Tuesday 4 November 2014

Traffic Signal Pie: A Midnight Feast (For Some)

Richard Tracey of the GLA Conservative Group has recently released a paper proposing to switch off traffic signals in London at night in order to save London's motorists £40m.

Tracey, the GLA Conservatives spokesman for transport, states;

“Every year Londoners waste over 170 million hours sitting in traffic, costing London’s economy £4bn. Many of these journeys in our city are unavoidable. But rather than hurting motorists with ridiculous charges and taxes, we should look at innovative ways to cut congestion and make traffic flow more smoothly. 

Turning off traffic lights at night, like they do in parts of Europe and North America, is one measure which would boost the economy and help the environment. A common sense approach in the right places would cut idling and therefore vehicle emissions, motorists would save cash as less fuel is wasted, and journey times would be slashed meaning deliveries are completed quicker and cabbies are able to take on more jobs. Even if lights were turned off for just six hours overnight, accounting for non-suitable junctions, drivers could save £40m over four years in saved time and fuel alone.”

Obviously, my "common sense" and "smoothing traffic" klaxons went off together and I thought it might be interesting to explore the paper and offer some other views. Let's start with the statement above. 170 million hours wasted in traffic sounds like a huge number (and I will come to the source shortly), but if London has about 8 million residents then this means that each person spends (on average) 21.25 hours a year stuck in traffic. Or 3.5 minutes a day. Doesn't seem that bad to me.

The source of this data is a 2006 report which is based on data from 2003 which in my book might make for some interesting historical reading, but is so out of date as to be irrelevant. The 170 million hours is derived from the report's suggestion that there is an annual delay of 10,250 million minutes (Table 2) and this figure is arrived at through some complex calculations;

But, hang on, the people for whom Tracey is so concerned are sitting in traffic for 170 million hours between 7am and 7pm, so what on earth has this got to do with switching off traffic signals at night? I can answer that, it has nothing to do with it. This is pure, populist politics. The data is not derived based on people sitting at traffic signals (although this might be one of many reasons for congestion - more on that in a bit). But, let's stick with it.

The paper suggests that this is costing London's economy some £4 billion a year, although the 2006 report suggests £1.6 billion (paragraph 5.2). Sadly, Tracey's reference isn't hyperlinked and doesn't show up in searches. There is reference to the £4 billion in the Roads Task Force Technical Note 11, which seems to have come from the same source which was for 2008/09, published in 2010. Certainly, the 2006 report uses the Department for Transport's COBA Manual (Table 1/1) figure of 930 pence per hour for the "average" vehicle (an average for drivers, passengers, buses, commuting, non commuting etc). I assume that there will be annual uplifts for the figures to get us to the £4 billion (which is quoted at £17 per hour). Of course, these "costs" do not factor in those accrued by pedestrians trying to cross busy roads or any costs to those riding bikes being delayed.

The RTF Technical Note 11 references the US Federal Highway Administration (FHWA) factors for congestion and network reliability (with a bit of London spin!):

Category 1 – Traffic-Influencing Events

1. Traffic Incidents – disruption to the normal flow of traffic usually by physical impedance.

2. Road Works – activities that result in temporary physical changes in road capacity.

3. Weather – certain conditions can lead to changes in driver behaviour that affect traffic flow.

Category 2 – Traffic Demand

4. Fluctuations in Normal Traffic – Day-to-day variability in demand leads to some days with higher traffic volumes than others. If the capacity of the network is fixed this can lead to variable travel times without any Category 1 events occurring.

5. Special Events – Are a special case of demand fluctuations where traffic flow in the vicinity of the event will be radically different from “typical” patterns.

Category 3 – Physical Highway Features

6. Traffic Control Devices – Intermittent disruption of traffic flow by control devices such as poorly timed signals can contribute to congestion and travel time variability.

7. Physical Bottlenecks (“Capacity”) - the maximum amount of traffic capable of being handled by a given highway section. Typical bottlenecks in London are caused by fixed bridge and tunnel capacity. Other factors that impact traffic capacity include bus lanes and cycle lanes.

For interest, I reproduce the FWHA's suggested causes of congestion (which is in the US of course). It is interesting as it shows that the causes of congestion are many and the FWHA's report admits that the matter is complicated and congestion can vary at location and day.

