You are on page 1of 9

Transportation Letters

The International Journal of Transportation Research

ISSN: 1942-7867 (Print) 1942-7875 (Online) Journal homepage: http://www.tandfonline.com/loi/ytrl20

Modelling the major stream delay due to U-turns

Malaya Mohanty & Partha Pratim Dey

To cite this article: Malaya Mohanty & Partha Pratim Dey (2017): Modelling the major stream
delay due to U-turns, Transportation Letters, DOI: 10.1080/19427867.2017.1401701

To link to this article: http://dx.doi.org/10.1080/19427867.2017.1401701

Published online: 15 Nov 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ytrl20

Download by: [UNIVERSITY OF ADELAIDE LIBRARIES] Date: 16 November 2017, At: 00:30
Transportation Letters, 2017
https://doi.org/10.1080/19427867.2017.1401701

Modelling the major stream delay due to U-turns


Malaya Mohanty and Partha Pratim Dey
School of Infrastructure, Indian Institute of Technology, Bhubaneswar, India

ABSTRACT KEYWORDS
This study assesses the delay faced by approaching through vehicles at uncontrolled median openings due Approach delay; median
to U-turning movement. Under limited priority situation, the U-turning vehicles accept smaller gaps which opening; mixed traffic;
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

compel the through vehicles to reduce their speeds and experience delays. First, the slowdown section U-turns; regression; dummy
is identified and the delay faced by through vehicles from the starting of slowdown section to start of variable
median opening is estimated. This is termed as approach delay. The approaching vehicles shift laterally as
they approach toward the median opening. The concept of zone has been introduced to study this lateral
movement. This study shows that the vehicles moving through the zone adjacent to curb face more delay
as compared to the vehicles moving through the zone adjacent to median. Finally, a regression model is
developed for estimating approach delay faced by any vehicle category traveling through a particular zone at
a given U-turning traffic volume.

Introduction
increases, resulting in increased waiting times for the U-turning driv-
The rapid increase in vehicular traffic has imposed a burden on ers. Meanwhile, if the duration of the waiting time at the median
transportation infrastructure (Mohapatra, Bhuyan, and Rao 2012). opening is greater than a particular threshold, the U-turning drivers
To meet the increasing demand for vehicular traffic, many multilane will start to accept smaller gaps, forcing some of the drivers on the
roads have been constructed in the recent years along with widening through traffic stream to reduce their speeds and experience some
of two-lane roads. Usually, raised median is provided on multilane delays. Furthermore, due to the discourteous behavior of the drivers,
roads to segregate the opposing traffic movements. Openings are the rule of priority is habitually violated, and the U-turning vehicles
provided in the raised median so that vehicles can take U-turns forcefully enter the median opening even in smaller gaps forcing
and reverse their direction of movement by merging with through the approaching through vehicles to slow down. In these situations,
traffic stream in the opposing lane (Mohapatra, Dey, and Chandra the priority of approaching stream becomes shared, which is usu-
2016a). The U-turning movement at a median opening is highly ally known as limited priority situation (Ma et al. 2013). Mohapatra,
complex and risky when compared with turning movements at Dey, and Chandra (2016b) observed that 60–80% of approaching
intersections (Mohapatra and Dey 2015). It is because of the high through vehicles are forced to slow down to enable the U-turning
speed and heavy traffic volume of opposite flow and also the turn- vehicles to merge. A similar observation is also reported for uncon-
ing vehicle has to make an l80-degree movement and merge with trolled intersections in India (Ashalatha and Chandra 2011). This is
the approaching through traffic stream in which it is seeking an not necessarily due to high traffic volume, but is due to the impatient
acceptable gap (Aldian and Taylor 2001). Assessing the process of and discourteous behavior of low priority movement (U-turning vehi-
how approaching through vehicles absorb U-turning vehicles is very cles). Due to these peculiar characteristics of road users and vehicles,
much interesting. The U-turning movement has lower priority than the approaching through vehicles experience delay due to limiting or
the through movement because the U-turning movement leads to reversal of priority situations.
merging with the approaching through traffic movement. Moreover, Over last few years, many researchers have tried to discuss the
one essential assumption is that the approaching through traffic characteristic of traffic operations in limited priority systems. Wang et
exercises absolute priority, which implies that none of the approach- al. (2013) developed a model to study the speed and travel time based
ing through vehicles experience any delay in passing through the on log-normal distributions. Minh et al. (2010) estimated delay under
median opening. heterogeneous traffic conditions. Bonneson and Fitts (1999) assessed
The situation is grimmer in India as most of the unsignalized inter- major stream delay to through vehicles at two-way stop controlled
sections are uncontrolled, i.e. they do not have a stop or yield sign, no intersections. Caliendo (2014) developed a delay time model for
traffic police, and no traffic signal to control the vehicular movements. unsignalized intersections. Besides these studies, some developments
Upon arrival at the median openings, the U-turning vehicle waits for a on critical gap estimation and minor stream delay at uncontrolled
suitable gap in the approaching through traffic to initiate the merging intersections have been made by few researchers (Troutbeck and Kako
maneuver. The number of gaps accepted by U-turning drivers during 1999; Bunker and Troutbeck 2003; Ma et al. 2013). However, research
a unit of time, however, will decrease as the through traffic streamflow on delay estimates for through traffic stream at uncontrolled median

