Académique Documents
Professionnel Documents
Culture Documents
net/publication/239391432
CITATIONS READS
18 501
3 authors, including:
Satish Chandra
Indian Institute of Technology Roorkee
92 PUBLICATIONS 1,177 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Development of Performance Parameters for Two-lane Roads under Mixed Traffic Condition View project
Effect of Countdown Display on Traffic Flow Characteristics and Driver Behaviour at Signalised Intersections View project
All content following this page was uploaded by Satish Chandra on 26 December 2014.
Abstract: Several studies are available in the literature on delays at unsignalized intersections under homogeneous and lane-disciplined
traffic conditions. The present study introduces a service delay model based on microscopic analysis of delay data under mixed traffic
conditions. The data for the study were collected at five uncontrolled intersections located in different parts of India. The data were
divided into several parts according to the category of the subject approach vehicle and type of the movement. The conflicting traffic was
measured under two categories: light and heavy vehicles. The proportion of heavy vehicles in the conflicting traffic was found to greatly
affect the service delay. The proposed service delay model is compared with Kyte’s linear model for service delay estimation.
DOI: 10.1061/共ASCE兲0733-947X共2009兲135:6共323兲
CE Database subject headings: Delay time; Vehicles; Intersections; Traffic management.
1
Professor, Dept. of Civil Engineering, Indian Institute of Technology, Review of Literature
Roorkee 247667, India. E-mail: satisfce@iitr.ernet.in
2
Junior Consultant, Frischmann Prabhu India, 315, Balgovind Wadi, Literature on uncontrolled intersections with mixed traffic where
New Prabhadevi Rd., Prabhadevi, Mumbai 400025, India. large proportion of the traffic does not follow the rules of the road
3
Lecturer, Dept. of Civil Engineering, College of Engineering, Thiru- is extremely limited. Kaysi and Alam 共2000兲 used a simulation
vananthapuram, Kerala 695 004, India.
technique to model driver behavior such as impatience, aggres-
Note. This manuscript was submitted on July 17, 2006; approved on
January 22, 2009; published online on May 15, 2009. Discussion period
siveness, experience, and traffic stream interaction at nominally
open until November 1, 2009; separate discussions must be submitted for priority unsignalized intersections. Troutbeck 共1999兲 used the
individual papers. This paper is part of the Journal of Transportation concept of limited priority at situations were minor stream ve-
Engineering, Vol. 135, No. 6, June 1, 2009. ©ASCE, ISSN 0733-947X/ hicles forced their entry into a major stream and developed rela-
2009/6-323–329/$25.00. tionships to predict the capacity and maximum delay to minor
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
street vehicles. Other important works related to delay at priority Intersection Location and type
intersection are also reviewed here for the benefit of the readers. 1. T-junction in Southern State of Kerala
Kimber et al. 共1977兲 estimated delays occurring at priority junc- 2. T-junction in Northern State of Uttar Pradesh
tions operating at or near capacity. A practical form of the flow/
3. T-junction in Northern State of Uttar Pradesh
delay relationship was suggested, which incorporated the time-
4. Four-legged intersection in Northern State of Uttar
dependent behavior of queues. Kimber and Hollis 共1979兲 Pradesh
presented methods for predicting queue length and delay at road 5. Four-legged intersection in Southern State of Kerala
junctions. The approach is quite comprehensive and takes into
account the stochastic nature of traffic demand and capacity.
Troutbeck 共1986兲 modified Tanner’s delay equation by quantify-
At each intersection, data were collected by video recording
ing the effect of bunching in the major stream. The study indi-
technique on a typical weekday. The video camera was placed at
cated that the degree of bunching in the major streams had a
a suitable vantage point near the intersection to record an
significant effect on the delay experienced by an isolated minor unobstructed view of all approaches and turning movements and
stream vehicle 共Adam’s delay兲, average delay, and the maximum data were recorded for about 2 h. The turning traffic volumes and
minor stream entry flow. Heidemann 共1991兲 analyzed formulas composition of traffic entering each leg of the five intersections
for the calculation of mean queue lengths and average delays for are given in Tables 1 and 2, respectively. The videotape was
vehicles on the minor roads at priority intersections. Kyte et al. played in the laboratory several times to get the conflicting traffic
共1991兲 developed a set of preliminary empirical models to esti- volume count and the service delay experienced by each subject
mate delay and capacity at two-way stop-controlled 共TWSC兲 in- approach vehicle. Both crossing and merging types of conflicts
tersections. A linear equation was suggested for minor street were taken into account while noting the conflicting traffic for
service time based upon the volume on the conflicting ap- each maneuver during the data extraction process. The conflicting
proaches. Akcelik and Troutbeck 共1991兲 presented a generalized traffic volume was divided into two categories as light vehicles
time-dependent delay formula for roundabouts and unsignalized and heavy vehicles 共bus and truck兲. The vehicles on subject
intersections. It can be applied to peak, off peak, as well as post- approach were divided into three categories: two wheelers, cars,
peak periods. The delay estimation process is significantly sim- and heavy vehicles. The number of vehicles observed for delay
plified. Madanat et al. 共1994兲 developed a probabilistic delay ranged from 20 共heavy vehicle, through movement兲 to 86 共car, RT
model at stop-controlled intersections. The model was based on from major兲 for different movements.