The London spin of course cannot resist mentioning bus lanes and cycle lanes as an impact on traffic capacity which of course completely disregards the efficient movement of people - a bus lane only "takes" capacity for other drivers when it is built, but it must be offset by passengers who have a more efficient journey. Cycle lanes would only take capacity if from existing traffic lanes when a new facility is built which cannot be used by motorists (i.e. mandatory lanes or cycle tracks). 

The three categories are related, but if we strip out unplanned things like weather, planned capacity reductions such as roadworks and special events, we are left with demand and capacity. Where demand exceeds capacity, we get congestion and journey reliability gets worse. Perhaps if Tracey had explored some of these issues, the paper might have come across as being a little more authoritative.

The RTF Technical Note 11 also gives an interesting little bit of information which was some modelling done on Victoria Embankment in preparation for the Olympic Route Network signal timing review. Essentially, the modelling looked at the relationship (if any) between the degree of saturation (DoS - there is more detail in that link) and journey time reliability. It showed that with a DoS above 80%, journey time reliability decreases.

If 100% is the theoretical capacity, then running at 80% will mean the flow is smooth (laminar to extend a water flowing through a pipe analogy). At 100%, there is "friction" (people pulling into and out of side roads, people braking and people behind them following suit, people crossing the road when invited etc). The DoS for a road link will vary because of the impacts on it, but the point is that 100% is not the most efficient flow. Of course, there has been a reducing traffic trend in London for a while now (although current road building policies seem determined to reverse this) and so if this continues, then congestion would reduce and hence the economic cost for those driving during the daytime which Tracey uses as a prop for the argument of people being stuck in traffic because of traffic signals in use at night.

The third reference in the paper is from 2010 where it is stated that London has 6,000 sets of traffic signals (up from about 4,800 in the year 2000). It is no secret that the Mayor of London is pro-car and has had a policy of "smoothing traffic flow" which includes removing traffic signals. In his 2008 paper "Way To Go!" the Mayor stated:

"It is incredible that we have successfully deterred tens of thousands of vehicles from entering the Congestion Charge zone, and yet congestion – and frustration – have continued to rise.

We are now in the process of reviewing all of London’s 6,000 traffic lights, and already we have shaved seconds off red on about 150 of them. A couple of extra seconds on green can cumulatively make a huge difference to traffic flow, and where it is possible to make a difference without prejudicing the rights of pedestrians, we will do so.

We are reviewing 1,000 lights per year, and in many cases we will be asking what the lights are doing there in the first place. Why were they put there? What risk were they addressing? Can we address that in any other way, without bringing traffic to a standstill?"

So, let's be honest, Tracey is just repeating the company line. Don't get me wrong, there is nothing intrinsically wrong with reviewing traffic signals and as well as timing reviews (because traffic patterns can change on a local or network level over time) there may be justification for removals (for example a bypass might render a set of signals in the bypassed area redundant). 

According to the guys at work who have been around a while, before TfL came into existence, traffic signals in London were run by the TCSU (Traffic Control Systems Unit) and it was the case that new signals were funded by the TCSU (the "civils" works were paid for by the borough promoting the scheme). At least in my area, I know of stand alone pelican crossings which were put in because it was cheaper to do so than zebra crossings as someone else funded the signals kit (at least at the point of use - the costs were paid through precepts). TfL now expects that boroughs fund new signals as part of their schemes and with budgets being the way they are, people tend to use them more sparingly these days (and of course there are annual maintenance charges). But, I digress (as usual).

Why do we use traffic signals? According to Tracey's paper;

"There are three reasons why traffic lights are used. The first of these is safety. There are some junctions where traffic signals play an important role in reducing the chance of accidents or injuries.

The second is amenity. Traffic signals can be useful in providing a safe route to a school or reducing the effect of a major road segregating a community. 

The third reason – and the one that has been the key driver in the growth of traffic lights over the last decade – is to help implement wider transport policies, such as bus and cycle priority."

Safety is a loaded point. There are layouts which could be redesigned so that traffic signals could be removed and with any given set of signals (junction or crossings) statistically, there will be collisions occurring as a consequence of how people behave with them. For example, people jump red signals when other people have priority or people stop when they should and get shunted for their trouble. If someone walking or riding a bike is involved, the outcome can be serious or fatal. In 2012, nearly 5000 people were hurt in collisions at traffic signal-controlled junctions (Table 5.2). Statistically, a signalised crossing could attract a higher rate of collisions than it did pre-signalisation.

