Effects of Two-Wheelers on Saturation Flow at Signalized

Intersections in Developing Countries

C. S. Anusha 1; Ashish Verma, Aff.M.ASCE 2; and G. Kavitha 3

Abstract: Two-wheelers (TW) constitute a major proportion of urban traffic in developing countries and therefore their effect on the sat-
uration flow at signalized intersections could be substantial. This paper attempts to study and analyze the effect of two-wheelers on the
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

saturation flow of signalized intersections by collecting data at a few signalized intersections in Bangalore, India. A strong correlation
is observed between the measured saturation flow and the proportion of two-wheeler traffic, which suggest that two-wheelers have significant
impact and should be considered in the capacity analysis of signalized intersections. In this paper, the effect of two-wheelers on saturation
flow rate is incorporated in a previous model by calibrating and introducing a new adjustment factor for two-wheelers. Results show that
saturation flow measured using the modified HCM equation is closer to observed saturation flow values. DOI: 10.1061/(ASCE)TE.1943-
5436.0000519. © 2013 American Society of Civil Engineers.
CE Database subject headings: Traffic capacity; Traffic signals; Traffic flow; Heterogeneity; Intersections; India; Developing countries.
Author keywords: Traffic capacity; Traffic engineering; Traffic signals; Heterogeneity; Intersections; India.

Introduction to analyze the capacity of signalized intersections in developing

countries, which comprises predominantly heterogeneous traffic
Traffic signals are the most common form of traffic control mea- conditions, there is a need to modify these models to suit the pre-
sures used in urban areas of both developed and developing coun- vailing conditions.
tries. Most of the developed countries have models in place, Saturation flow rate is the basic parameter used to derive
developed based on their conditions, to analyze the capacity of capacity of a signalized intersection. It is determined based on
the signalized intersections. These models are best suited for de- the minimum headway that the lane group can sustain across
veloped countries where the flow is homogeneous and follows the stop line. Several attempts have been made previously to
lane discipline. In most of the developing countries like India, model saturation flow. Sarna and Malhotra (1967) conducted stud-
two-wheelers are the major mode of personal transportation. In ies on saturation flow at a number of different intersections with
the city of Bangalore, India, two-wheelers constitute more than varying approach road widths. They developed the relationship
70% of the total volume, while cars comprise 15%, autos 4%, between the saturation flow and the approach road width at sig-
and the remaining includes other vehicles such as buses, vans, nalized intersections. Also, the effect of approach volume and
and tempos (http://www.bangaloretrafficpolice.gov.in). Because increasing percentage of bicycles on the saturation flow was stud-
of this high composition of two-wheelers, the urban traffic char- ied. The study has shown that the saturation flow increases with
acteristics in developing countries are significantly different from the increase in approach volume. Zegeer (1986) conducted a field
those of developed countries. Two-wheelers, privileged by their survey to find saturation flow and to verify saturation flow and
size, make use of the small gaps between the bigger vehicles traffic volume adjustment factors used in various capacity manuals
to reach the head of the queue. Lane changing and overtaking throughout the United States at signalized intersections. Saturation
by these two-wheelers are more commonly observed. Due to such flow headways for more than 20,000 observations were collected.
complex maneuvring characteristics of two-wheelers, it is not pos- Various factors like road geometry, traffic characteristics, environ-
sible to use the U.S. Highway Capacity Manual (US-HCM) model mental, and signal cycle lengths were considered to develop a
directly because it has been developed for a homogenous lane- series of modified adjustment factors to determine modified sat-
based traffic flow. Hence, when the US-HCM model is being used
uration flow rates while calculating the signalized intersection
1 capacity.
Research Scholar, Dept. of Civil Engineering, Indian Institute of
The US-HCM 2000, developed by Transportation Research
Science (IISc), Bangalore 560012, Karnataka, India. E-mail: hoshatti@
gmail.com Board (TRB), includes a model to calculate saturation flow rate
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of considering the effect of various factors like width, gradient,
Science (IISc), Bangalore 560012, Karnataka, India (corresponding parking activity, heavy vehicles, area type, turning movements,
author). E-mail: ashishv@civil.iisc.ernet.in; rsashu@yahoo.com and pedestrian blockage (TRB 2000). It assigns an adjustment
3
Assistant Professor, RASTA, Centre for Road Technology, Extension factor to each of these parameters, which can be calculated using
centre for VTU, Peenya phase 1, Bangalore 560058, Karnataka, India. empirical formulas proposed in US-HCM 2000. These adjust-
E-mail: kavi_gsy@yahoo.com ment factors are multiplied to the base saturation flow so, which
Note. This manuscript was submitted on May 3, 2012; approved on
is considered to be 1,900 passenger cars (pc) per hour of green
November 6, 2012; published online on November 8, 2012. Discussion
period open until October 1, 2013; separate discussions must be submitted time per lane for a signalized intersection, to obtain the satura-
for individual papers. This paper is part of the Journal of Transportation tion flow rate S of the intersection approach. Following is the
Engineering, Vol. 139, No. 5, May 1, 2013. © ASCE, ISSN 0733-947X/ model proposed in US-HCM 2000 to determine the saturation
2013/5-448-457/$25.00. flow rate:

448 / JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013

J. Transp. Eng. 2013.139:448-457.

 S ¼ so Nfw f HV f g fe fbb f a f LU fLT f RT f Lpb f Rpb ð1Þ two-wheelers, auto rickshaws, and light commercial vehicles
(LCVs), and hence it is important to also consider the effect of
where S = saturation flow rate for the lane group in vehicles per mixed-traffic flow on saturation flow rate.
hour (veh=h); so = base saturation flow rate per lane (pc=h=lane); This has been recognized in past studies and attempts have been
N = number of lanes in a lane group; fw = adjustment factor made to model the effects of mixed-traffic flow on saturation flow.
for lane width; fHV = adjustment factor for heavy vehicles; Arasan and Jagadeesh (1995) proposed a probabilistic approach
f g = adjustment factor for approach grade; f e = adjustment fac- based on the first-order second-moment method to estimate satu-
tor for parking activity; f bb = adjustment factor for blocking ef- ration flow at signalized intersections under heterogeneous traffic
fect of local buses; f a = adjustment factor for area type; flu = conditions. They make a comparison between the conventional
adjustment factor for lane utilization; f lt = adjustment factor for method of estimating saturation flow, i.e., the headway method,
left turn; f rt = adjustment factor for right turn; fLpb = pedestrian and their newly proposed probabilistic approach. The authors
adjustment factor for left turn; and fRpb = pedestrian adjustment
found the probabilistic approach to be more appropriate for the
factor for right turn.
Indian condition.
The Indian Roads Congress (IRC) also proposes an empirical
Hossain (2001) analyzed the traffic characteristics and opera-
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

formula to estimate saturation flow. Following is the equation

tions at the signalized intersections of Dhaka, Bangladesh, and con-
proposed by IRC:
cluded that there is a need for a different modeling approach to
S ¼ 525
ðwÞ · passenger car units ðPCUsÞ per hour ð2Þ analyze the saturation flow rates at the intersections of developing
nations, and the concept of PCU, which is widely used as a signal
where w = width. This expression is valid for widths from 5.5 to design parameter, is not applicable in the case of mixed traffic com-
18 m. This empirical equation does not cater for many other factors posed of both motorized and nonmotorized vehicles. Hossain also
like category of vehicle, area type, and gradient. tried to develop a new microscopic simulation technique in which a
As can be seen from Eq. (1), the effect of the type of vehicles is coordinate approach to modeling vehicle location is adopted. Based
considered only in terms of the heavy vehicle adjustment factor, on these simulation results, an equation was developed to estimate
which is obtained using the following equation: the saturation flow from the influencing variables like road width,
1 turning proportion, and percentage of heavy and nonmotorized
f HV ¼ ð3Þ vehicles.
1 þ PTðPCET − 1Þ
Arasan and Vedagiri (2006) also proposed a simulation model,
where f HV = heavy vehicle adjustment factor; PT = percent heavy HETEROSIM, to estimate the saturation flow rate of heterogeneous
vehicles; and PCET = heavy vehicle’s passenger car equi- traffic. Simulation results were used to study the effect of road
valent (PCE). However, as mentioned previously, the vehicle com- width on saturation flow measured in PCUs per unit width of road.
position in a country like India is composed of vehicles such as Studies showed that there is a significant increase in saturation