CONTACT  Malaya Mohanty  mm14@iitbbs.ac.in


© 2017 Informa UK Limited, trading as Taylor & Francis Group
2   M. MOHANTY AND P. P. DEY

openings has not been studied till now. The present work attempts t2 = time when vehicle is at point 2 (start of the median opening)
to estimate the delay faced by the approaching through traffic under t2′ = time when vehicle is at point 2 if traveled with steady speed (Vs)
limited priority condition. L1 = location of point 1
L2 = location of point 2
Vs = steady speed of the approaching through vehicle
Delay measures
Several types of delay are identified as the approaching through vehi-
Median delay
cles move forward toward uncontrolled median openings such as (a)
approach delay and (b) median delay. These delay concepts are illus- Median delay is the delay experienced by the approaching through
trated graphically in Figure 1. vehicles as they clear the median opening area. It is measured as the
additional time spent due to deceleration within the median open-
ing, the time spent stopped (if any) within the median opening, and
Approach delay the acceleration delay up to the start of the median opening. For the
Approach delay (Da) is the delay experienced by the approaching purpose of obtaining the median delay, the time required to cover
through vehicles while decelerating up to the beginning of the median the distance from point 2 to point 3 is noted, and median delay (Dm)
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

opening. For the purpose of obtaining the approach delay, the location is computed as:
of point 1 (starting of deceleration) needs to be identified. Then, the ( )
time required to cover the distance from point 1 to the starting of L2 − L 3
(2)
( )
Dm = t3 − t2 −
median opening (point 2) is noted, and delay is computed as: vs
where
( )
L1 − L 2
(1) t2 = time when vehicle is at point 2 (start of the median opening)
( ) ( ) ( )
Da = t2 − t2� = t2 − t1 − t2� − t1 = t2 − t1 −
vs
t3 = time when vehicle is at point 3 (end of median opening)
where L2 − L3 = length of the median opening
t1 = time when vehicle is at point 1 and deceleration begins Vs = steady speed of the approaching through vehicle

L3 F
End of median opening Acceleration starts
I E
Center of the median opening
D
Start of median opening G C Deceleration stops
L2
Direction of traffic

B
Direction of traffic

L1
Distance

Deceleration starts

A
t t1 t2 t3 Time

Figure 1. Time–distance diagram of approaching through vehicles.


TRANSPORTATION LETTERS   3

Data collection and extraction For the proper identification of starting of slowdown section, a
length of 80 meters divided into eight equal divisions of 10 meters
In order to study the delay caused by U-turns to approaching through each, in the upstream of the starting of median opening was consid-
traffic, seven different median openings on six-lane divided urban ered as shown in Figure 2. The markings were done on the pavement
roads were selected. The test sections were selected so that the traf- surface by using temporary paints. The paintings were done around 3:
fic flow at these sites is not influenced by horizontal curvature, the 00 am when the traffic volume is very less. In this microscopic analysis,
presence of downstream or upstream intersection, bus stop, parked the movement of each approaching through vehicles is considered
vehicles, pedestrian movements, or any kind of side friction. All the separately. The real time as each vehicle crosses all the marked eight
road sections had road width in the range of 9.4–9.8 m in each direc- divisions (at an interval of 10 m) is noted, and the subsequent time
tion of travel with raised curb on either side of the road. The details required to cover the successive divisions is obtained. Finally, the
of road geometry of all sections are given in Table 1. speeds maintained at every 10 m in the downstream direction are
At each test section, data were collected by video recording tech- estimated and compared with the steady speed maintained by the
nique on a typical weekday. The study conducted in the field is mainly subject vehicle. The steady speed is measured by using a trap length
to assess the followings: of 25 m at a distance of 140 meters in the upstream of the starting
(1) Identification of starting of possible slowdown sections of median opening. The section is so chosen so that it is free from
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

