The effect of cycle lanes on the proximity between motor traffic and cycle traffic

An experiment collected proximity data of motor traffic overtaking cycle traffic on roads with and without cycle lanes using an instrumented bicycle. The work enhances previous research which has considered the riding position of the cyclist and whether or not the cyclist was helmeted, while controlling for vehicle type. The analysis shows that significantly wider passing distances are adopted by motorists in the condition without a 1.45 metre cycle lane, with posted speed limits of 40mph and 50mph with a 9.5 metre wide carriageway. These findings were not replicated for a similar width road with a posted speed limit of 30mph and a 1.3 metre cycle lane. The results suggest that in the presence of a cycle lane, drivers may be driving within the confines of their own marked lane with less recognition being given to the need to provide a comfortable passing distance to cycle traffic in the adjacent cycle lane.


Background
Cycling has environmental, social, energy and congestion benefits through reduced motor vehicle use and confers health benefits on the user. However, the perceived risk of cycling is a deterrent to its wider uptake, as discussed in, for example Henson et al. (1997), Davies et al. (1997) and Gardner (1998). Routes for cycle traffic include every part of the highway network (apart from, for example, motorways, which are restricted to motor traffic use) and other off-highway routes which may form convenient shorter routes between parts of the highway network (such as permissive routes across, for Cycle lanes may offer a greater degree of separation between the cyclist and the motorist. They may also: usefully direct cycle traffic to the most appropriate position within the carriageway; provide a legal means for cycle traffic to undertake motor 3 traffic in queues approaching junctions; provide a degree of continuity and conspicuousness of routes for cycle traffic (Lancashire County Council, 2005).
However, as cycle traffic may wish to carry out a variety of manoeuvres within the carriageway, then the presence of a cycle lane may adversely affect the way cyclists use the carriageway, for example, feeling unnecessarily inhibited in moving to the right of a motor traffic lane when carrying out a right turning manoeuvre (left hand rule of the road). Motorists may also wrongly assume that the presence of a cycle lane means that the remaining parts of the carriageway will be free of cycle traffic.
The experience of one of the author's in training UK traffic engineers indicates that Dutch cycle design guidance is often regarded as being the most appropriate guidance available for western countries. This may be based on the false notion that high bicycle usage in The Netherlands is entirely due to high design standards and implementation.
There could be many other explanations for high levels of use in The Netherlands including a long history of a culture of cycling. The guidance (CROW, 1993 and2006) does, however, helpfully differentiate between highway cross-sections as follows: 'spacious', which allow motor traffic and cycle traffic to pass each other comfortably without encroaching into oncoming traffic in the adjacent lanes of an undivided carriageway; 'tight', which are so narrow as to require cycle traffic to follow motor traffic and vice versa; and 'critical', which are some way between 'spacious' and 'tight' and may create the most risky situations because overtaking may occur, but within a carriageway which is not sufficiently wide to allow this to happen comfortably.

4
Dutch research (CROW, 2006) shows that motorised traffic will nearly always pass cycle traffic when the bicycle to motor vehicle distance is 0.85 metres or greater. At 30mph, and where the overall width permits, the passing distance is typically around 1.05 metres. According to the Federal Highway Administration (FHWA, 1975), such a clearance produces a lateral force of under 9 Newtons (2lb). While this force may be relatively low and less than the force under normal service braking, the fact that it is induced in the cyclist by the actions of others (the passing traffic), then, from a psychological point of view, this may exceed a level deemed comfortable. The passing dimensions are usefully depicted in cycle guidance prepared by Lancashire County Council (LCC, 2005) as shown in Figure 1 for a carriageway 8.5 metres wide.
[Insert Figure 1 here] A wider carriageway of 9.5 metres, and assuming equal gaps between motor vehicle and bicycle and motor vehicle and motor vehicle, suggests a passing distance of 1.38 metres.
With a significant presence of Heavy Goods Vehicles of width 2.6 metres, the carriageway width would need to be 10.1 metres wide. The Dutch advice has also been carried through into recent current United Kingdom guidance (DfT, 2008), which suggests ideal total minimum widths for overtaking on a carriageway with a speed limit of 30mph (48kph) of 4.3 metres, or 5.05 metres with significant numbers of heavy goods vehicles.