the gap acceptance functions developed through empirical data
collected from a T-intersection in Indiana 共U.S.兲. Al-Omari and
Benekohal 共1999兲 developed empirical models at undersaturated Data Extraction
TWSC intersections, to estimate service delay as a function of
conflicting traffic volume and variance of service delay as the Data extraction for the present study was done using microscopic
function of conflicting traffic volume and the average service analysis as described by Kyte et al. 共1991兲. The microscopic
delay. analysis requires the definition of the conflicting traffic flow as
Most of the above studies have been conducted in developed seen by each subject approach vehicle. Let t0 = time of arrival of
countries where rules of priority are fully respected. In a mixed the subject approach vehicle at the reference line; td = time of
traffic situation of the type prevailing in India, the personal expe- departure of the subject approach vehicle; n = number of observed
rience shows that lane driving, respect for the stop sign and the
stop line are not strictly observed. Due to loose structure of the
regulatory system, the vehicles sometimes force their entry to the
intersection, even in very small gaps in the conflicting traffic. This Table 1. Volume of Turning Traffic at Intersections
causes delays to major street vehicles also. Under these situations, Turning movements 共veh/h兲 Total entry
the results of delay studies conducted in developed countries can- volume
not be applied to mixed traffic conditions. The present study was Intersection From Left turn Straight Right turn 共veh/h兲
undertaken to carry out microscopic analysis of service delay at 1. South 310 — 195 505
three legged and four legged uncontrolled intersections in a mixed East 45 610 — 655
traffic situation. West — 490 310 800
2. South 246 — 110 356
East 66 945 — 1011
West — 905 165 1070
Data Collection 3. South 464 — 30 494
East 35 497 — 536
In order to study the service delay 共delay experienced by a vehicle West — 434 428 862
at the stop line兲, five intersections located in different parts of 4. North 145 90 95 330
India were selected. The intersection sites were in nonurban areas South 85 110 85 280
to avoid the effect of any upstream junction, parking, or bus stop East 105 520 40 665
on arrival rate. All the intersections had four-lane divided major West 60 475 125 660
streets and two-lane undivided minor streets. An important traffic
5. North 76 240 200 516
feature at all five intersections was that the queue formation on
South 120 184 60 364
the minor street approach was very rare. Three of the five inter-
East 72 800 102 974
sections were of T-type and the remaining two were four legged
West 160 570 74 804
as described below.
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 2. Traffic Composition at Intersections
Traffic composition
Total entry
Intersection From volume 共veh/h兲 Two wheelers Three wheelers Car Heavy vehicle Nonmotorized
1. South 505 152 106 66 181 —
East 655 127 135 279 114 —
West 800 245 95 341 119 —
2. South 356 121 — 128 107 7
East 1,011 344 — 422 245 25
West 1,070 362 — 391 317 11
3. South 494 115 — 256 123 18
East 536 195 — 244 97 21
West 862 312 — 386 164 23
4. North 330 119 — 112 99 15
South 280 82 — 127 71 7
East 665 230 — 212 223 24
West 660 211 — 245 204 16
5. North 516 139 98 114 165 —
South 364 102 55 93 114 —
East 974 343 117 275 239 —
West 804 278 132 240 154 —
conflicting vehicles for the subject vehicle, including the and con- Computation of Service Delay
flicting vehicle passing just after departure of the subject ap-
proach vehicle; and tn = time of arrival of nth conflicting vehicle at The microscopic analysis carried out in the present study pro-
the reference point. vided large a amount of data on service delay and the correspond-
The definition of the conflicting flow rate as seen by the par- ing instantaneous conflicting flow. The analysis was done
ticular subject approach vehicle is the number of observed con- separately for three categories of vehicles: two wheelers, cars, and
flicting vehicles divided by the observation time heavy vehicles and for two types of movements: right turn from
major, and right turn from minor 共left-hand drive rule is followed
n in India兲 at all three T-junctions. At four-legged intersections,
Conflicting flow rate = 共1兲 through traffic from a minor street was also analyzed in addition
共tn − t0兲
to the right turns from major and minor streets. Scatter plots were
This definition differs from the standard macroscopic method of made between the service delay 共Ts, s兲 and the corresponding
estimating flow rates, in which averages are reported for some conflicting traffic volume 共CT, veh/s兲. The analysis of these data
fixed period, usually 15 min or 1 h. The service delay can be is given in the following sections.