from where the approaching through vehicles start decel- the effect of U-turns and the presence of any upstream intersection.
erating and Similar to start of slowdown section, the end of the section was iden-
(2) Estimation of delay. tified by studying the speed of the approaching through vehicles in the
downstream direction. Primary field observations revealed that the vehi-
Due to the limited priority situation, the approaching through cles started to speed up within the median opening area, and accordingly,
vehicles start decelerating in the upstream of the median open- the length of the median opening was divided into five equal divisions.
ing. Therefore, a possible slowdown section exists for the vehicles In conducting the field surveys, four cameras were used to record the
approaching toward the median opening. Ma et al. (2013) considered traffic data as shown in Figure 2. The video cameras were placed simul-
40 meters before the starting of intersection to be the possible slow- taneously at suitable vantage points to record unobstructed view of all
down section. However, the situation is totally different at uncon- traffic movements, and data were recorded for about 10–12 h, including
trolled median openings and that too under mixed traffic condition the peak and off-peak periods. Camera C4 recorded the time required
in India where lane discipline and rules of priority are hardly followed. for the vehicles to cover a distance of 25 meters (140–165 meters).
Moreover, this starting of slowdown section varies with category of Camera C3 and C2 were used to cover the distance of 80 meters from the
vehicles and prevailing traffic volume. Though, it is observed that the start of median opening. Camera C1 covered the median opening area.
variation is not significantly influenced by the category of vehicles. On The recorded films were played on a large screen TV monitor with
the other hand, field observations revealed that the start of slowing the help of software (Avidemux version 2.6) which plays the videos at
down, identified by observing the continuous brake light, is also not frame per second at which it was recorded. For the present study, the
uniform at all traffic volume levels. Preliminarily, this was found that camera used had a precision level of 25 frames per second (fps). All
the vehicles stated to decelerate continuously at a distance of around vehicles in the traffic stream were divided into six predominant cat-
30–50 meters in the upstream of the start of median opening, and egories as motorized two-wheelers (2 W), motorized three-wheelers
for this purpose, the speed profile of the each and every approaching (3 W), passenger cars, sports utility vehicles (SUV), light commercial
through vehicles moving toward the median opening was studied. vehicles (LCV), and heavy vehicles (HV). From the video, the follow-
ing information was noted:
• Classified traffic volume count of U-turning traffic,
Table 1. Geometric details of different test sections. • Time required by individual vehicles to cover each and every
Section No. 1 2 3 4 5 6 7 division marked temporarily on the road surface, and
Width of median opening (m) 20 15.7 19.8 14.8 20 20.3 20.1
• Lane wise position of each and every approaching through
Width of raised median (m) 1.3 1.0 1.2 1.2 1.3 1.2 1.3 vehicle when they enter the possible slowdown section and
Width of carriageway (m) 9.6 9.6 9.8 9.5 9.4 9.5 9.4 median opening.
End of median opening

Start of median opening

10 40 50 70 80 140 165
5 equal parts 20 30 60

C1 C2 C3 C4

Figure 2. Camera setup for field survey.


4   M. MOHANTY AND P. P. DEY

Analysis of data to −0.2x, −0.2x to −0.4x, −0.4x to −0.6x, −0.6x to −0.8x, and −0.8x to
−x) were compared to the steady speed of the approaching through
The microscopic analysis of delay was studied with a large amount of vehicles. The term ‘x’ refers to the length of median opening (distance
data for each category of vehicles at various traffic volume levels. The from start of median opening to the end of median opening). The
analysis is divided into two major parts as follows: speed differences are quite small when the vehicles are away from
• Identification of possible slowdown sections and the median opening, and it increases as the vehicles approach toward
• Analysis of approach delay. the median opening area. Perco, Marchionna, and Falconetti (2012)
reported that speed differences lower than 2.2 km/h in a speed profile
are considered practically constant. The speed difference is found to
be more than 2.2 km/h at a distance of 40–50 m when the hourly
Identification of possible slowdown section
U-turning traffic volume is less than 900 vph. This indicates that the
In the exercise of identifying the slowdown sections, the speed profiles vehicles are likely to slow down their speed at a distance of 40–50 m.
of around 1250 approaching through vehicles at varying U-turning Similarly, it is observed that when U-turning traffic volume is more
traffic volume levels (300 vph to more than 1500 vph) were studied. than 900 vph, the possible slowdown section starts from 30 to 40 m.
The average speeds of the approaching through vehicles at every vol- Therefore, it is clear that the slowdown section varies with U-turning
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