5
The new UK guidance suggests cycle lanes should be 2 metres wide on busy roads or where traffic is travelling in excess of 40 mph (64 kph), but that 1.5 metre lanes may generally be acceptable on roads with a 30mph speed limit. The guidance notes that cyclists may need to move away from the kerb to avoid surface hazards and this may 'give motorists misplaced confidence to provide less clearance while overtaking than they would give in the absence of a cycle lane'. This assertion has potentially serious implications, particularly for narrower cycle lanes, and is tested by the research presented here.
Most Northern European countries assume driver liability in collisions with pedestrians and cycle traffic for insurance purposes, with the burden of proof falling on the driver to prove that he or she was not liable. This is sometimes (inaccurately) referred to as 'strict The data collected for the analysis presented here were obtained using an instrumented bicycle to measure the passing distance of vehicles relative to a cyclist along six sections of road. The sections have posted speed limits of 30mph, 40mph and 50mph and were sub-divided into sections with and without cycle lanes. Section 2 reviews previous research in the field. Section 3 describes the methodology, Section 4 presents an analysis of the results. A discussion of the results is presented in Section 5 and Section 6 draws conclusions.

Previous research
Motor traffic passing a cyclist exerts a lateral force because of the air turbulence created. The Federal Highway Administration report (FHWA, 1975) suggest a tolerance limit is defined as 16 Newtons (3.5lbs), equivalent to heavy goods vehicle traffic travelling at 50 mph, 1.2 metres from the cyclist. This is a little less than the force experienced during normal service braking.
Considering the physical presence and effect of traffic in a psychological way, Sorton and Walsh (1994) showed that cyclists could recognise aspects of the mental effort of cycling as being related to levels of traffic volume, motor vehicle speed and lane width. The Federal Highway Administration reviewed the operation and safety of similar sections of route with and without cycle lanes (FHWA, 1999) and found significantly higher rates of conflict between cycle traffic and motor traffic at sites with bikes lanes as compared to sites without, although the rates were small compared with rates with and without cycle lanes on the approaches to junctions.

8
In the United Kingdom, Guthrie et al. (2001) attempted to create an index of 'cyclability' on a ten point scale (1 bad for cycling, 10 very good for cycling) based on cyclists' assessment of road and traffic conditions. Fifty-one cyclists rode a 9.2 kilometre route comprising eleven links which were generally non-urban in nature and the sample was biased towards male frequent cyclists. Lane width was included as a linear parameter in the model and contributes 1.03 times the lane width to the cyclability score, a higher proportion than estimated by Landis et al. (1997) andHarkey et al. (1998). In addition, separate 'safety', 'effort' and 'pleasure' ratings were considered and lane width was found to correlate significantly (p<0.001) with safety (-0.46) and pleasure (-0.42).
In a simulated environment, Basford et al. (2002) found that the provision of cycle lanes appears to increase driver confidence and hence risky behaviour such as higher speeds and less speed reduction when a cyclist is encountered. Stone and Broughton (2003) tabulate incidence and fatality rates for cycling accidents during 1990-1999 from over 30,000 accidents reported using the United Kingdom STATS19 road accident reporting mechanism. They note with interest the much greater fatality rate for cyclists hit from the rear than from the front.
As part of work for the Warrington Cycle Campaign, Owens (2005) asserts from photographic evidence alone that cycle lanes have the effect of reducing overtaking distances, suggesting a demand for further knowledge and understanding about overtaking distances amongst the cycling community. In a survey using an instrumented 9 bicycle on roads in Bristol and Salisbury, Walker (2007) found that the further out in the carriageway the cyclist rode, the less space is received from overtaking vehicles; drivers generally pass closer to a helmeted cyclist; and drivers of buses and heavy goods vehicles pass closer than other types of vehicle. The work did not take account of available carriageway widths or the widths of the passing vehicles. He recommended that the effect of on-road cycle lanes be investigated, and this demands that proper attention is given to available road width.
In order to improve on research into the perception of cycling that had hitherto only considered links, Parkin et al. (2008) used video clips of routes and junctions from the point of view of a cyclist and presented them to cycling and non-cycling commuters.
The Risk Ratings for combinations of routes and junctions which mimicked real potential journeys by the respondents were on a scale of 1 (lowest perceived risk) to 10 (highest perceived risk) and were used as the dependent variable in a logistic regression model constrained to lie within the Risk Rating range. The model did not explicitly consider width, but flow passing the cyclist was found to be significant.
The literature suggests no common definition of the disutility associated with cycling: sometimes it is considered on a measure purporting to be a 'level of service', sometimes a 'compatibility' or 'cyclability' index, the components of which include issues connected with safety, effort and pleasure, or it has been considered as a risk rating.
There is no commonly emerging functional form for the inclusion of passing distance as a measure of disutility, and the contribution of passing distance to the overall disutility appears to vary between studies. Design standards appear to be based on observed passing distances, but there is no correlation suggested between these observed distances and perceived comfort for the cyclist. The effect on passing distance of the presence of a cycle lane needs to be more fully understood.

