The London Assembly is currently holding an enquiry into traffic congestion in London, and I’ve been asked to give evidence in front of the Transport Committee. I’ve done so several times before, and I’ve been impressed by the level of debate: it’s definitely something you need to prepare for! So I thought I’d start writing a few thoughts on this blog. Comments are very welcome.
First thoughts: where does congestion come from? The Assembly says ‘The causes are complex, but boil down to more vehicles using London’s roads without a corresponding increase in road space. In addition some road space has been closed to motor vehicles to make space, either temporarily for roadworks or permanently for upgrades such as widened pedestrian areas or the Cycle Superhighways.‘
Being pedantic, ‘traffic’ and ‘vehicles’ are not the same thing. Cycles are vehicles, but pedestrians are not. Under Part 2 of the Traffic Management Act 2004, which gives transport authorities the duty to keep traffic moving, ‘traffic’ is defined explicitly as including pedestrians. However, much work on congestion assumes that ‘traffic’ and ‘vehicles’ are motor traffic and (usually private) motor vehicles – including this recent interesting report from London First, of which more in another post.
But if we are really talking ‘traffic congestion’, then presumably we might want to talk about the pressure on footway space generated by public transport improvements, for instance. And a focus on ‘vehicle congestion’ might want to include looking at how close to capacity cycle superhighways are, or indeed bus journey times. Pedestrian congestion is likely to be almost wholly separate in principle from congestion affecting motor vehicles, given London’s roads include footway space from which motor vehicles are generally excluded. (Yes, I know people have to cross roads – but it’s not immediately obvious to me that more road congestion would always lead to more pedestrian congestion in crossing – it might in some circumstances make crossing easier).
By contrast, cycling congestion and bus congestion will be more closely related to that experienced by other vehicles. London has good bus priority compared to many cities (not just bus lanes; also traffic management measures such as prioritisation at junctions) but there are still many places where traffic jams hold up buses. We’re starting to get better cycle networks, but these are still very limited and much of the time cyclists must mix with motor traffic.
A review of congestion indicators finds that basically, there are two ways of measuring congestion. One is in relation to travel times, i.e. these measures ‘contrast the observed or congested traffic conditions (using travel time or journey speed) with some reference level.’ (See also this document from VTPI).
This is what the satnav studies linked to above do; they compare the time taken to drive a certain route in free-flow conditions with the time taken when roads are busier. TfL developed the ‘excess delay’ method as part of the congestion charging scheme (i.e. to measure what actually happened to congestion) and have since applied the principle specifically to buses (because buses both have different expected travel times to cars, and are affected differently by traffic jams, due to bus priority measures).
The other method examines bottlenecks, measuring either time in a queue or the seconds spent standing still, depending on the characteristics of the study area. In the first report TfL produced on the Congestion Charge, both definitions were used, although with travel time rates seen as the main indicator:
Definitions of congestion
Congestion is a consequence of the balance of the capacity of the road network and the intensity of traffic flow. The Department for Transport defines congestion as the average ‘excess’ or ‘lost’ travel time experienced by vehicle users on a road network. Excess travel time is the time spent over and above that which would be experienced under ‘uncongested’ or ‘free-flow’ conditions. These are best measured in terms of ‘travel rates’ rather than speeds.
Congestion inside the charging zone
The uncongested network travel rate for the charging zone is 1.9 min/km. In 2002 the average travel rate in the zone was 4.2min/km, which means that traffic inside the zone was experiencing delays or congestion of 2.3 min/km. This compares to an average across large urban areas in England of 0.4 min/km.
Speed distributions in the charging zone
Another way of assessing traffic conditions is in terms of speed distributions, or, in other words, the proportion of time spent driving within various speed bands. The latest figures show that during charging hours over half of the time spent by vehicles within the charging zone is spent either stationary or at less than 10km/h.
A report produced as part of TfL’s Roads Task Force Evidence Base clarifies a key indicator for measuring congestion:
Congestion on the network is estimated on a daily basis using journey time data from automatic number plate recognition technology. The congestion level is defined as follows: Congestion level = (level of journey time / reference level observed at night [10pm-6am]) – 1.