ume levels are presented in Table 2 and shown in Figure 3. The steady traffic volume, and these are used to estimate the approach delay at
speed of the approaching through vehicles was measured at a dis- different volume levels.
tance of 140–165 m in the upstream of median opening. Subsequently, From Figure 3 and Table 2, it can be seen that up to volume level
to check if speed performance varies in the proximity of U-turns, 900 vph, the vehicles are found to reduce the speed at more or less uni-
the speed values at all the definite locations (140–165 m; 70–80 m; form rate. The maximum reduction in speed (compared with respect
60–70 m; 50–60 m; 40–50 m; 30–40 m; 20–30 m;10–20 m; 0–10 m; 0 to steady speed) is observed at the center of the median opening

Table 2. Speed of approaching through vehicles (km/h).

U-turning traffic volume (vph)


Distance from start of median opening (m) 300–600 600–900 900–1200 1200–1500 More than 1500
Within the median opening (x) −0.8x to −x 45.770 34.681 24.614 21.901 22.997
−0.6x to −0.8x 43.757 33.684 23.096 19.012 18.612
−0.4x to −0.6x 41.864 32.434 21.874 17.007 15.654
−0.2x to −0.4x 40.630 31.388 20.254 17.049 16.977
0 to −0.2x 41.916 32.534 23.243 21.392 20.970
Upstream from the start of median 0–10 42.210 34.433 28.867 28.100 26.900
opening 10–20 43.525 37.233 32.967 31.150 29.400
20–30 45.381 40.867 39.333 32.650 30.400
30–40 46.797 43.767 41.967 39.600 32.300
40–50 48.212 46.400 44.133 43.550 38.400
50–60 49.100 47.245 44.567 44.150 39.000
60–70 49.467 47.500 44.767 44.850 39.500
70–80 50.067 47.767 45.000 45.150 40.050
140–165 50.467 49.200 45.600 45.650 40.500
Note: −ve values indicate downstream from the starting of median opening.

55.0
50.0
45.0
40.0
Speed, kmph

35.0
450 vph
30.0
750 vph
25.0
1050 vph
1350 vph 20.0

1650 vph 15.0


10.0
160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 -10 -20 -30
Distance, m
Direction of traffic

Direction of traffic

Figure 3. Identification of slowdown section.


TRANSPORTATION LETTERS   5

and in the order of 30%. This is attributed to mainly two reasons: (i) Effect of U-turning traffic on approach delay
due to the less number of U-turning vehicles and (ii) at low traffic The effect of U-turning traffic volume on delay to individual category
volume, the approaching through vehicles are found to shift laterally of vehicles was studied. However, HV has not been considered in
to avoid conflict with the U-turning vehicle. At volume level, 900 this study due to its very less proportion in the traffic stream. The
vph or more, the vehicles reduced their speed in two stages. First, delay statistics for each category of approaching through vehicles were
they reduce the speed slightly as they enter the slowdown section, estimated from the collected data at different traffic volume and are
and finally, they reduce the speed very fast as they are very close to presented in Table 3. Table 3 clearly shows that at any particular traf-
median opening area. At higher U-turning traffic volume, the flexi- fic volume level, the difference in delay experienced by individual
bility of the approaching through vehicles to shift laterally gradually category of vehicle is negligible. Therefore, a t-test was conducted at
disappears, and thus, these vehicles are forced to reduce the speed to 5% level of significance, and it revealed that there is no significant
avoid collision with the U-turning vehicles. Further, it is also observed difference among the delay values. It can also be observed that the
that this deceleration continues up to the center of median opening. delay faced by a specific category of vehicle increases with increase
This marks the end of slowdown section. After the vehicles cross the in U-turning traffic volume.
center of median opening, they start accelerating until they reach
desired speed. Lane wise distribution of approaching through traffic
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