Methodology
An Using footage of overtaking manoeuvres collected whilst cycling on roads both with and without cycle lanes and using the front wheel of the passing vehicle as a reference point, the proximities of the overtaking vehicles were established. The bicycle was checked to ensure that it was calibrated correctly after each period of data collection.
Three sites were selected for analysis and had posted speed limits of 30mph, 40mph and 50mph (48kph, 64kph and 80kph). The characteristics of the spread in speed for the roads surveyed is provided in Table 1. Each site contained stretches of road with and without cycle lane. The sites were all virtually straight and flat in order to eliminate horizontal and vertical geometry variables.
[Insert Table 1 here] Site 1 (50mph) is on the A6 at Cabus, near Garstang, Lancashire, England. The width of the cycle lane is 1.45 metres with an overall road width of 9.57 metres 1 . In the area without a cycle lane, the overall road width is 9.64 metres. Figure 3 shows the two sites.
Site 2 (40mph) is on the A6 at Broughton, north of Preston, Lancashire. The average width of the cycle lane is 1.45 metres with an overall road width of 9.57 metres. In the area without a cycle lane, the overall road width is 9.37 metres. Figure 4 shows the two sites. Site 3 (30mph) is in Westgate, a suburb of Morecambe in Lancashire. The width of the cycle lane is 1.30 metres with an overall road width of 9.45 metres. In the area without a cycle lane, the overall road width is 9.49 metres. Figure 5 shows the two sites.

15
The significantly wider passing distance offered by motorists on the A6 at Broughton without a cycle lane is all the more noteworthy when it is realised that the carriageway without the cycle lane is 200 millimetres narrower than the carriageway with the cycle lane.
The passing distance will be influenced by the available width to the motor vehicle driver, with passing distances being smaller on narrower carriageways. The available gap will also vary depending on whether or not traffic is coming towards the overtaking vehicle, a variability which we have not be able to account for in this experiment.
Considering a car overtaking the bicyclist on the A6 at Cabus in the condition without a cycle lane, it may be seen that the total gap available in the lane is (9570/2)-800-1819=2166mm. The proportion of this dimension which the motor vehicle driver leaves between the motor vehicle and the cyclist provides an indicator of the way that the motorist uses the available road space. Table 3 shows the mean passing distance of cars as a proportion of available space in lane.
[Insert Table 3 here] The mean proportions are all greater than 0.5 and this implies that the motorist is leaving more than half of the available space between the motor car and the cyclist as  Table 3.
The difference between the proportions on the A6 at Cabus with and without a cycle lane (0.701 and 0.772) is significant (p=0.000), as is the difference on the A6 at Broughton (0.520 and 0.579, p=0.000). This is not the case at Westgate. These results simply parallel the results based on the measured passing distance, which is to be expected because the widths of neither the roads nor the motor vehicles themselves vary greatly, at least not in comparison with the variation in measured passing distances.
Further inconclusive analysis has been performed to determine the distribution of the passing distances. It was hypothesised that in the circumstance where there is a superabundance of space within the lane, the passing distance would be distributed standard normal, but in cases where the cross-section is tight a skewed distribution would obtain. The data do not support such hypotheses.

Discussion
The data collected provide evidence that motor traffic passes cycle traffic at closer The lack of a significant difference in passing distances between the with and without cycle lane condition on the road with a posted speed limit of 30mph may be due to drivers not making a conscious overtaking manoeuvre in the condition without a cycle lane.
The data do not support a view as to what a comfortable passing distance should be and this would require further research considering objective measures of comfort, such as lateral force and noise, as well as self-reported ratings of comfort.

Conclusions
It may be concluded that in circumstances where a cycle lane is insufficiently wide for the speed of general motor traffic, drivers provide greater passing distances to cyclists on stretches of road without cycle lanes. Cycle lanes therefore do not appear to provide greater space for cyclists in all conditions. The limited data available on different vehicle types suggest that motor vehicle overtaking proximity also varies depending on vehicle type, and this confirms Walker's finding.
These results should encourage further investigation into the effectiveness of cycle lanes in separating cycle traffic from motor traffic. Differences in lateral separation may affect risk of collision, but may equally affect the perception of journey ambience for cyclists, also an important consideration.