Input: Intersection geometrics,

Cycle details, Volume data,
Saturation flow

Theoretical saturation flow Correlation b/w measured saturation flow

using US-HCM-2000 model & type of vehicle & approach vol.

Comparison b/w Theoretical

Identification of factors to be considered to
saturation flow & measured
modify US-HCM-2000 model
saturation flow

Derivation of new
adjustment factors
Error No
accepted

Validation of No
modified
Yes model

Yes

End

Fig. 1. Proposed methodology

JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013 / 449

J. Transp. Eng. 2013.139:448-457.

 flow rate with the increase in the width of the approach road, and modifications. Considering this, the objective of the current
HETEROSIM measures saturation flow effectively in the study is to study the impact of various categories of vehicles
Indian scenario. on saturation flow rate and to modify the US-HCM 2000 mo-
Kockelman and Shabih (2000) analyzed the impacts of differ- del to suit Indian conditions incorporating the effect of two
ent light-duty trucks (LDTs) on the capacity of signalized inter- wheelers.
sections. Rahman et al. (2004) conducted studies in the city of
Dhaka, Bangladesh, to analyze the effect of nonmotorized and
motorized auto rickshaws on the capacity of signalized intersec- Methodology
tions. They proposed new passenger car equivalent values for auto
Fig. 1 depicts the proposed methodology for calibration of
rickshaws for capacity analysis of intersections. Minh and Sano
the US-HCM 2000 (TRB 2000) saturation flow model for the
(2003) conducted studies at signalized intersections in Hanoi,
Indian condition. The following are the steps involved in the
Vietnam, and Bangkok, Thailand, to analyze the effect of motor- methodology:
cycles on saturation flow rate. They concluded that effect of mo- 1. The input data required include intersection geometric details,
torcycles is significant and should be taken into account in
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

signal cycle details, approach volume data, and measured

geometric design and operation of signalized intersections. Patil saturation flow.
et al. (2007) developed simple regression models to estimate 2. The theoretical saturation flow is calculated using the
saturation flow at signalized intersections having heterogeneous US-HCM 2000 model (TRB 2000). Also, the correlation be-
traffic. The model was developed based on parameters like road tween the measured saturation flow and types of vehicle and
width, percentage of heavy vehicles, and proportion of turning approach volume is calculated.
vehicles, and was validated on the basis of data collected from 3. A comparison is made between the theoretical and measured
Mumbai, India. Zhang and Chen (2009) conducted a saturation saturation flow. If they are comparable within the error limits,
flow study in Nanjing City, China. They concluded that the meth- then it can be concluded that the US-HCM 2000 model (TRB
odology for saturation flow rates put forward by the HCM can 2000) is good enough for Indian conditions, and the process
also be used in China provided the parameters involved are sys- ends. In case the error is more than the acceptable limit, the
tematically calibrated for Chinese conditions. next step is followed.
Summarizing the review of past literature, it is clear that the 4. Based on the correlation analysis, factors to be considered
model proposed by US-HCM 2000 [Eq. (1)] (TRB 2000) can for calibration of the US-HCM 2000 (TRB 2000) model
be adapted to developing countries after necessary calibration/ are identified.