Typical heterogeneous traffic is characterized by the presence of mul-


tiple vehicle types and non-lane-based movement (Mathew, Munigety,
Analysis of approach delay
and Bajpai 2013). Because of the off-centered positions of the vehicles,
In the previous section, it has been shown that the length of slowdown it is difficult to study complete lane-changing behavior in heterogene-
section changes with traffic volume. Moreover, at any particular traffic ous traffic flow. Unlike discrete lane changes in the case of lane-based
volume, the delay experienced by all the vehicles is also different. traffic, in the case of heterogeneous traffic, some vehicles are found to
The approach delay was assessed microscopically by noting down shift laterally not making a complete lane change (Mathew, Munigety,
the following details: and Bajpai 2013). The pattern of these lane-changing behaviors of
approaching through vehicles at different traffic volumes has been
• U-turning traffic volume, studied by noting their lane wise position at start of slowdown section
• Lane wise position of the subject vehicle while just entering the (entry lane) and start of median opening (exit lane). The vehicles
slowdown section, hereafter this is referred as position on entry involved in lateral shifting by half lane are termed as moderate lane
lane, and change, whereas in case the lateral shifting is one lane or more, it is
• Lane wise position of the subject vehicle just before entering termed as significant lane changing. The lane wise distribution of the
the median opening area, henceforth termed as exit lane. approaching through vehicles is tabulated in Table 4.
The above table clearly depicts that only the lane wise study is not
good enough, as many vehicles shared two lanes while traveling. The
Table 3. Effect of U-turning traffic volume on approach delay.
vehicles may enter the slowdown section through any of the combi-
More nations namely median lane (henceforth lane 1) only, sharing lane 1
U-turning traffic 900– 1200– than and central lane (henceforth lane 2), lane 2 only, sharing lane 2 and
volume (vph) 300–600 600–900 1200 1500 1500
curb lane (henceforth lane 3), and lane 3 only. Likewise, the vehicles
Approach 2 W 0.246 0.307 0.402 0.542 0.707
are found to enter at the beginning of median opening area also. From
delay (s) 3 W 0.271 0.34 0.421 0.571 0.757
CAR 0.238 0.289 0.489 0.577 0.771 the above table, it is also clear that around 40% of vehicles are not
SUV 0.245 0.302 0.48 0.594 0.770 involved in lane changing, whereas around 60% are found to change
LCV 0.252 0.295 0.476 0.587 0.769 their lanes as they approach toward the median opening. The lane
Average delay (s) 0.249 0.308 0.426 0.557 0.730 wise (entry lane and exit lane) distribution of approaching through

Table 4. Lane wise distribution of approaching through vehicles.

Lane-changing Percentage of total


Entry lane Exit lane maneuver vehicles Moderate lane change Significant lane change
1 1 No lane change 4.02 53.67 25.52
1&2 Moderate lane change 10.37
2 Significant lane change 4.93
2&3 Not found 0.00
3 Not found 0.00
1&2 1 Not found 0.00 69.65 0.00
1&2 No lane change 4.80
2 Moderate lane change 11.02
2&3 Not Found 0.00
3 Not found 0.00
2 1 Not found 0.00 26.49 25.48
1&2 Not found 0.00
2 No lane change 24.56
2&3 Moderate lane change 13.55
3 Significant lane change 13.03
2&3 1 Not found 0.00 74.65 0.00
1&2 Not found 0.00
2 Not found 0.00
2&3 No lane change 2.20
3 Moderate lane change 6.48
3 1 Not found 0.00 0.00 0.00
6   M. MOHANTY AND P. P. DEY

Number of vehicles 500 0.9


0.8
400 0.7
0.6
300

Delay, s
0.5
200 0.4
0.3
100 0.2
0 0.1
1 1.5 2 2.5 3 0
0 250 500 750 1000 1250 1500 1750
Lane U-turning traffic volume, vph
Entry Exit
Zone 1 Zone 2 Zone 3
(a)
Figure 5. Effect of different zones on approach delay.
500
Numer of vehicles
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

