So the Transport Research Laboratory (TRL) have published their findings into the safety of cycle track design at junctions – or, more specifically, Trials of segregation set-back at side roads [pdf], to give the report (PPR703) its full title. This report was commissioned by TfL.
I’m going to go into some detail about it, but in short –
- Good (Dutch) junction design is completely ignored by this trial.
- Confusing give way markings are employed – Dutch markings, employed the wrong way round.
- The report recommends ending cycling tracks, and ‘merging’ cyclists with motor traffic, some 20 metres from junctions on roads with higher speeds.
- It then suggests that this ‘merging’ corresponds to Dutch design practice.
It’s worryingly bad.
From the report summary –
This report provides an overview and interpretation of the key findings from four trials carried out by TRL on behalf of Transport for London (TfL) to investigate the effects of ‘setting-back’ a kerb-segregated cycle track at different distances from a side-road junction.
What do TRL mean by “‘setting-back’ a kerb-segregated cycle track”? There is an explanation in one of the photographs in the report –
Clearly, ‘set-back distance’ is being used to refer to the distance from the junction at which the cycle track becomes a cycle lane. So, this TRL report investigates the different consequences of different ‘set-back distances’ – i.e., how far from the junction the kerb separation ends.
And nothing else.
No other forms of junction design incorporating cycle tracks (designs we’ll come to in a moment) are investigated.
Why was this study so narrowly focused? The explanation comes in the summary –
A review of existing international guidance and research on approaches for taking cycle lanes across side-roads identified two distinct design strategies. Either:
• Cyclists are returned to the carriageway level at least 20m before the junction, so as to establish their presence in the traffic, or
• Segregation is brought right up to the junction (typically <=5m) and very tight geometry (and often raised crossings) used to keep turning speeds down and encourage vehicles to cross the cycle lane at close to 90 degrees. [my emphasis]
Amazingly, these two strategies – ending the segregation more than 20m from the junction, or ending it 5m or less from the junction – are the only two distinct design approaches TRL identify, and consequently the only ones they investigated.
Both these strategies involve turning a cycle track into a cycle lane at the junction. The only difference is the point at which that change occurs. Other design approaches – those commonly employed in the Netherlands at side roads – have been completely ignored. These include –
Continuing a cycle track through a junction, at the same raised level, alongside a continuous footway. Not investigated by TRL.
Setting back the cycle track from the carriageway, providing an area in which motorists can wait, both to enter the main road (without obstructing the cycle track) and also to pause, yielding to people cycling. Not investigated by TRL.
This technique can also be employed with two-way tracks; again, set back from the carriageway, with a waiting area, and good visibility as cyclists and motorists cross perpendicularly. Not investigated by TRL.
To repeat (I can’t labour this enough) – these kinds of techniques are completely ignored by the authors of this TRL report. The only two ‘distinct design strategies’ investigated amount to nothing more than on-carriageway cycle lanes across junction mouths, with no investigation of designs that continue a cycle track through the junction at a raised level, with continuity, with or without ‘set back’ from the carriageway.
This despite the fact that Britain itself already has a few isolated examples of reasonably well-designed cycle tracks across junctions, that correspond approximately to Dutch design. One of them – this one – is only two miles from the Transport Research Laboratory!
I can’t begin to understand this oversight.
So the results of this trial are really very narrow in scope, and essentially amount to nothing more than discussion of where it is best to revert to an on-carriageway treatment on the approach to a junction.
The trial examined ending the physical segregation 30 metres from the junction, up to 5 metres from the junction, in 5 metres increments. The ‘tightest’ geometry still involved the cycle track ending 5 metres before the side road.
The study found that with the kerb divider continuing closer to the junction (but still 5 metres from it), drivers turned into the side road more slowly, and crossed the cycle lane (for this is what it is, and how it is described in the report) at an angle closer to perpendicular.
