Back to Part 2: Time, Money, and Routes
While the previous section focused on broader, area-level modelling, I also wanted to say something about junction modelling, and how it deals with cycling. While choice models covering broader transport networks could – if they dealt with cycling better – help us think about changing patterns of cycling at a local and regional level, they are not the type of model most often encountered by local advocates.
Most often, if you hear ‘the model says there will be unacceptable delays’ people are not talking about a large scale network model, but about a small scale model of traffic flow through a particular facility, such as a junction. Many of these are ‘aggregate’ models, although some more sophisticated ones are micro-simulation models, which predict individual vehicle behaviour. In theory these latter could deal with cycling in quite sophisticated ways too, although in practice, I believe they generally don’t and cycles are modelled, if at all, as being small and slow car-type objects that impede traffic flow. (With aggregate models, cycles are explicitly equated to cars as described below).
But cycles are not cars. For example, seasonality of cycle travel is very different from seasonality of car travel. Cycling has specific patterns of concentration by month and time of day (given the greater ability of commuters to cycle in current UK conditions). While in London cycling might be 2% of all trips, on particular routes this is much higher and so on a summer Friday morning rush hour over one of the river bridges, cycles might make up half of all vehicles; yet make up far fewer than this on a rainy November afternoon peak. Cycling environments have to cope with existing maximal flows (not just flows in ‘normal months’, which tend to be outside cycling’s seasonal peaks) as well as planning for – and encouraging – growth. But often, even if the desire is there to model pro-actively for cycling in this way, data will be lacking.
When dealing with junctions, relatively simple aggregate models are used to calculate how many vehicles can clear a given arrangement. Each vehicle type is allocated a PCU (“passenger car unit”) figure, which for cycles is set at 0.2. In other words, five bicycles are assumed to take up the same space as one car. In 2009 David Carrignon argued that in ‘heterogeneous traffic conditions’, the actual PCU of a bicycle could vary between 0.19 and 0.3, depending on the lane width. Where you have narrow lanes, in other words, bicycles will slow ‘traffic flow’ more than when there are wide lanes. However, these findings are not generally taken into account when calculating flow and delay. Another factor that would influence the effective PCU of bicycles would be bicycle flow: with higher bicycle flows, cyclists may ride (or stop, at a junction) in more of a ‘cohort’, taking up more space than if in a narrow line. Related research on motorcycle behaviour (much from East Asia) has demonstrated similar problems with assuming that motorbikes behave like little cars – and cycles are even less like little cars than motorcycles are.
Where comparisons are being made between different infrastructural scenarios (bus lanes, wide and narrow general traffic lanes, cycle tracks, cycle lanes, etc.) then information about how PCUs vary could be important. Providing for cycling is often seen as (and criticised for) threatening motor vehicle capacity, but, potentially, such comparisons fail to take account of how traditional British infrastructure design (integrating cyclists with motor traffic and providing multiple narrow general purpose lanes at junctions) itself generates motor vehicle capacity restrictions.
More fundamentally (in my view), modelling cycles as cycles (not as little cars or fixed fractions of a car, which ignores their changing characteristics in relation to infrastructure, cycle flow, and motor vehicle flow) could lead to us thinking more seriously about how we are limiting cycle capacity, and how infrastructure changes (permeability for cycling, dedicated infrastructure, etc.) might affect cycle well as motor vehicle capacity. If we think about current cycling environments as suppressing cycling capacity (by excluding many people who would cycle if conditions were better), this opens the door to assessing and measuring this, and potentially then including a measure of cycling capacity within flow assessments. But cycle capacity would need to be understood as not just how many bikes you can fit in (as with cars), but also how much cycling the junction will allow (or prevent), given that we know most people (including almost all children and older people) simply won’t cycle in heavy or fast moving motor traffic.
Forward to Part 4: Concluding thoughts
Two observations:
1. Junction Capacity Models don’t normally take into consideration the used/wasted capacity within the vehicles. The bicycle, unlike the car or bus, is always at 100% capacity. See my blog for an example of what the occupancy can be http://kenningtonpob.blogspot.co.uk/2012/10/huge-amounts-of-capacity-on-lambeth.html
2. The bicycle is weighted at 0.2 PCU (Passenger Car Units) on the understanding that you can get 5 bicycles through in the time one car would take. A bus may be 2.5 PCU as it’s longer than a car and takes the time/space of 2.5 cars. Improving bicycle facilities at a junction can result in a significant increase in capacity if one factors in occupancy. Five bicycles, 1 PCU, passing through means five people have passed through; 1 car , 1 PCU, passing through with just a driver means one person has passed through. A bus, full, is very efficient; a bus empty is very inefficient. Using bicycles has results in a five fold increase in capacity over the single occupancy car. The easiest way to increase junction capacity is to penalise vehicles with lots of empty seats.
Pingback: Governing the transport transition | NewCycling