Content uploaded by Tom Cherrett
Author content
All content in this area was uploaded by Tom Cherrett on Jan 16, 2015
Content may be subject to copyright.
MODELLING THE IMPACTS OF SHARED FREIGHT-PUBLIC
TRANSPORT LANES IN URBAN CENTRES
Fraser McLeod, Tom Cherrett
Transportation Research Group, School of Civil Engineering and the Environment, University of
Southampton
Introduction
This paper investigates the effects of allowing freight vehicles access to a central business district
(Winchester) using bus lanes, in terms of the benefits/disbenefits to the different vehicle types
involved. The work undertaken here was done as part of an EU-funded (FP7), information and
communications technology (ICT) project called SmartFreight (www.smartfreight.info), which is
interested in the dynamic control of individual goods vehicles in urban centres by a third-party ‘freight
traffic controller’, an idea which is similar in concept to air traffic control. Besides shared use of lanes,
other concepts being studied in SmartFreight are the use of bus stops at certain times of day for
freight unloading; the use of freight holding areas outside the city centre; and provision of targeted
traffic information to individual lorry drivers to enable them to make their deliveries more efficiently.
Review
Bus lanes are widely used in the UK and around the world to provide buses and their passengers with
priority access to urban centres. They are designed to allow buses to bypass queues that form on
approaches to traffic signals, roundabouts or other junctions. Typically, the end of a bus lane is set
back from the junction to maintain capacity at the junction. At traffic signals, guidelines suggest that
the setback distance (in metres) should be twice the signal green time (in seconds) (DETR et al,
1997). Problems can occur when bus lane use is not controlled and observations in Hanoi, Vietnam,
showed that high volumes of motorcycle traffic could significantly impede buses where the lanes were
not strictly enforced (Tranhuu et al, 2007).
For safety reasons, cyclists are normally permitted to use bus lanes in the UK and in some cases,
taxis are also allowed as long as they can be readily identified (IHT, 1997). Although the use of ‘bus
lanes’ by goods vehicles has been proposed (DETR et al, 1997), there have been few examples of
actual implementations. Such a shared-use lane was introduced in the city of Newcastle-upon-Tyne in
1992 with further ‘no-car’ lanes subsequently (DfT, 2006), but little evaluation appears to have been
undertaken to quantify the benefits. It was reported that only a small proportion of goods vehicles used
the initial lane in Newcastle due to a ‘lack of certainty among lorry drivers and local characteristics of
the road’ (DETR et al, 1997). A ‘local characteristic’ inhibiting use was that buses waiting to turn into a
bus station caused an obstruction to lorries (Binder, 2005). A small-scale trial of shared freight/bus
lanes, involving only 8 lorries, took place in Gothenburg, Sweden in 2004, and was reported to be
successful but was discontinued due to objections from bus operators (Binder, 2005).
Another idea for shared-use bus lanes is to allow general traffic to use the lane whenever there is no
bus using it. Clearly, this arrangement would only be feasible where bus frequency is low. This idea of
an ‘intermittent bus lane’ was introduced by Viegas (1996) and was piloted in a 6-month
demonstration in Lisbon, Portugal (Viegas et al, 2007). The authors reported that bus speeds
increased by 15%-25% with no noticeable impact on general traffic. A similar system has been
running successfully for trams in Melbourne, Australia since 2001, but with lower performance gains
compared to the Lisbon trial (Currie and Graham, 2008) and Sakamoto et al (2007) reported a similar
shared-use bus priority lane in Shizuoka, Japan. A related idea proposed within the SmartFreight
project is to permit freight unloading in bus lanes when there is no bus present or expected during the
planned period of unloading.
The use of dedicated lorry lanes has tended to be limited to inter-urban roads, typically motorways. A
simulation study of a motorway section in the Netherlands suggested that a lorry lane could improve
safety and efficiency through improved merging of traffic (Minderhoud and Hansen, 2001). In some
places, for example in the USA, “truck lane” refers to the fact that trucks are restricted from using other
lanes not that other vehicles are banned from the truck lane (NCTCOG, 2009) so there can be
confusion in the terminology.
Methodology and assumptions made
A simulation model of Winchester city centre and its main entry/exit routes was built using the traffic
micro-simulation package AIMSUN (Figure 1). The model included all main roads and five major car
parks but did not include little-used side streets, on-street parking and the park-and-ride car parks to
the south of the city. The road network was derived from a digital map to ensure accurate geometry.