Fig. 2. ESI Hospital junction, Rajajinagar, Bangalore, India

450 / JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013

J. Transp. Eng. 2013.139:448-457.

 5. With the input from Step 4, new adjustment factors are derived Modification of Saturation Flow Model for Indian
and the modified US-HCM 2000 (TRB 2000) model is Conditions
validated for Indian conditions by comparing the modified the-
oretical saturation flow with the measured saturation flow. If The modification (calibration) of the US-HCM 2000 saturation
the error is within the acceptable limit, then the process stops; flow model [Eq. (1)] (TRB 2000) for Indian conditions is done
otherwise Step 4 is repeated until the modified US-HCM 2000 by collecting data at a few signalized intersections in Bangalore,
model is validated. India. All signals identified are operating as pretimed signals.
The next section presents the case study. The intersections are selected considering the geometry of the
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. 5th Main Chamarajpet junction, Basavangudi, Bangalore, India

Table 1. Geometric Details, Volume, and Saturation Flow for Selected Junctions
Intersection Study approaches Area type Width (m) Gradient (%) Cycle time (s) Green time Volume (veh=h) Saturation flow (veh=h)
ESI Hospital junction Approach Aa CBD 5.2 −1.97 118 40 1,422 3,852
Approach Bb CBD 4.15 −3.83 118 40 649 2,532
Approach Cc CBD 4.95 −3.3 118 39 824 2,156
Approach Dd CBD 4.7 −3.34 118 30 721 3,930
Chamarajpet junction Approach 1e CBD 6.2 −3.36 205 104 1,847 4,614
Approach 2f CBD 5.9 2.16 205 44 699 3,845
Approach 3g CBD 6.2 −1.71 205 60 1,153 2,630
Approach 4h CBD 6 −0.61 205 95 651 2,964
Note: CBD = Central Business District.
a
Toward Majestic.
b
Toward WOC.
c
Toward Rajaji Nagar Police Station (RPS).
d
Toward Magadi Road.
e
Toward Majestic (Chamarajpet).
f
Toward Basavangudi.
g
Toward Mysore Road.
h
Toward K.R. Market.

JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013 / 451

J. Transp. Eng. 2013.139:448-457.

 Table 2. Correlation Coefficients (r)
r
Intersection Study approaches Two-wheelers Cars/jeeps Auto rickshaws Heavy vehicles LCVs Buses
Chamarajpet junction Approach 1 0.94 0.3 0.75 0.45 — —
Approach 2 0.71 0.49 0.67 0.17 — —
Approach 3 0.6 0.69 0.28 −0.05 — —
Approach 4 0.87 0.51 0.6 0.44 — —
ESI Hospital junction Approach A 0.78 0.007 0.5 — 0.13 0.05
Approach B 0.66 0.53 0.17 — 0.02 —
Approach C 0.42 0.55 0.1 — 0.34 —
Approach D 0.65 0.39 0.76 — 0.05 0.01

intersection and availability of high-rise buildings near the intersec-

Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

tion (for data collection using video camera). Because the objective
of the study is to find the impact of vehicle type, particularly two-
wheelers, on saturation flow, the intersections are selected based on
reducing the impact of other factors. Hence, intersections with uni-
form geometry, minimum gradient, and without bus bays and park-
ing bays are selected. From the initial set of 12 intersections, two
intersections were selected for calibration and another two for val-
idation. Following are the details of the two intersections selected
for calibration.
• ESI Hospital Junction, Rajajinagar, Bangalore, India: This is a
four-legged intersection at Rajajinagar, Bangalore, India (Fig. 2).
The surface condition of the roads was good during the time of
study, hence good platoon speed is observed. The highest traffic
flow is observed from west of Chord Road (WOC) toward
Fig. 4. Difference between field saturation flow and saturation flow
Majestic. The traffic composition includes two-wheelers, cars,
estimated using US-HCM 2000 (TRB 2000)
auto rickshaws, buses, and LCVs.

