Friday, 30 March 2018

Traffic Signal Pie - Time For T

I've been grappling with an issue of what we can do where a cycle route in a quiet network of back streets meets a busy urban mixed-use A-road.

There is a real-world site I have in mind, but I won't name it as it is too close to work. However, this doesn't matter as the geometry is more important that location and it has been good to flex my brain on the issue.

The main road in question is a typical long-distance single carriageway A-road which was never planned as such and so ends up connecting a series of shopping centres with housing between each; plus the occasional school and community facility. In fact, these mixed-use streets are everywhere and often end up being the de-facto long-distance road network.

The existing layout is shown above. The main road runs east-west and carries around 20,400 vehicles per day (over 24-hours), with 17,500 of them being cars. There are (unsurprisingly) 200 people per day cycling, 500 bus/ coach movements and the rest various sizes of HGV.

The total highway width gets up to around 20.5 metres with about 11.5m being carriageway (with advisory cycle lanes). To the south of the junction, we have a primary school. The side street 7m wide, flaring out to a huge 21m as it means the main road - probably because of the former industrial estate (now housing with a couple of small workshops). The side street feeds a residential area and because of modal filters, it's the only motor vehicle entrance into the estate. There is also a signalised pedestrian crossing to the west of the junction on the main road.

Aside from the main road being pretty awful to cycling along, the principal issue is one of helping people cycling to turn right into and out of the side street. The main road needs cycle tracks to do anything about the woeful cycling numbers and the twice a day congestion, but the links are pretty simple, the junctions are the challenge.

My starting point was a junction I had studied in Amsterdam;

The junction of Middenweg with Zaaiersweg is not a million miles away from my study location. Aside from the central bus/ tram lanes, there is only one general traffic lane in each direction and the road is flanked by one-way cycle tracks and footways. The cycling here feels nice and safe. The issue here is one of turning left (i.e. UK right) from the side street (Zaaiersweg) which is the main access to a filtered estate. 

Turning left required people cycling to cross the entire carriageway to get to the cycle track going in the opposite direction (towards the city in this case). At least when I was there, the main road was quiet enough to find a gap to cross in one go, but I could imagine times when this would be harder. After mooching around in Streetview, I found a junction with Postjesweg elsewhere in the city which provided space to cross the road in two parts (with cycle space) and a pedestrian crossing on one side and this gave inspiration for my crack at a UK version;

Although I hadn't been to this junction, I have crossed large urban Dutch Roads with a refuge and priority remaining with drivers such as here;

As you can see from below, the highway space has been reworked as you might expect to see in a place such as Amsterdam.  

For people cycling, the side street has a similar entrance treatment to Zaaiersweg and so the right turn out (Dutch left) has people cycling coming from the shared carriageway of the side road and across the cycle crossing point with motor traffic retaining priority. Those turning right in have a similar provision, but they end up in the shared carriageway (blue is right out, purple is right in);

People on foot retain a crossing, but it's now a zebra rather than being signalised.  I put the sketch out on Twitter and as usual it provided for some good debate. David Hembrow did sound a note of caution in terms of the UK probably having far more motor traffic than the Netherlands and so it might not be suitable, also citing Dutch experience with the same. He also flagged the issue of needing to divert the traffic [from the main road] before trying to 'civilise the street'.

This is where the UK constantly comes up against a wall. In my example, there is a high-speed (parallel) dual-carriageway about 750m to the north and to reduce through-traffic on the main road, we would have to divert it to the trunk road. This would mean having people driving longer distances switching to the trunk road and at the same time making it less favourable to use my main road example as a long-distance route. The problem is (at least for now) is the trunk road is often heavily congested at peak times and a major change like this would need political cooperation across a number of boroughs and Transport for London.

So, I wondered if a pragmatic approach could be signalising the junction with the justification that it helps feed a quiet cycle route to the north. If a borough were going it alone (like one of the mini-holland boroughs) then this could be a treatment which protects those who we would like to get cycling (such as children going to the primary school) and as such, any loss of capacity for long-distance drivers is given back to local people who will want to cycle.

In playing with the space, it was clear that space was tight into the side road and so providing a set of three parallel crossings with "floating" crossings couldn't quite fit. The sketch below shows my thinking (ignore the southern cycle-only arm).

This approach would make for a simple method of control whereby the east-west motor traffic runs together, then the side road (north) runs an then finally the three parallel pedestrian and cycle crossings run together. I had to think a bit differently;

The image above shows the side street as having those cycling and driving mixing as opposed to the separation in the example above. The traffic signal method of operation took some thinking about, but it runs like this;

1. East-west motor traffic runs together (black),

2. Traffic released from side street. Motor traffic joins carriageway of main road (black) and cycles enter the cycle tracks (purple),

3. Pedestrians cross both arms of the main road (red) and cycle traffic turning right from the main road into the side street runs in parallel,

4. East-bound and left turning cycle traffic runs with pedestrians still crossing both arms of the main road.

3 and 4 could be swapped so that eastbound/ left turning cycle traffic runs before cycle traffic turning right into the side street, but either way, 3 and 4 are essentially a single pedestrian stage within which two cycle sub-stages are squeezed because people cycling need less time to cross. Don't ask me what the timings might be with this as I am not a traffic engineer!

The drawbacks of this design are that people cycling east are going to have their progress halted (and there will be red light jumping). In the UK, we couldn't have people cycling ahead being put into conflict with left turning drivers within traffic signals, although for some reason this is perfectly acceptable at uncontrolled side roads with advisory cycle lanes!

Alternatively, left turning drivers could be held on red signals (so-called 'hold the left turn'), but this requires stacking space which isn't available in my example - see the layout here in Mile End, East London;

The cycle track on the left can run with the ahead motor traffic movement when the left turning motor traffic is held. It is all very complex and needs lots of motor traffic stacking space!

So where does this leave us? Signalisation can be costly and not very flexible for walking and cycling (after all, signalisation is generally about stopping conflict with motor traffic). With the first sketch, we could have parallel zebra crossings, but having one on each side of a junction in such close proximity would be extremely unusual (if not unheard of in the UK).

I don't have the answers, just suggestions and so it is going to have to be down to the context. On a mixed-urban A-road, we are not going to be able to signal every single side street, although we can be clever with filtering them. As David Hembrow has said, the motor traffic reduction is key - how we get there is one for the politicians. 


  1. Something like this junction in Groningen could work here. It has simultaneous green for all cyclists and pedestrians.

    1. Indeed - I have an early sketch which does this. The sketch in the blog has a right turn waiting area which does make the sim green a bit harder. Perhaps a pair of parallel zebras could be a solution too!