In other words, the congestion level is equal to 1 when motorists experience a level of journey times which is double the one they usually experience at night.
It’s worth noting that TfL only seem to report on congestion related to the ‘Network of Interest’, i.e. the TLRN, BRPN and SRN (Transport for London Road Network, Borough Principal Road Network, and Strategic Road Network), which I think is aligned to the coverage provided by their ONE model. (Also, those are the routes for which DfT traffic counts are available – see below).
TfL say in a recent report on ‘Total Vehicle Delay for London’: “The DfT have suggested that recent falls in average speeds across London [and hence increases in travel rates] may be partly attributed to a reduction in speed limits in some London boroughs due to the introduction of the 20mph speed limits [see e.g. this], as well as an increase in traffic levels.”
Briefly, this is interesting, because it reminds us – whether DfT are right or not in this particular case – that sometimes an apparent increase in ‘congestion’ might be partly due to planned decreases in speed via policy instruments (rather than just stop-start traffic). (As an aside, it also occurs to me that a further disproportionate increase in congestion might then be measured, if compliance with speed limits is lower at night.) It also argues for using the bottleneck indicator at least alongside others, because that gives you a sense of whether motorists are driving at a fairly constant 15mph (for instance), or half the time stopped and half the time at 30mph. They have rather different implications for injury, emissions, etc., but potentially could lead to the same measured congestion compared to night time speeds.
What are the costs of congestion? I think we need to be careful about some of the scary headline figures that we see. “Total Vehicle Delay in London” comes up with: ‘The estimated total cost of annual vehicle delay on the network of interest within Greater London is £5.5 billion‘.
Where does this come from? The delay measurement is based on TrafficMaster data, also used by DfT to generate congestion statistics, which uses in-vehicle GPS and is skewed towards vans. This is assumed to represent delays to all vehicular modes (but I suspect it doesn’t represent buses well, and am pretty sure it doesn’t represent cycles). Then, in accordance with the standard ‘excess travel time’ methodology, the extra time taken during the peak and interpeak along a link (compared with the time taken at night during the same link) is calculated, and then multiplied by traffic volumes (as per DfT statistics) along that link. (There are a few assumptions made, which interestingly ‘may introduce positive error in the calculations, so the estimates of delay are likely to be maximum values.’)
This ‘total cost’ calculation – like many similar headline-grabbing ‘cost of congestion’ numbers – is in my view pretty spurious. I wouldn’t disagree that congestion, or factors associated with it, may have negative outcomes, including business costs – and I’ll be writing some more about ‘costs of congestion’. But that £5.5 billion doesn’t relate to actual costs, such as, for instance, an increased cost to businesses having to pay contractors to deliver before the morning peak. Neither does it reflect, for example, the potential impact on bus patronage of bus delays. Rather the £5.5 billion is just the WebTAG “value of time” calculation, allocating monetary value to people’s time based on mode and trip purpose (taxi users trump car drivers trump bus users, business trips trump trips to the shops or doctor).
So firstly a plea to look at actual impacts of congestion, including costs (and potentially, benefits – two recent reviews show some studies suggest congested conditions can be safer, on some road types) of congestion. Secondly, if congestion matters, to think about it in relation to all users. TfL have good figures on bus journey times, which allow us to look separately at impacts on buses. But cycles and pedestrians are less well served. Both may be affected by motor vehicle congestion, but the impacts may be different in size and even direction to the impacts on motorists. And both also may have ‘their own’ congestion on separated infrastructure.
Finally, it’s worth remembering that the impacts of congestion will be affected by how we measure ‘congestion’: the comparison to night time journeys may be affected differently by changes to speed limits, depending on whether motorists adhere to them at different times of day. And it’s not clear how this comparison would work in relation to pedestrians and cyclists: both may for instance be affected by adverse traffic light phasing at times when there are fewer pedestrians and cyclists (although again this would also be affected by compliance).
More thoughts to follow…