Compared with situations where there are no signals, it can be difficult for people to get out of side roads or cross roads and so signals help and so there may be suppressed demand anyway - the introduction of signals therefore allow people to move through junctions or across roads where it might have been difficult or dangerous to do so without and this in part could lead to an increase in collisions. Signals are a tool which can enable people, especially those walking and cycling and particularly the elderly and the young. It kind of links to the second point, but this is of course far from "amenity" (what a curious word to use) for many people signals are vital.

The third reason is absolutely fair and what it tells me at least, is that in order to provide for cycling (and indeed walking), we are going to need to use an awful lot more traffic signals to make things feel safe for people on many roads (using or crossing) and this links back to the first point anyway. Of course, this cannot possibly square with the Mayor's policy for removing traffic signals - but they are a vital traffic management tool and so we should be prepared to pay for them to be used.

The paper also quotes a Greater London Authority paper - "Economic Impact of Traffic Signals" to justify turning signals off at night. The paper uses modelling of 5 London junctions to compare a "do minimum" (essentially tweak the signals for efficiency) and remove the signal control. There was a big assumption;

"In modelling traffic movements some assumptions are needed as to how traffic will react without signals. When the traffic signals are removed traffic is assumed to give-way to the right as normal on roundabouts, to give-way to traffic on  the right on 4-arm junctions and to revert to major-minor road status for 3-arm junctions."

It would be interesting to see how a 4-arm junction would be laid out with signals turned off as the modelling assumes that people would give way to the right which is not a UK convention in the event of a traffic signal failure. For a 3-arm junction, reverting to major/ minor would mean that people entering/ exiting the minor arm may well be disadvantaged. As for signals on roundabouts, aside from the ability to incorporate stages for walking and cycling, signal control is often used at intersections of big roads to stop them locking each other up - I am thinking of places such as Bow, Redbridge and Hanger Lane.

In terms of impacts on pedestrians, the GLA report does not cover benefits and disbenefits and for road safety, the commentary suggests that where traffic signals fail this could equate to around 9 PIAs per year (personal injury accidents) compared to 2.4 PIAs per year normally. Failed signals is not a proxy for turning them off, but interesting background nonetheless.

When we finally get to the proposal to switch off traffic signals at night (where congestion is least and journey reliability highest), the GLA report considered modelling of the study junctions at various times of the day and this included an "off peak" between 22:00 and 01:00. The study looked at the economics of removing the signals, but stresses that this is for the study junctions rather than something applicable to the whole of London - each case on its merits then, which is what good engineering advice is about. There is a slight economic benefit in all 5 cases for no signals during the off peak (22:00 to 01:00) period which I have to assume is the thrust of Tracey's point.

You should read the full conclusions of the GLA report, but like any good research, more work is needed;

"In the UK legislation does not allow for the use of switching all signals at a junction to flashing amber at less busy times, a measure which is commonplace in a number of European countries. We recommend discussions should take place with the appropriate European traffic authorities to obtain evidence and ascertain their views on the impact that such traffic control methods have on safety, vehicle and pedestrian movement."

Of course, other European countries also have vehicles turning right on red (giving way to pedestrians and bikes going ahead), arrow-based red signals (for lane control) and where flashing amber signals are used, pedestrians can have priority as the crossings revert to zebra crossing equivalents (please correct me or add in the comments as I don't know enough about international set-ups). I have to conclude that even if a flashing amber was used (in lieu of a complete switch-off) then we need to change our regulations and given that the Department for Transport is killing off the Pelican crossing in the proposed changes to the Traffic Signs Regulations & General Directions, we will have no flashing amber signals at all!

In the economic part of the paper, Tracey argues that £10m per year could be saved across London by turning traffic signals off at night (i.e. £40m over 4 years). The saving is accrued by drivers of course and does not add in disbenefits to pedestrians. TfL is currently rolling out SCOOT which allows junctions to adapt (within parameters) to changing traffic patterns. At night, this translates to more efficient detection and cycle times and probably renders switching off signals as irrelevant now. Of course, Tracey's calculation makes no assessment as to what the costs in terms of new collisions could be - I would lay odds that would more than £10m a year.