computed using Eq. 共2兲
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
It indicates that the two models given by Eqs. 共4兲 and 共5兲 are a
16
statistically better representation of the data than a combined
14 model given by Eq. 共3兲. In other words, the effect of heavy ve-
12
hicles on service delay is statistically significant. A similar exer-
cise was done for other conditions also and the effect of heavy
Service delay (s)
10 for light and heavy vehicles are shown in Figs. 6 and 7, respec-
tively.
8
Later, all delay data were combined irrespective of the vehicle
6 type and a general model was developed 共Fig. 8兲. Mathematical
models for the curves shown in Figs. 6–8 are given in Table 6.
4
The delay data were further generalized, ignoring the percentage
2 of heavy vehicles in the conflicting traffic as shown in Fig. 9. The
equation for this aggregate service delay model is given below
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Conflicting Traffic, veh/s
Ts = 2.0692e2.3036CT R2 = 0.653 共6兲
where Ts = average service delay to a priority movement, s 共RT
Fig. 2. Effect of heavy vehicles on service delay to two wheelers and TH from minor and RT from major combined兲, and CT
going right from major = conflicting traffic, veh/s.
The model, which is available in the literature, is only for the
Table 3. Statistical Parameters of the Models Developed minor street movements. Therefore, the data corresponding to the
Equation Sum of squares of Degrees of freedom right turn from major street were excluded and the delay model
number residuals 共SSR兲 共dof兲 R2 exclusively for the minor street was developed as shown in Fig.
共3兲 335.705 63 0.420 10. The proposed minor street service delay model made for com-
共4兲 69.429 34 0.610 parison purposes is given by Eq. 共7兲
共5兲 55.086 27 0.821
Ts = 2.1955e2.0868CT R2 = 0.629 共7兲
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 4. Models for Service Delay
Heavy vehicles in Service delay
Type of movement Subject vehicle conflicting traffic 共%兲 equation R2 value
RT from major Two wheelers 艋 25 Ts = 1.4649e 2.3347 CT
0.610
⬎25 Ts = 2.9251e2.3161 CT 0.8208
Car 艋25 Ts = 1.6049e2.0982 CT 0.5844
⬎25 Ts = 2.9733e2.2562 CT 0.8363
Heavy vehicle 艋25 Ts = 1.4975e2.6580 CT 0.6428
⬎25 Ts = 2.3947e2.8740 CT 0.7751
TH from minor Two wheelers 艋25 Ts = 0.7649e3.8457 CT 0.7261
⬎25 Ts = 1.6212e3.8829 CT 0.7847
Car 艋25 Ts = 1.4358e3.0281 CT 0.8243
⬎25 Ts = 2.7497e3.0243 CT 0.6731
Heavy vehicle 艋25 Ts = 0.9675e3.7660 CT 0.7757
⬎25 Ts = 2.1832e3.6241 CT 0.7443
RT from minor Two wheelers 艋25 Ts = 1.1362e2.8721 CT 0.8608
⬎25 Ts = 1.4779e3.5037 CT 0.8796
Car 艋25 Ts = 0.9185e3.2051 CT 0.7602
⬎25 Ts = 2.2334e3.1805 CT 0.7839
Heavy vehicle 艋25 Ts = 2.5751e1.8420 CT 0.9488
⬎25 Ts = 3.9585e2.1142 CT 0.9744
Note: RT= right turn; TH= through, CT= conflicting traffic volume, veh/s; and Ts = service delay, s.