Fig. 5. Capacity versus category of vehicle

452 / JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013

J. Transp. Eng. 2013.139:448-457.

 Table 3. PCE Values Used in the Calculation Table 5. R-Square Values for Different Model Combinations
Serial number Class of vehicle PCE values Combinations R-square
1 Car, jeep 1.0a ftw, proportion of two-wheelers, and volume 0.64
2 LCV(minitruck) 1.5a ftw, proportion of two-wheelers, and 1=volume 0.72
3 Auto rickshaw 0.4a ftw, proportion of two-wheelers, and width divided 0.62
4 Two-wheeler 0.3a by volume
5 Bus 2.0b ftw, proportion of two-wheelers, and volume divided 0.933
6 Truck 2.0b by width
a
Justo and Tuladhar (1984).
b
US-HCM-2000 (TRB 2000).
conditions (Patil et al. 2007) have observed that the start-up lost
time in Indian traffic conditions is much lower than 10 s and is
• 5th Main Chamarajpet Junction, Basavangudi, Bangalore, India: considered to be 5 s after the green is displayed. Now the video
This is a four-legged intersection at Chamarajpet, Bangalore, clip is reversed to the original position and replayed. This time,
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

India (Fig. 3). This intersection has a substantial number of a classified vehicle count is done for each direction of movement.
vehicles arriving from Majestic and K.R. Market moving to- It is not possible to take a classified vehicle count for all directions
wards Basavangudi. The surface condition of the roads is good of movement at the same time, therefore the video is replayed a
with good road markings. The traffic composition mainly con- number of times and every time a vehicle count of one or two cat-
sists of two-wheelers, auto rickshaws, cars, buses, LCVs, and egories is done. The previous procedure is repeated for each cycle
trucks. Figs. 2 and 3 also give the geometrical representation of the recorded period. Adopting the same procedure, saturation
of the two intersections. Traffic studies were conducted at flow is measured for all eight approaches of the intersections con-
the selected junction to collect the following data: sidered and the values are tabulated as shown in Table 1. When
• Classified traffic volume with turning movements using these measured saturation flows were plotted against widths of
video camera; the approach roads (Fig. 4), it was observed that it did not follow
• Intersection geometry; and the popular trend that an increase in the width of the approach road
• Saturation flow. increases the saturation flow. Instead, low saturation flows were
observed on even wider roads. This depicts that there are other fac-
tors that affect saturation flow.
Saturation Flow Measurement
Saturation flow rate is defined as the maximum discharge rate dur- Derivation of Adjustment Factor
ing green time. It is expressed either in PCU/hour or vehicles/hour.
Saturation period and direction-wise classified traffic volume is Initially the correlation between the measured saturation flow and
necessary to calculate saturation flow for a particular lane group. category of vehicle is calculated to observe the significance of
The procedure for measuring prevailing saturation flow is summa- correlation. Obtained correlation coefficients are tabulated in
rized subsequently. Table 2. From the table, it is found that the highest correlation
An observation point is selected by playing a video clip re- of 0.94 is observed with two-wheelers, which clearly shows that
corded for the particular approach of the intersection. The obser- saturation flow is very sensitive to the two-wheelers’ population.
vation point is normally the stop line. Start of the green is noted Positive correlation infers that an increase in population of two-
down from the video camera timer. The available video camera wheelers increases the saturation flow. A strong positive correlation
gives time with accuracy of 1 min. A conventional stop watch timer is also observed with cars and auto rickshaws for some approaches.
is used to measure time in seconds. The video clip is paused at the Further, the capacity of each approach is calculated by using the
moment the signal turns green, and the stop watch is set to zero. following standard equation:
The video is played until the last vehicle in the queue crosses the c ¼ S
g=C ð4Þ
observation point. The saturation period is noted down from
the stop watch timer. The period of saturation flow begins when where c = capacity of approach (veh=h); S = saturation flow rate
the green is displayed for 10 s. Saturation flow ends when the rear (veh=h); g = average effective green time (s); and C = average cycle
axle of the last queued vehicle at the beginning of the green time time (s).
crosses the stop line [US-HCM 2000 (TRB 2000)]. This initial 10 s Graphs (Fig. 5) of capacity versus category of vehicles are
is the allowance for start-up loss, however past studies on Indian plotted in order to find out variation in capacity with respect to