The modelled vehicles were cars, vans, lorries and all bus services, including the park-and-ride bus
service.
Figure 1: AIMSUN model of Winchester
Data sources and model calibration
Traffic flow and origin-destination movements
Traffic flow data on all the arterial roads entering and exiting Winchester were obtained from a
previous study (MIRACLES, 2004). Although these data are now somewhat dated, having been
collected between 2001 and 2003, they were from the most complete and comprehensive survey
undertaken in the area and included counts of light goods vehicles (LGVs) and heavy goods vehicles
(HGVs). On average, these revealed that LGVs comprised 11% and HGVs, 1.9% of all vehicles
respectively. Detailed car park occupancy data were available from Hampshire County Council (HCC)
for one week, in mid-December 2008, giving an indication of car park usage, albeit at a particularly
busy time of year for shopping. A freight vehicle origin-destination (O-D) survey was undertaken by
HCC in Winchester during 2007, however it only included four of the eight entry/exit routes so was of
partial value only; no O-D data were available for general traffic. Given this lack of O-D
measurements, the O-D matrices required as input to AIMSUN were estimated using the entry/exit
flow data, car park data and estimated O-D movements. Adjustments were made to the matrices, in
the calibration process, to ensure that traffic flows at the critical junctions in the network matched the
real data.
Bus services
Bus service information was obtained from the Travel Line website (www.traveline.org.uk) and was
corroborated by on-site observations and bus-stop timetable information. All bus services and bus
stops were included in the model. A total of 18 different services and 54 buses per hour operate during
the modelled time period (0800-1000), that is, an average service frequency of 20 minutes. The park-
and-ride service is the most frequent (7.5 minutes) while several of the services are hourly. The
majority of buses (36 per hour) travel on St. George’s Street, the proposed site for the shared access
lane.
Traffic signals
Traffic signal timing data were obtained from the SCOOT traffic control system. These included
pedestrian phases (although pedestrians were not explicitly modelled here). Standalone pedestrian
crossings were not included in the model with the exception of one on St. George’s Street, which it
was felt was important to include due to its effect on traffic in the particular area of interest (Figure 2).
Performance measures
Performance is measured here in terms of percentage increases or decreases in travel times for the
different vehicle types (buses, lorries, vans and cars). Travel times were measured for sections of road
covering the shared lane on St. George’s Street and its approach roads, with the upstream measuring
points (different ones for buses and for general traffic as buses enter from a different road) being
upstream of queues caused by the shared lane.
Modelling scenarios
Time period and numbers of vehicles
The modelled time period was from 0800 to 1000 hours. This time period was chosen as it was
relatively busy for freight deliveries, buses and for general traffic (Cherrett et al, 2009). Separate O-D
matrices were used for each hour to reflect the greater numbers of vehicles in the first hour. The
numbers of vehicles entering the network are shown (Table 1).
Number of vehicles entering network in time period (% of total)
Vehicle type
0800-0900 0900-1000 0800-1000
Car 3569 (85.7%) 3004 (85.7%) 6573 (85.7)
Bus 54 (1.3%) 54 (1.5%) 108 (1.4%)
Lorry 89 (2.1%) 60 (1.7%) 149 (1.9%)
Van 453 (10.9%) 387 (11.0%) 840 (11.0%)
Total 4165 (100%) 3505 (100%) 7670 (100%)
Table 1: Modelled vehicle composition by time period
Location of shared bus/freight lane
The proposed bus/freight lane is located on St. George’s Street, a one-way street in Winchester city
centre (Figure 2). Issues that had to be considered were where to start and end the bus lane to allow
efficient entry and egress. During experimentation, with the bus lane starting at the junction with Upper
Brook Street, it was observed that turning traffic from Upper Brook Street had difficulty getting into the
correct lane. With the lane starting 20m further upstream (on Upper Brook Street) this problem was
alleviated. Buses enter St. George’s Street from a different approach road from general traffic, as
shown in the diagram. The end of the bus lane was set back from the junction with Jewry Street by
60m, in accordance with the guidelines (DETR et al, 1997). This allowed turning traffic (both left and
right) to reach the appropriate lane. The bus lane length on St. George’s Street was 160m. The traffic
signals were modelled as actuated control with a minimum green time on St. George’s Street of 30s.