400
Table 7. Statistical parameters of regression analysis.
300
Approach delay (s) Coefficient P > |t| ANOVA
200
U-turning traffic volume 0.0003 0.000 ANOVA (6,1462) = 791.56,
100 Zone −0.009 0.037 p = 0.000
two-wheeler 0.03 0.031
0 three-wheeler 0.055 0.015
1 1.5 2 2.5 3
Car 0.021 0.035
Lane SUV 0.018 0.046
Entry Exit
(b)
the approaching vehicles are traveling. However, a slight variation
Figure 4. Lane wise vehicle count at (a) low traffic and (b) high traffic volumes. is observed at higher volume levels (more than 1000 vph). The data
presented also show that as the vehicles are away from the median lane
(lane 1), there is an increase in the delay at higher U-turning traffic
Table 5. Division of zones according to entry and exit of vehicles. volumes. This is attributed to the fact that the approaching through
Zone Entry lane Exit lane vehicles exhibit a general trend of lateral shifting toward the curb lane
1 1, 1 & 2 1, 1 & 2, 2 (lane 3) as they approach close to median opening. As a result, the
2 2, 2 & 3 2, 2 & 3, 3 distribution of approaching traffic in the lane 1 becomes least, and
3 3 3 it is maximum in lane 2. Hence, apart from the U-turning vehicles,
there is also more interaction among through vehicles traveling in
zone 2. Hence, the vehicles in zone 2 face a greater amount of delay
Table 6. Zone wise approach delay at different traffic volumes.
than vehicles moving in zone 1. Interestingly, though in zone 3, the
  Approach delay (s) traffic volume is less than zone 2, the vehicles traveling through zone
U-turning traffic volume (vph) Zone 1 Zone 2 Zone 3 3 experience maximum delay. It is obvious due to the fact that zone
300–600 0.236 0.26 0.262
3 is mostly used by the vehicles which want to avoid the main traffic
600–900 0.299 0.304 0.31 flow and also ready to compromise their speed to be safe on the road.
900–1200 0.37 0.381 0.382
1200–1500 0.577 0.626 0.752
>1500 0.697 0.771 0.856 Approach delay modeling
Average 0.4358 0.4684 0.5124
In the previous sections, it is observed that the approach delay is
mainly influenced by the U-turning traffic volume. Moreover, it is
vehicles is shown in Figure 4. From the figure, it is clear that at both observed that the zone through which the vehicle is approaching
the volume (low and high), there is a tendency of shifting laterally toward median opening also influences the approach delay. Also,
toward the curb as the vehicles move toward the median opening. the experienced delay is not same for all the categories of vehicles.
Table 4 clearly shows that the vehicles entering the slowdown Therefore, in the microscopic study, it is required to develop a gener-
sections at lane 1 usually shift lanes maximum up to lane 2 while alized model to estimate the delay considering upon all the mentioned
entering the median opening. Similarly, vehicles entering the slow- influencing factors.
down section in lane 1 and lane 2 also mostly shift up to lane 2. The A multiple linear regression was carried out to estimate the
abovementioned vehicular movements (entry through lane 1 and lane approach delay experienced by any particular category of approach-
1 and lane 2) were considered in one group and termed as movement ing through vehicle. The independent variables used for the analysis
through zone 1. Similarly, zone 2 and zone 3 are identified, and details are U-turning traffic volume, zone in which the approaching through
are given in Table 5. vehicle is traveling, and the category of approaching through vehicle.
The delay faced by the approaching through vehicles is not same Prior to regression modeling, all the independent variables (U-turning
as they move forward toward the median opening. Rather, it varies volume, zone, and category of vehicles) were checked for multicolline-
with zone in which they are traveling toward the median opening. The arity, and no significant correlation among these three variables at 5%
average delay experienced by the vehicles as they travel via different significance level was observed. The category of vehicles is an indicator
zones is summarized in Table 6 and shown in Figure 5. variable. These variables usually indicate the presence or absence of
Figure 5 clearly shows that up to U-turning traffic volume of 1000 a quality or an attribute (Gujarati 2003). One way to ‘quantify’ such
vph, the average delay does not vary with the zones through which attributes is by constructing artificial variables that take on values of
TRANSPORTATION LETTERS   7