16 18
Conflicting traffic with 25% or less HV Conflicting traffic with 25% or less HV
14 Conflicting traffic with more than 25% HV 16 Conflicting traffic with more than 25% HV
14
12
12
Service Delay, s
Service Delay, s
10
10
8
8
6
6
4
4
2 2
0 0
2W CAR HV 2W CAR HV
Vehicle Type Vehicle Type
Fig. 3. Service delay to RT from major street Fig. 5. Service delay to RT from minor street
18
Conflicting traffic with 25% or less HV 50
16 Conflicting traffic with more than 25% HV Conflicting traffic with 25% or less HV
45
Conflicting traffic with more than 25% HV
14 40
12 35
Service Delay, s
Service Delay, s
30
10
25
8
20
6
15
4 10
2 5
0 0
2W CAR HV 0 0.2 0.4 0.6 0.8 1 1.2
Vehicle Type Conflicting Traffic, veh/s
Fig. 4. Service delay to TH from minor street Fig. 6. Service delay to light vehicles irrespective of their movement
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 5. Statistics of t-Test
t-statistic
Vehicle types
Movement type considered Computed Tabulated Inference
RT from minor Car and two wheelers 1.476 2.052 At ␣ = 0.05 and degree of freedom 共dof兲 = 27, the null
hypothesis is accepted.
Car and heavy vehicle 0.674 2.060 At ␣ = 0.05 and dof= 25, the null hypothesis is accepted.
RT from major Car and two wheelers 1.122 2.013 At ␣ = 0.05 and dof= 47, the null hypothesis is accepted.
45 35
Conflicting traffic with 25% or less HV
40 Conflicting traffic with more than 25% HV 30
35
25
30
Service Delay, s
Service Delay, s
25 20
20 15
15
10
10
5
5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Conflicting Traffic, veh/s Conflicting Traffic, veh/s
Fig. 7. Service delay to heavy vehicles irrespective of their move- Fig. 9. Aggregate service delay ignoring the effect of heavy vehicles
ment in conflicting stream
35
50
45 30
Conflicting traffic with 25% or less HV
40 Conflicting traffic with more than 25% HV
25
Service Delay, s
35
Service Delay, s
20
30
25 15
20
10
15
10 5
5
0
0 0 0.2 0.4 0.6 0.8 1 1.2 1.4
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Conflicting Traffic, veh/s
Conflicting Traffic, veh/s
25
Table 6. Service Delay Models for Aggregate Data
20
Heavy
vehicles in
15
conflicting
Vehicle category traffic 共%兲 Service delay equation R2 value
10
Light vehicles 共Fig. 6兲 艋25 Ts = 1.5831e2.2610CT 0.5515
5
⬎25 Ts = 2.7933e2.2772CT 0.5829
Heavy vehicles 共Fig. 7兲 艋25 Ts = 1.9033e2.0606CT 0.6892 0
⬎25 Ts = 3.5782e2.0161CT 0.7909 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Combined 共Fig. 8兲 艋25 Ts = 1.6651e2.1852CT 0.5421 Conflicting Traffic, veh/s
⬎25 Ts = 2.8727e2.2341CT 0.6201
Fig. 11. Comparison of proposed service delay model with Kyte’s
Note: CT= conflicting traffic volume, veh/s; Ts = service time or service
delay, s. model
Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Kyte et al. 共1991兲 developed a linear equation for minor street scenario at uncontrolled intersections. It is mainly due to impa-
service delay, as given below, after conducting microscopic stud- tient behavior of drivers and difference in the static and dynamic
ies at four different intersections in the Pacific Northwest. This characteristics of vehicles in India and other developing countries.
relationship was developed for a conflicting traffic volume range The proposed model is developed for mixed traffic conditions of
of 183 veh/ h to 1,100 veh/ h 共0.051 veh/ s to 0.31 veh/ s兲 the type prevailing on Indian roads and can be used to better
estimate the service delay to various priority movements at un-
Ts = 17.28 · CT 共8兲
controlled intersections. Service delay is related to the departure
Fig. 11 compares the service delay as obtained from Eqs. 共8兲 headway, which in turn, is the inverse of capacity of a movement
and 共9兲. As may be seen, the prediction of delay by the proposed type. Therefore, service delay models suggested here can also be
model is on the lower side of Kyte’s equation beyond a conflict- used by practicing engineers to estimate the capacity of a priority
ing traffic of 0.20 veh/ s. Hence, it can be said that service delay movement at an unsignalized intersection under mixed and unruly
varies linearly with conflicting traffic up to a conflicting traffic of traffic behavior.