This minimum green time was normally needed, as the traffic flow was relatively high, and was used to
ensure that the signals did not gap out too early, which was otherwise observed to happen under
some queuing circumstances (typically when a slow-moving lorry was in the right-hand lane). In
reality, the signals operate under SCOOT control, however, this could not be readily modelled.
Freight unloading point
The modelled unloading point was on the left hand side of St. George’s Street just at the end of the
bus lane (Figure 2). Unloading vehicles were modelled in AIMSUN as ‘section incidents’, which meant
that no vehicles were permitted to enter the area during the modelled unloading times. Unloading
times of 20, 40 and 60 minutes each hour were considered. The average dwell time for an unloading
delivery vehicle in Winchester was estimated to be 18 minutes, based on a survey of 83 Winchester
High Street store business managers, however, some deliveries were observed to take up to an hour
(Cherrett et al., 2009). Four unloading scenarios were considered:
1. No unloading
2. Unloading between 0820-0840 and 0920-0940 (i.e. 20 minutes in the hour)
3. Unloading between 0810-0850 and 0910-0950 (i.e. 40 minutes in the hour)
4. Unloading between 0800-1000 (i.e. 60 minutes in the hour or continuous unloading)
Figure 2: Modelled bus/freight lane on St. George’s Street
Lane use
Various categories of lane user were considered for the shared lane: (i) any vehicle (existing base
case situation); (ii) bus + LGV + HGV; (iii) bus + HGV; (iv) bus only. In practice, taxis are sometimes
allowed into ‘bus lanes’, however, this was not modelled here. Use of the lane by taxis may be
considered in further research.
Random variation
Ten modelling runs, each with a different random seed determining vehicle entry times onto the
network, were made for each scenario. The same set of 10 random seeds was used for each
scenario. The average results from each set of 10 runs are presented here. Ten runs for each
scenario was considered to be adequate to reduce the effects of random variation while keeping the
run time within acceptable bounds. The variation in the results between different random seeds was
very much dependent on the unloading scenario being modelled: when no unloading was assumed
the coefficient of variation (=standard deviation / mean) of the total travel time through the study area
was very low, ranging between 0.05 and 0.08 for the different lane uses; at the other extreme, with
continuous unloading being assumed, the coefficient of variation was rather high, ranging between
0.12 and 0.37, reflecting the fact that traffic conditions had become much more chaotic and volatile.
Reassignment
In the runs made here it was assumed there would be no reassignment of vehicles from one route to
another, that is, no vehicles would divert when faced with longer than usual queues. Vehicles were
assumed to follow their quickest route under free flow conditions. Due to the nature of Winchester city
centre’s one-way system there is very little route choice, so this assumption was considered to be
reasonable. The study did not consider any longer-term impacts which might occur in practice if a
significant change is made to the road network (e.g. mode choice, time of departure, route choice).
Results
Impact of shared lane with no unloading activity
The first set of results considers the impact of the lane, and its various uses, in the case where freight
vehicles do not stop to unload on St. George’s Street (Figure 3). Allowing freight vehicles to pass
through a central area with minimum disruption in terms of stops would reduce emissions and improve
overall traffic flow. Although this does not tend to happen at present, it can be considered to be a
scenario where unloading is banned and there is strict enforcement in place. Figure 3 shows the
Upper Brook St.
freight unloading
pedestrian crossing
Jewry St.
bus approach road
start of bus lane
end of bus lane
St. George’s St.
traffic signals
traffic signals
increase (or decrease) in total travel time suffered by different vehicle types for the three different lane
uses proposed, compared to the existing situation where both lanes are available for all vehicles.
Travel times were measured here on St George’s Street and its approach roads with t-tests being
undertaken to determine whether there were any significant differences in mean journey times
between each of the scenarios and the base case. Significant differences at the 95% level are
indicated in bold text with non-significant results in plain text. From Figure 3 it can be seen that:
(i) The dedicated bus lane provided a modest bus time saving through the area of 2% but
increased travel times for traffic as a whole by 12% (t=3.41, P=0.0042). It was observed that
one reason why buses did not benefit more was that at the end of the bus lane, right-turning
buses (20 of the 36 buses per hour turn right) suffered a small amount of delay in gaining
access to the right-hand lane.