If I put human cost to one side, the current economic costs of collisions are as follows;

Essentially, if a programme of turning signals off at night went ahead and there were 5 fatal collisions (I prefer that term rather than "accidents"), then that wipes out the "savings" made by drivers - seriously, is it worth even trying this on pure economic grounds. I don't think so. Is it worth doing to save a few people a few seconds on their night time journeys even if others get hurt or killed? Of course it's not.

Fancy coming back at night for a bit of cross-traffic chicken?
On the TfL trunk road network, the main roads are often set with 40mph and 50mph speed limits and the traffic signals rest on green on these routes anyway. Side road traffic triggers a demand and the high-speed arms get a red to allow side road traffic to access or cross. Certainly, when there is a traffic signal failure on these junctions, TfL's normal protocol is to close any right turns off the trunk roads and to prevent traffic crossing the trunk roads (in essence you can turn left on or off the trunk road). I find it hard to imagine TfL being happy turning signals off at night on high speed roads.

What about in Central London? Lots of areas are busy well into the night with traffic, bikes and pedestrians (I am thinking the West End for example) and so I cannot see those junctions being suitable for switching off. Where does Tracey suggest we turn signals off? Well, he gives some recommendations;

1. Where it is deemed safe, TfL and London’s Boroughs should turn off London’s traffic lights between midnight and 6am.

2. TfL should, as a pilot, turn off traffic lights at 100 junctions during daytime off-peak hours. London’s Boroughs should look for opportunities to do the same.

3. TfL should perform regular wide scale reassessments of their current traffic lights to see if any are redundant under their current standards.

In response to the first point, I have given a couple of types of location where it wouldn't be safe to switch off signals at night. I have also thought about my own area and I would not wish to put my name to any report or decision to switch signals off - good luck with that one.

On the second point. Oh, hang on, the paper was about switching off signals at night and now Tracey fancies some during the day too - how on earth are people going to be able to cross the road under those conditions? Yes, nothing wrong with the third point as things change over time and actually signals may no longer be needed and other things could be looked at (such as roundabouts with zebra crossings on all arms).

At the end of the day, Tracey has (like me of course) picked on snippets of information and used them to justify a political position which is chiefly to turn off traffic signals at night. He has mixed up various studies and papers to get the result he desires and as far as I am concerned, I simply do not think it holds up to scrutiny and I am sure the modelling experts and academics can pick a few more holes in the ideas too. To be fair to Tracey, a politician's job is to question the status quo and to ask searching questions about accepted wisdom or even why is something done a certain way - yes, we should review how traffic (in the widest sense to include all highway users) operates in our city, but Tracey comes at it from the put down motorist's point of view;

"rather than hurting motorists with ridiculous charges and taxes, we should look at innovative ways to cut congestion and make traffic flow more smoothly."

That is where I take issue on all of this. Tracey's congestion is that experienced by the private motorist. His traffic is the private motorist. Congestion (and indeed journey reliability) is complex and the solutions can be complex, but it always comes down to the point where demand exceeds capacity. The ideas in the paper are tweaks to improve night-time capacity, but this is when congestion is not an issue anyway and so makes this a political dogma - similar to that of people wanting to drive faster on motorways at night, or ignore 20mph limits outside of school times.

All in all, the paper is an extension of the Mayor's car-centric side and is in direct opposition to making London a more walkable and bikeable (indeed liveable) place. The position also has to be anti-bus, as technology is used to give buses priority through junctions. It also hints of a disregard for road safety in the widest sense - all to save £10m a year in a city of 8m people. If only there were other ways of moving lots of people around efficiently.


  1. In the Netherlands, all Dutch light controlled junctions also have standard give way markings. This allows them to work as priority junctions when the lights are turned off at night (they flash amber rather than going off completely).

    1. Thanks Paul, so to be applicable here, that is another raft of changes for the DfT to deal with. Expect the won't as they are rushing to get TSRGD2015 out before the elections!

  2. [it doesn't seem to be letting me comment under psychobikeology for some reason]

    a political dogma - similar to that of people wanting to drive faster on motorways at night, or ignore 20mph limits outside of school times.