0.20 veh/ s, beyond which an exponential trend is followed. The
Kyte’s linear model is valid only for a low conflicting traffic of up References
to 0.20 veh/ s, beyond which it would overestimate the service
delay. When conflicting traffic is very high, say more than Akcelik, R., and Troutbeck, R. 共1991兲. “Implementation of Austrian
3,500 veh/ s, the available gaps in the major stream vehicles are roundabout analysis method in SIDRA.” Proc. 1st Int. Symp. on High-
unacceptably low and service delay to low priority movements way Capacity, Balkema, Karlsruhe, Germany, 17–34.
increases rapidly. Such high traffic volume occurs under mixed Al-Omari, B., and Benekohal, R. 共1999兲. “Hybrid delay models for un-
traffic because of the undisciplined movement leading to cluster- saturated two-way stop-controlled intersections.” J. Transp. Eng.,
ing of traffic. Linear models would certainly fail under such con- 125共3兲, 291–296.
ditions. Heidemann, D. 共1991兲. “Queue length and waiting time distributions at
priority intersections.” Transp. Res., Part B: Methodol., 25共4兲, 163–
174.
Highway Capacity Manual 共HCM兲. 共2000兲. “Highway capacity manual.”
Conclusions SR, 209, Transportation Research Board, National Research Council,
Washington, D.C.
This study provides preliminary exponential models to estimate Kaysi, I., and Alam, G. 共2000兲. “Driver behavior and traffic stream inter-
service delay to different categories of vehicles for various prior- actions at unsignalized intersections.” J. Transp. Eng., 126共6兲, 498–
ity movements and for varying proportions of heavy and light 505.
vehicles in the conflicting traffic. It is observed that the service Kimber, R. M., and Hollis, E. M. 共1979兲. “Traffic queue and delays at
delay to a vehicle does not depend much on its type. It depends road junctions.” DOE, DTp. TRRL Rep. No. LR 909, Transport and
only on the amount of conflicting traffic and its composition in Road Research Laboratory, Crowthrone, U.K.
terms of proportion of heavy vehicles. The delay data show a Kimber, R. M., Marlow, M., and Hollis, E. M. 共1977兲. “Flow/delay rela-
tionships for major/minor priority junctions.” Traffic Eng. Control,
distinct demarcation when the proportion of heavy vehicles in the
18共11兲, 516–519.
conflicting traffic is greater than 25%. It is found to increase the
Kyte, M., Clemow, C., and Khisty, C. J. 共1991兲. “Capacity and delay
service delay by about 1.7 to 2.4 times that with less than 25% characteristics of two-way stop-controlled intersections.” Transporta-
heavy vehicles in the conflicting traffic. The delay to a vehicle tion Research Record. 1320, Transportation Research Board, National
from the minor street 共through or turning兲 is about 1.1 to 1.4 Research Council, Washington, D.C., 160–167.
times more than that for a right turn movement 共for left-hand Lu, J. J., and Lall, B. K. 共1995兲. “Empirical analysis of traffic character-
drive conditions兲 from the major street. The delay for a right turn istics at two-way stop-controlled intersections in Alaska.” Transpor-
from the minor street is almost similar to that for a straight tation Research Record. 1495, Transportation Research Board,
through movement from the minor street. Earlier studies have National Research Council, Washington, D.C., 49–56.
shown that the service delay is linearly dependent upon the con- Madanat, S. M., Cassidy, M. J., and Wang, M.-H. 共1994兲. “Probabilistic
flicting traffic volume, especially under low ranges of conflicting delay model at stop-controlled intersection.” J. Transp. Eng., 120共1兲,
21–36.
traffic. The comparison of the proposed delay model with Kyte’s
Troutbeck, R. J. 共1986兲. “Average delay at an unsignalized intersection
model shows that the linear variation of service delay is valid up with two major streams each having a dichotomized headway distri-
to a conflicting traffic volume of 0.20 veh/ s, beyond which it is bution.” Transp. Sci., 20共4兲, 272–286.
exponential. Troutbeck, R. J. 共1999兲. “Capacity of limited priority merge.” Transpor-
The models developed under homogeneous and lane disci- tation Research Record. 1678, Transportation Research Board, Na-
plined traffic conditions cannot be applied to the mixed traffic tional Research Council, Washington, D.C., 269—276.
View publication stats Downloaded 07 Jul 2010 to 59.163.196.106. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org