(ii) Allowing HGVs (lorries), (approximately 40 of them per hour), to use the shared lane reduced
delays, compared to the bus only option, although there was an overall increase in travel times
over all vehicles of 8% (t=2.62, P=0.018). Lorries gained a marginal reduction in travel time
(not statistically significant) through the area but buses lost a small amount of the already
small benefit from the lane.
(iii) Allowing LGVs (vans) (approximately 158 of them per hour) to use the shared lane reduced
the overall delays again, compared to the two options above, with an increased travel time of
4% (not statistically significant). Vans achieved a marginal benefit (not statistically significant)
and the bus benefit had now disappeared.
(iv) Since cars dominate the traffic flow (approximately 900 of them per hour on St. George’s
Street), the overall results reflect those for cars. The results show that all of the shared lane
options would have an overall negative impact, with the benefits shown for buses, lorries and
vans being somewhat marginal and not statistically significant.
-6%
-4%
-2%
0%
2%
4%
6%
8%
10%
12%
14%
16%
bus+LGV+HGV bus + HGV bus only
vehicles pe rm itte d in shared lane
increase in journey time
bus lorry van car all vehicles
Figure 3: Effect of shared lane use on different vehicle types (no unloading)
Impact of unloading activity
The impacts of the different unloading scenarios were examined, initially for the existing situation of
both lanes on St. George’s Street being available for use by any vehicle (Figure 4). These results
show the increase in travel time through St. George’s Street and its approach roads compared to the
base case (scenario 1) of there being no unloading on street. From the results it can be seen that:
(i) Travel times increase substantially as the time spent unloading on street increases - for
example, with the main unloading point being used for 20 minutes in the hour the overall delay
for all vehicles was about 6% ,increasing to 19% (t=4.46, P=0.0008) for 40-minute unloading
in the hour and to 75% (t=3.68, P=0.005) for continuous unloading throughout the modelled
time period.
(ii) Buses tended to be less affected by the blockages than other vehicles since they did not suffer
the queues that were generated in Upper Brook Street (Figure 2), however, they did suffer
queues in St. George’s Street itself.
0%
10%
20%
30%
40%
50%
60%
70%
80%
20 40 60
unloading activity (minutes in the hour)
increase in travel time
bus lorry van car all vehicles
Figure 4: Impact of unloading activity
Impact of shared lane with loading activity
The third set of results considers the impact of the shared lane when there is unloading activity in St.
George’s Street (Figure 5). Again the shared lane scenarios are measured against the existing
situation of there being no shared lane. The unloading scenario considered here was 40 minute
unloading in the hour, that is, from 0810 to 0850 and from 0910 to 0950. The results suggested that:
(i) The dedicated bus lane now seems to benefit buses more than previously (Figure 3) with a
saving of 9% (t=2.93, P=0.011). Also it has a less detrimental effect on other traffic with the
results for the other vehicle types not being statistically significant.
(ii) The shared bus and HGV lane sees a 5% benefit for buses, a 3% benefit for lorries and a 6%
disbenefit for cars, none of which were statistically significant.
(iii) Allowing vans into the shared lane sees small benefits for vans, lorries and buses but again a
6% disbenefit for cars (none statistically significant).
-10%
-8%
-6%
-4%
-2%
0%
2%
4%
6%
8%
bus+LGV+HGV bus + HGV bus only
vehicles pe rm itte d in share d lane
increase in journey time
bus lorry van car all vehicles
Figure 5: Effect of shared lane use on different vehicle types (40 minute unloading)
The final set of results examines how the introduction of 40-minute unloading activity in the hour would
affect the different lane use scenarios (Figure 6). Here, the results were compared with there being no
unloading activity, so, for example, the bus-only result compares “bus-only, with 40-minute unloading”
against “bus-only, with no unloading”. A general trend was observed that the fewer the vehicles
allowed to use the shared lane, the less the impact of the 40-minute unloading time. Since the
unloading point is right at the end of the shared lane (Figure 2) this may be because there are fewer
vehicles in the shared lane that need to pull out into the right-hand lane in order to overtake the
obstruction. This result doesn’t show that a dedicated bus lane, for example, is beneficial, however, it
shows that, once in place, the dedicated bus lane is least affected by any unloading that may take
place.
0%
5%
10%
15%
20%
25%
any vehicle bus + HGV + LGV bus + HGV bus only
vehicles permitted to use shared lane
increase in travel time
bus lorry van car all vehicles
Figure 6: Effect of “40-minute in the hour unloading” by lane use scenario
Conclusions
This paper has examined the combined effects of shared lane use and of freight unloading activity.