Table 8. Model validation. different median opening of six-lane divided road. The data of this
Average approach delay (s)
validation section have not been used for analyzing the approach
Category of Difference in delay. A random sample of around 60 data points for each category of
vehicle Proposed model Field observation percentage vehicles was used for the validation purpose. Table 8 shows the com-
2 W 0.411 0.391 4.86 parison of average delay of all the categories of vehicles estimated from
3 W 0.584 0.556 4.79
Car 0.451 0.442 1.99
the proposed model and actual delay as obtained from the field data.
SUV 0.511 0.498 2.54 From Table 8, it can be observed that the values obtained from
LCV 0.493 0.481 2.43 proposed model are in close agreement with the observed field values.
It can be marked that the minimum error was found in case of car,
i.e. 1.99%. The maximum difference in the values was found in 2 W
1 or 0, 1 indicating the presence (or possession) of that attribute and which is around 5%. Therefore, it can be concluded that the proposed
0 indicating the absence of that attribute. Such variables that assume model is good enough to predict the approach delay.
such 0 and 1 values are called dummy variables (Gujarati 2003). For
each qualitative regressor, the number of dummy variables introduced Conclusions
must be one less than the categories of that variable (Gujarati 2003).
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

Therefore, in the present analysis, four dummy variables have been The priority of the approaching through vehicles at uncontrolled
used to indicate the effect of five categories of vehicle on approach median openings is partly shared as the U-turning traffic flow
delay. The dummy variables considered in the present analysis are 2 W, increases, and some approaching through vehicles will experience
3 W, car, and SUV. If approach delay faced by 2 W is to be estimated some delay as they adjust their positions to avoid road crashes.
from the model, 2 W takes the value of 1, and rest dummy variables In the present study, under mixed traffic conditions, the average
take the value of 0. Likewise, the delay experienced by 3 W, car, and approach delay faced by the approaching through vehicles due to
SUV can be estimated. If all the four dummy variables take the value U-turning traffic at uncontrolled median opening was assessed. The
of 0, it is obvious that the model will give delay for the fifth category location on the upstream of median opening from where the vehicle
of vehicle, i.e. for LCV. The equation is as follows. started decelerating was first identified to know the start of slowdown
section. The stretch of road from the start of the slowdown section up
Da = (0.0003 ∗ U) − (0.009 ∗ Z) + (0.03 ∗ 2W) to the start of median opening has been used to estimate approach
(3) delay. It is observed that with the rise in U-turning traffic volume,
+ (0.055 ∗ 3W) + (0.021 ∗ Car) + (0.018 ∗ SUV)
the approach delay of all the five categories of vehicles (2 W, 3 W, car,
(Adjusted R2 = 0.77) SUV, and LCV) also changes, and the relationship is exponentially
where Da = the average approach delay faced by ith category of increasing. It is also quite common to observe that no proper lane
approaching through vehicle, s; discipline is maintained by the vehicles, and also vehicles are found to
U = U-turning traffic volume, vph; and shift laterally but not doing a complete lane change as they approach
Z = zone in which the approaching through vehicle is traveling (1 toward the median opening. Furthermore, most of the vehicles are
for zone 1; 2 for zone 2; and 3 for zone 3). found to move by sharing two adjacent lanes. Therefore, these effects
It can be observed from Equation (3) that the coefficient for 3 W are assessed by introducing the concept of zones which explains the
is highest, whereas it is lowest for SUVs. Therefore, it can be said that movement of vehicles from the start of slowdown section up to the
the highest delay is faced by 3 W followed by 2 W, cars, and SUVs. start of the median opening. Depending upon the movement pattern,
The p-value for all regression coefficients and ANOVA has been three zones have been introduced, and it is observed that up to 1000
provided in Table 7. It can be seen that the p-value for all the inde- vph U-turning traffic volume, the delay in all the three zones is more
pendent variables is less than 0.05 which indicates their significance at or less same. But when the traffic volume increases above 1000 vph,
5% level of significance. Similarly, the p-value obtained from ANOVA delay faced by the vehicles in zone 3 is found to be more than that in
is also less than 0.05 which shows that there is a significant difference zone 2, and least delay is observed for the vehicles in zone 1.
between all the independent variables. Therefore, each independent Finally, a model is developed to estimate the approach delay by
variable has a significant effect on the outcome of dependent variable considering the U-turning traffic volume, zone in which the vehicle is
(approach delay). The standard error of the estimates was calculated traveling, and the category of approaching vehicle. The accuracy of the
and is found to be 0.2729. Hence, it can be said that the regression is proposed model is checked by comparing the delay values obtained
well fitted, and the model developed is statistically correct. from the model with the observed delay values in the field, and the
maximum error is around 5%. Thus, the proposed models yield results
N that are in close comparison with the field data, and it can be used to
(4) assess the performance of uncontrolled median opening.