    Well quite. My first thought as I read about this report in your post, was to recall my brief period of being a london "motorist" (hawk, spit). Driving late at night was just lovely - it came close to matching the fantasy of driving that we all have lurking in the cultural imagination - and ooh, if we could just eliminate the need to stop occasionally, it might match the fantasy completely. I don't suppose Tracey was all concious of this - I'm sure he sincerely thinks this is all about sober economic benefit and whatnot (only us hairy green hippies ahve opinions connected with feelings for goodness sake) - but I bet that's what underlies it.

    1. My driving fantasy is the A39 Atlantic Highway in Devon and Cornwall, but I have used it when on holiday rather than every day and I am sure it would get boring!

      The paper and the references do exclude data and costs (on the whole) for the disbenefits accruing on the safety issues - and that includes the safety of drivers, no hairy green hippy analysis there!

      My problem is that it is an idea which doesn't stack up to any half decent scrutiny. I am not a signals expert or an economics expert, but if I can find flaws, I would like to see what cleverer people can make of it all!

  3. They do ask for feedback at the end of the paper:
    "Connect with us online and tell us what you thought about this paper."

    So feel free to contact them & let them know what you think:
    Twitter: @assembly_tories
    Facebook: facebook.com/glaconservatives
    Email: assembly.tories@gmail.com

    1. OK, in the interests of fairness and their right of reply, I will tweet them the link to this post!

  4. From my (admittedly relatively limited) experience I can imagine that many people who regularly walk in the centre of London spent a lot more than 3.5 minutes a day waiting (often in the cold and rain) at badly timed pedestrian crossings. Presumably however nobody considers it a priority to do anything about this - if anything getting rid of lights will only make things worse for people on foot.

    1. Loads of places where crossing waits are long and crossing times are short. Some of the worse are where it takes several separate crossings to cross diagonally - take Henlys Corner: http://goo.gl/maps/7XPdY

  5. Turning off the lights (or even better, using flashing amber lights in all directions, eliminating their use with pelican crossings, and adding yield sign and intersecting road signs to assign which direction goes first, adding a few things like a detection system to determine when and where there is a volume of traffic too high to allow the lights to remain unsignalized. This would be complicated math formulas I have no idea which ones apply here but either way, (I should have clarified that yield signs are now North Americans talk about give way signs) attaching those to the traffic light pole, adding flashing yellow walking man lights, and flashing yellow arrows, left and right turns, and flashing yellow bicycle signals, those all go off. Some intersections are so big that it would be best if pedestrians and bikes could go under the intersection like this : https://bicycledutch.wordpress.com/2009/07/18/new-bicycle-tunnel-s-hertogenbosch/ or over the lights like this: https://bicycledutch.wordpress.com/2013/02/28/cycle-bridge-enschede/. This keeps cyclists and pedestrians from having to cross roads way to large even when that road is low volume during off peak times. Having median refuge islands, not staggered, when crossing, ideally between each and every lane, or at least between the left turn lane and through lanes, through lanes and right turn lane and the middle. In addition to safe cycle paths, 3-4 metres wide for bidirectional use, 2-2.5 metres for unidirectional use, and footways 2-4 metres in width, this makes them safe. If you add a raised table or humps on crossings for pedestrians and bikes designed to be taken comfortably at the speed limit or less, it reduces speeding. Cameras also help. Narrower lanes make it feel not good to speed by design. Using zebra type crossings for pedestrians and elephants feet in the direction that pedestrians and bicycles have priority, (even then, median refuges help a lot), low turning radii, they all help a lot.

    1. Unfortunately, none of that is permitted under UK rules!

    2. Perhaps a personal recommendation to get the DfT to actually metricate the British road system and properly implement the Vienna Convention would help.

    3. Here is Munich we have similar to the Dutch intersections. Lower capacity intersections have signs which show priority for each direction. Also the lights blink yellow to advise drivers of the condition.
      In the US the lights seem to work all the time. Some having a trip device to let the less travelled road remain red and then turn green after the device in triggered. In the US when the lights at an intersection are not operating the rule is that it is to be treated as a 4 way stop. There woiuld have to be an effort to put in a flashing yellow and/or signs. Plus education because way too many drivers don't seem to know, or care, about the intersection being a 4 way stop.