Table 4. Adjustment Factors and Calculated Saturation Flow

Study Number of Saturation flow (veh=h)
approaches fhv(al) fw fg frt flt lanes (US-HCM-2000)
Approach A 1.02145 0.942 1.01 1 0.966 2 3,211
Approach B 1.101322 0.906 1.02 1 0.988 2 3,439
Approach C 0.97561 0.878 1.017 0.99 0.962 2 2,837
Approach D 1.18 0.878 1.017 0.970 0.993 2 3,485
Approach 1 1.042753 0.967 1.02 0.99 0.981 2 3,416
Approach 2 1.18624 0.956 0.989 1 0.995 2 3,817
Approach 3 1.051525 1 1.003 0.98 0.993 2 3,525
Approach 4 1.123596 0.961 1.005 1 0.914 2 3,392
Note: fhv(al) = adjustment factor for category of vehicle (heavy vehicle, LCV, and auto rickshaw); fw = adjustment factor for width of the approach; fg =
adjustment factor for gradient of the approach; frt = adjustment factor for right-turning movement; and flt = adjustment factor for left-turning movement.

JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013 / 453

J. Transp. Eng. 2013.139:448-457.

 Table 6. Statistical Analysis where PAR = percentage of auto rickshaws; and PCEAR = PCE
Regression statistic Value value of auto rickshaws. Adopting this modified equation
[Eq. (5)] and adopting the PCE values for each category of vehicle
Multiple R 0.966091
given in Table 3, f HV values were estimated for all the approaches,
R-square 0.933331
Adjusted R-square 0.899997
along with other adjustment factors required for the calculation of
Standard error 0.07696 the saturation flow using US-HCM 2000 (TRB 2000). The same
Observations 7 are tabulated in Table 4.
While comparing the saturation flow values in Table 4 with
field saturation flows, it is observed that the values are still not
matching within the acceptable error range, with the difference
Table 7. t Stat and P-Value
in values as high as 35% (Fig. 4). Hence, this clearly indicates that
Model variables Coefficients Standard error t stat P-value the effect of two-wheelers should also be considered in order to get
Intercept 0.378608 0.242114 1.563757 0.192918 the saturation flow closer to field saturation flows. In order to fur-
Two-wheelers −0.80027 0.523187 −1.52961 0.200851 ther incorporate the effect of two-wheelers into the US-HCM 2000
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

Volume/width 0.004763 0.00067 7.112176 0.002065 equation (TRB 2000), a new adjustment factor ftw is introduced.
The variation between the proportion of two-wheelers and capacity
is clearly understood from Fig. 5, i.e., an increase in the proportion
category of vehicle. From the graphs, it is clear that apart from of two-wheelers increases the capacity. Capacity and saturation
two-wheelers in which capacity increases with an increase in pro- flow are directly proportional and hence an increase in the propor-
portion of two-wheelers, the other classes of vehicles, including tion of two-wheelers increases the saturation flow. Therefore, the
auto rickshaws, which are predominant in Indian roads, result in a new adjustment factor ftw should increase with the increase in the
decrease in capacity with the increase in their respective proportion. proportion of vehicles.
Kockelman and Shabih (2000) observed a similar kind of variation The residual variation in measured field saturation flow and
for LCVs and appended the adjustment factor for heavy vehicles theoretical saturation flow is assumed to be explained by the sub-
[Eq. (3)] as follows: stantial presence of two-wheelers on Indian roads and its effect is
considered by introducing a multiplicative adjustment factor for
1 two-wheelers (ftw) in the US-HCM 2000 model (TRB 2000).
f HV ¼ ð5Þ
1 þ PTðPCET − 1Þ þ PLTðPCELT − 1Þ An attempt is made to develop a multiple linear-regression equation
to calculate this factor, taking into account the independent varia-
where PLT = percent LCVs; and PCELT = LCVs’s PCE. bles, namely, proportion of two-wheelers, approach width, and vol-
To incorporate the effects of auto rickshaws, Eq. (4) was further ume. Table 5 gives the details of various combinations tried and
appended in a similar manner as Kockelman and Shabih (2000). corresponding R-square values obtained. Finally, the following
The following is the modified form of equation: equation is found to be appropriate:
1
f HV ¼  
1 þ PTðPCET − 1ÞþPLTðPCELT − 1ÞþPARðPCEAR − 1Þ vol
ftw ¼ 0.378 − 0.8 · ptw þ 0.004 · ð7Þ
ð6Þ w

