadway and create a waiting space, or is it bent towards the roadway (if space is limited) to place the cyclist/pedestrian in to direct line of sight of the turning driver (this is not so good for HGV left hook but then the Dutch have HGV bans). http://pedestrianiselondon.tumblr.com/post/52244534872/small-traffic-con...

On through roads where there is more space, full segregation and separate turn lanes/stages are used. Turning motors are held at red if bikes are detected in the cycleway (via a induction loop).http://pedestrianiselondon.tumblr.com/post/27284356735/traffic-controlle...

Typically, bikes only need a very short dedicated stage (depends on circumstance of course) as 1) junctions are far apart so cyclists won't come across many so waiting isn't so big a problem and induction loops can be used to give priority when needed 2) bikes are much quicker at crossing a junction from a stop than motors are. So you can do a straight on stage for all, then after 6-10 seconds if no more bikes are detected, show red to the cycleway and start the motors left turn stage.

Rik's description of having all turns going at once is going to be very problematic. It is not legal in the UK to have such a set up. When shown a green arrow (filter arrow), unlike the green ball (standard green traffic signal) vehicles have absolute right of way when making that turn. Having a potential conflict with traffic turning from the opposite direction would not be allowed, even where there are two exit lanes. This is why the approach of having dedicated turn lanes and 4 stages as I have described I think is preferable, if not as efficient. Also Rik's design leaves no time for pedestrians to cross unless you have an all directions red for motors/bikes (or allow motors to turn through a green man which is not legal).

The all ways "scramble" green is another option that he does not discuss which is preferred by Hembrow although I've never witnessed one myself (they seem to be relatively new and confined to places I haven't visited). They would work well iff there are induction loops to add a bike green stage quickly after spotting a bicycle rather than waiting for the full light cycle to complete first.

To me the key is the reduction in the number of complex junctions rather than the solving of the junctions themselves. Complex junctions are always going to be troublesome and cause delays to everyone involved, so the less of them the better. Closing junctions and turning cross roads into t-junctions has the added benefit of removing rat runs, and junctions into and out of access only areas can be treated as standard priority junctions (motors have to mount the pavement to enter) rather than complex traffic controlled junctions.

Sorry this is so rambling. I hope it is useful for this evening. If it's useful I'd be happy to meet up with anyone from LCC to discuss these things over a pint.

onlyanotherandy

Interesting.

"When shown a green arrow (filter arrow), unlike the green ball (standard green traffic signal) vehicles have absolute right of way when making that turn. Having a potential conflict with traffic turning from the opposite direction would not be allowed, even where there are two exit lanes."

What if the exit lanes were also kerb-seperated by a large margin, with dashed lane markers through the junction to guide vehicles into the correct lane? Something like this. With some cycleways though.

 

Edit: Then again, I guess it just takes up more space and doesn't offer any benefits compared to phases mentioned on Pedestrianise London with 1 exit lane per arm.

Edit 2: Actually, in both turning phases those extra left turns could also go too: link

pete owens

What is astonishing is that anyone should even conteplate designing the phasing of the lights that gives green lights simultaneously for two streams of traffic that cross each others paths through a junction. What is even more astonishing that such a design could pass a safety audit. 

This is a fundamental requirement for safety (just as the requirement for filter arrows to indicate a clear run through a junction) - not some optional feaure that would be nice to have  when balanced against other competing design parameters.

 

onlyanotherandy

Quite common in the UK, just never with left/right filter lights.

CS2 "segretated" extension anyone? I don't really blame this driver.

Log in or register to post comments