ADavg = pi ∗ ADi
i=1

Disclosure statement
where ADavg = average approach delay faced by a traffic stream at a
definite traffic volume; pi = proportion of ith category of approaching No potential conflict of interest was reported by the authors.
through vehicle;
ADi = average delay faced by ith category of approaching through
References
vehicle; and
N = number of categories of approaching through vehicle. Aldian, A., and M. A. Taylor. 2001. “Selecting Priority Junction Traffic Models to
Determine U-turn Capacity at Median Opening.” Proceedings of the Eastern
Asia Society for Transportation Studies 3 (2): 101–113.
Ashalatha, R., and S. Chandra. 2011. “Critical Gap through Clearing Behavior of
Validation of the proposed model Drivers at Unsignalised Intersections.” KSCE Journal of Civil Engineering 15
(8): 1427–1434.
To validate the proposed delay model, approach delays computed by Bonneson, J. A., and J. W. Fitts. 1999. “Delay to Major Street through Vehicles
the model have been compared to those obtained from the field data. at Two-way Stop-controlled Intersections.” Transportation Research Part a:
For the validation purpose, separate field data were collected from a Policy and Practice 33 (3): 237–253.
8   M. MOHANTY AND P. P. DEY

Bunker, J., and R. Troutbeck. 2003. “Prediction of Minor Stream Delays Mohapatra, S. S., and P. P. Dey. 2015. “Lateral Placement of U-turns at Median
at a Limited Priority Freeway Merge.” Transportation Research Part B: Openings on Six-lane Divided Urban Roads.” Transportation Letters 7 (5):
Methodological 37 (8): 719–735. 252–263.
Caliendo, C. 2014. “Delay Time Model at Unsignalized Intersections.” Journal of Mohapatra, S. S., P. P. Dey, and S. Chandra. 2016a. “Conflicting Volume for
Transportation Engineering 140 (9): 04014042. U-turns at Uncontrolled Median Openings.” Proceedings of the Institution of
Gujarati, D. N. 2003. Basic Econometrics. 4th ed. New York: McGraw-Hill. Civil Engineers-Transport 169 (4): 195–204.
Ma, D., X. Ma, S. Jin, F. Sun, and D. Wang. 2013. “Estimation of Major Stream Mohapatra, S. S., P. P. Dey, and S. Chandra. 2016b. “Modeling the Critical Position
Delays with a Limited Priority Merge.” Canadian Journal of Civil Engineering of U-turning Vehicles at Uncontrolled Median Openings.” KSCE Journal of
40: 1227–1233. Civil Engineering 20 (1): 411–420.
Mathew, T. V., C. R. Munigety, and A. Bajpai. 2013. “Strip-based Approach for Perco, P., A. Marchionna, and N. Falconetti. 2012. “Prediction of the Operating
the Simulation of Mixed Traffic Conditions.” Journal of Computing in Civil Speed Profile Approaching and Departing Intersections.” Journal of
Engineering, ASCE 29 (5): 04014069. Transportation Engineering, ASCE 138 (12): 1476–1483.
Minh, C. C., T. H. Binh, T. T. Mai, and S. A. N. O. Kazushi. 2010. “The Delay Troutbeck, R. J., and S. Kako. 1999. “Limited Priority Merge at Unsignalized
Estimation under Heterogeneous Traffic Conditions.” Journal of the Eastern Intersections.” Transportation Research Part a: Policy and Practice 33 (3): 291–304.
Asia Society for Transportation Studies 8: 1583–1595. Wang, Y. B., W. Dong, L. Q. Zhang, and D. Chin. 2013. “A Note on Speed Modeling
Mohapatra, S. S., P. K. Bhuyan, and K. V. Rao. 2012. “Genetic Algorithm Fuzzy and Travel Time Estimation Based on Truncated Normal and Log-normal
Clustering Using GPS Data for Defining Level of Service Criteria of Urban Distributions.” Transportation Research Board (TRB) 92nd Annual Meeting,
Streets.” European Transport/Transporti Europei 52 (7): 1–18. Washington, DC.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 00:30 16 November 2017

You might also like