Fig. 6. Geometric details of Cash Pharmacy junction

454 / JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013

J. Transp. Eng. 2013.139:448-457.

Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Geometric details of Museum Road junction

Fig. 8. Geometric details of Kundalahalli Gate junction

JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013 / 455

J. Transp. Eng. 2013.139:448-457.

 Table 8. Geometric and Volume Data of the Intersections Considered for Validation
Study Area Width Gradient Cycle Green Volume Saturation flow
Intersection approaches type (m) (%) time (s) time (veh=h) (veh=h)
Cash Pharmacy Approach Ia CBD 17.7 0.98 160 120 4,960 8,860
Museum Road junction Approach IIb CBD 16.6 −2.96 130 80 5,684 12,511
Kundalahalli Gate junction Approach IIIc CBD 11.3 0.9 230 86 3,810 5,981
Note: CBD = Central Business District.
a
Toward Mayo Hall.
b
Toward Brigade.
c
Toward Whitefield.

Table 9. Validation Results

Study Number of Saturation flow (veh=h) Saturation flow Saturation flow
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

approaches fhv(al) fw fg frt flt ftw lanes (US-HCM-2000) (veh=h) (actual) (veh=h) (calibrated)
Approach I 1.04821 1.25 0.99 1 1 1.44 3 6,527 8,860 9,433
Approach II 1.04188 1.06 1.01 1 0.95 1.72 4 7,270 12,511 12,505
Approach III 0.9738 1.227 0.99 0.97 0.93 1.33 2 4,096 5,981 5,514
Note: fhv(al) = adjustment factor for category of vehicle (heavy vehicle, auto); fw= width of the approach; fg = gradient of the approach; frt = right-turning
movement; flt = left-turning movement; ftw = new adjustment factor for two-wheelers. The bold values are the values obtained by the developed equation.

where ftw = adjustment factor for two-wheelers; ptw = proportion Validation

of two-wheelers; vol = approach volume; and w = width of the
approach. Three different intersections in the city of Bangalore, India, were
Tables 6 and 7 give the statistical analysis carried out for the considered for validation:
selected equation. An R-square of 0.933 shows that the obtained • Museum Road junction;
equation is statistically stable. Further, the equation is validated • Cash Pharmacy junction; and
using data from another set of junctions. • Kundalahalli Gate junction.
The geometric and volume details of the intersection are repre-
sented in Figs. 6–8, respectively, and Table 8. The saturation flow is
measured at these junctions for selected approaches and is also
calculated using the original US-HCM 2000 model [Eq. (1)]
and using the calibrated US-HCM 2000 model (by incorporating
ftw) (TRB 2000). The validation results are shown in Table 9
and different saturation flows have been plotted as shown in Fig. 9.
It can be seen that saturation flow measured using the calibrated
US-HCM 2000 model is closer to the field-measured satura-
tion flow. Percentage errors before and after calibration of the
US-HCM 2000 model are shown in Fig. 10, which shows a signifi-
cant decrease in the percentage error after calibration.