Surprisingly, perhaps, it seems that drivers in the trial actually preferred segregation that continued closer to the junction. This was even the case for drivers of goods vehicles who – you would think – would prefer a less tight geometry, to manoeuvre their larger vehicles.
The preferred set-back distance for 62% of the [goods vehicle] drivers (who expressed an opinion) was one that maximises segregation from cyclists on the approach to the junction
cyclists were divided in preferences for short or long set-back distances. The differences reflect different views on the benefits of segregation, including cyclists’ concerns about being able to position themselves for passing the junction and that drivers wouldn’t give way when turning across their path.
‘Position themselves for passing the junction’ – i.e., compensate for poor design. These findings are reflected in this table –
While a clear majority of drivers preferred separation continuing as much as possible, a large number (nearly half) of the cyclists in this trial preferred to ‘join traffic’. The report comments
this suggests that cyclists may feel safer if the segregation ends before the junction so they can merge with the traffic before the turn.
So a large proportion of cyclists in this trial clearly like the idea of ‘merging’ with motor traffic before a junction. (At this point it is worth asking whether these cyclists are representative of the general population, or instead representative of a small subset of the population, namely the ‘traffic-tolerant’.)
However, on the other hand, the motorists in the trial didn’t really understand what on earth was going on with the concept of ‘merging’.
The purpose of the segregation set-back was not well understood [by motorists] – most believing it to be to make it easier for vehicles to turn [!], only a few referred to it providing space for cyclists and drivers to adjust to each other before the junction.
Could it be that the idea of ‘merging’ people cycling and driving isn’t all that intuitive?
This suggests that there is a lack of understanding amongst drivers of how cyclists will behave at the junction.
Amazingly, however, this ‘merging’ technique is actually recommended by this TRL report on roads with higher speeds.
The findings from the off-street trials suggest that two different strategies can then be considered:
- Bring segregation very close to the turning (<5m), sufficient to reduce the turning radius and so reduce turning speeds and position turning vehicles at right angles to the path of cyclists (this is similar to the principles behind the use of ‘continental geometry’ at roundabouts). This approach would be most appropriate where geometry is already tight and vehicle speeds comparably low, or where other measures to achieve this will also be implemented.
- End the segregation at least 20 m from the junction, giving cyclists sufficient space to re-introduce themselves into the traffic flow and for drivers to adapt to their presence. This would be more suitable where traffic speeds are higher and tight turning geometry is not considered to be appropriate.
Before then stating
These two situations are consistent with the two distinct design approaches adopted in the design practice sin countries such as Denmark and the Netherlands
Well, I’m sorry, I have never seen anything like this in the Netherlands, especially not on roads with higher speeds. It’s just terrible design.
And we know that this is bad design, because Transport for London have built the junctions on Stratford High Street like this, with predictable consequences.
The final boggling issue are those markings! – which make no sense whatsoever. Here’s how the report describes them –
The surface of the cycle lane was coloured green throughout… additionally using triangular give-way markings to highlight the cycle lane for turning vehicles. These markings are not an approved road marking in the UK, however somewhat similar versions are used in the Netherlands as a ‘give way’ marking.
‘Somewhat similar’ – except completely the wrong way round.
The Dutch use ‘sharks teeth’ as a give way marking, but crucially with the ‘sharp’ bit of the teeth pointing at the people who should be giving way. This trial, however, has managed to get this completely wrong, with the ‘teeth’ pointing at the people on the cycle track. In addition, the ‘sharks teeth’ markings have been employed across the whole of the junction, rather than just on the ‘entry’ side, where drivers would be giving way. So, used in conjuction with a British give way marking, is it any wonder people driving didn’t understand this marking?
This failure to get even the basics right is symptomatic of the general failure of this trial to assess proven Dutch junction design in a British context. How is it possible for the Transport Research Laboratory to have what seems to be absolutely no clue about how the Dutch design well at junctions?
What on earth is going on?