The results suggested that all of the shared lane options would have an overall negative impact due to
the negative effects on cars (which comprise approximately 85% of all traffic). In addition, the benefits
shown for buses, lorries and vans were rather small and were often not statistically significant. This
may be because the shared lane covered only a short section of road (160m) and vehicles wanting to
turn right into Jewry Street experienced some delay in gaining access to the right-hand lane. A
potential solution to this latter problem may be to use a pre-signal to provide buses (or other shared
lane users) with priority access to the downstream junction (Wu and Hounsell, 1998); this remains as
further research.
Increasing unloading activity was seen to result in escalating delays to all vehicles: the road was
observed to be able to cope with 20-minute unloading in the hour without too much disruption but the
delays experienced as a result of 40-minute unloading periods were around 19% for the existing lane
use (both lanes unrestricted), which might be considered unacceptable. Freight unloading had the
least impact when the left-hand lane was restricted to buses only. Also, the bus lane seemed to be
more beneficial to buses when there was a lorry using the unloading point. This was likely due to the
fact that there were longer queues in the right-hand lane for the bus to jump ahead of.
An issue to consider further is the behaviour of drivers in the right-hand lane when allowing or not
allowing priority vehicles in the left-hand lane to cross to the right-hand lane, either to overtake an
unloading lorry, or, at the end of the lane, when they want to turn right at the downstream junction. The
AIMSUN model makes certain assumptions about this kind of behaviour and has some adjustable
parameters affecting it. In this study, AIMSUN default parameters were used and, from observations,
the drivers in the right-hand lane appeared to be quite considerate in allowing priority traffic to cut
across.
References
• Binder, S. (2005). The potential for integrating light commercial freight into the Avenue bus
lane. B.Sc. student thesis, School of Civil Engineering and the Environment, University of
Southampton.
• Cherrett, T., McLeod, F.N., Maynard, S., Hickford, A., Allen, J. and Browne, M. (2009)
Understanding retail supply chains to enable ‘greener’ logistics. Logistics Research Network
conference, Cardiff, 2009.
• Currie, G and Lai, H. (2008). Intermittent and Dynamic Transit Lanes: The Melbourne
Experience. Transportation Research Board 87th Annual Meeting, Washington, USA.
• Department for Transport (2006). Bus priority: The way ahead. Available at:
http://www.dft.gov.uk/pgr/regional/buses/bpf/busprioritythewayahead12/ (Accessed 8 June
2009).
• Department of the Environment, Transport and the Regions (DETR), The Scottish Office and
The Welsh Office (1997). Keeping Buses Moving. Local Transport Note 1/97. The Stationery
Office.
• Institution of Highways and Transportation (IHT) (1997). Transport in the urban environment.
• Minderhoud, M. and Hansen, I. (2001). A microscopic simulation study after freight lanes near
merging areas. 8
th
World Congress on Intelligent Transport Systems, Sydney, Australia.
• MIRACLES (2004). Project website: http://www.winchestermiracles.org/ (Accessed 8 June
2009)
• North Central Texas Council of Governments (NCTCOG) (2009). Truck lane pilot study.
Available at: http://www.nctcog.org/trans/goods/trucklane/ (Accessed 8 June 2009).
• Sakamoto, K., Abhayantha, C. and Kubota, H. (2007). Effectiveness of bus priority lane as
countermeasure for congestion. Transportation Research Record: Journal of the
Transportation Research Board, 2034, 103-111.
• Tranhuu, M., Montgomery, F. and Timms, P. (2007). Modeling Bus Lane Priorities in a
Motorcycle Environment Using SATURN. Transportation Research Record: Journal of the
Transportation Research Board, 2038, 167-174.
• Viegas, J.M., Roque, R., Lu, B. and Vieira, J. (2007). Intermittent Bus Lane System:
Demonstration in Lisbon, Portugal. Transportation Research Board 86th Annual Meeting,
Washington, USA.
• Viegas, J.M. (1996). Turn of the century, survival of the compact city, revival of public
transport, bottlenecks. Transportation and the Port Industry (H. Meersman, ed). Antwerp,
Belgium.
• Wu, J. and Hounsell, N.B. (1998). Transportation Research A, 32(8), 563-583.