Conclusions
Fig. 9. Comparison of saturation flow before and after calibration of
the US-HCM 2000 model (TRB 2000) with actual field saturation Following are the conclusions drawn from the present study:
flows • The saturation flow analyzed for different approaches
shows that it does not depend only on width of the approach
(w); therefore the empirical formula 525w suggested for Indian
conditions in Special Publication (SP)-41 (IRC 1994) of the
Indian Roads Congress is inappropriate for obtaining saturation
flow.
• From the correlation coefficient computed between saturation
flow and type of vehicle, it is observed that two-wheelers have
significant correlation with saturation flow on most of the ap-
proaches and it varies from 0.42 to 0.94.
• While intersection capacity varied directly with the increase in
the volume of two-wheelers, it was inversely proportional to the
increase in volume of all other categories of vehicles.
• The saturation flow estimated using the calibrated US-HCM
2000 model (TRB 2000) is closer to field values, which implies
Fig. 10. Reduction in percentage error in saturation flows after
that the effects of two-wheelers and approach volume are to be
calibration
considered while modeling saturation flow in Indian conditions.

456 / JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013

J. Transp. Eng. 2013.139:448-457.

 This method can be treated as an alternative to the traditional Hossain, M. (2001). “Estimation of saturation flow at signalised intersec-
525W empirical formula method (IRC 1994), which has been tions of developing cities: A micro-simulation modelling approach.”
proved to be inappropriate. Transp. Res. Part A, 35(2), 123–141.
Indian Roads Congress (IRC). (1994). “Guidelines for design of at-grade
intersections in rural and urban areas.” IRC Special Publication, No. 41,
Future Scope Indian Roads Congress, New Delhi, India.
Justo, C. E. G., and Tuladhar, S. B. S. (1984). “Passenger car unit values for
US-HCM 2000 (TRB 2000) proposes adjustment factors to account urban roads.” J. Indian Road Congr., 45(1), 183–238.
for the disturbance caused by parked vehicles and bus blockages. In Kockelman, K. M., and Shabih, R. A. (2000). “Effect of light duty vehicle
India, parking is haphazard and buses allow boarding and alighting on signalized intersection.” J. Transp. Eng., 126(6), 506–512.
Minh, C. C., and Sano, K. (2003). “Analysis of motorcycle effects to sat-
even at stop lines of the intersection, which influences the satura-
uration flow rate at signalized intersection in developing countries.”
tion flow greatly. These two major factors could be considered fur- J. East. Asia Soc. Transp. Stud., 5, 1211–1222.
ther and modeled. Patil, R. G., Roa, K. V. K., and Xu, M. S. N. (2007). “Saturation flow es-
While measuring saturation flow in this study, it is assumed timation at signalized intersections in developing countries.” Proc.,
based on past studies that start-up lost time is 5 s in Indian con-
Downloaded from ascelibrary.org by Selcuk Universitesi on 01/30/15. Copyright ASCE. For personal use only; all rights reserved.

86th Annual Meeting (CD-ROM), Transportation Research Board,

ditions. However, from the field observations carried out in this Washington, DC.
study, practically no initial start-up lost time is observed. Hence, Rahman, M., Okura, I., and Nakamura, F. (2004). “Effects of rickshaws
it is necessary to carry out research to establish an appropriate stan- and auto-rickshaws on the capacity of urban signalized intersections.”
dard of saturation flow for Indian conditions. IATSS Res., 28(1), 26–33.
Sarna, A. C., and Malhotra, S. K. (1967). “Study of saturation flow at
traffic light controlled intersections.” J. Indian Road Congr., XXX-2,
References 303–327.
Transportation Research Board. (2000). Highway capacity manual,
Arasan, V. T., and Jagadeesh, K. (1995). “Effect of heterogeneity of Washington, DC.
traffic on delay at signalized intersections.” J. Transp. Eng., 121(5), Zegeer, J. D. (1986). “Field validation of intersection capacity factors.”
397–404. Transportation Research Record 1091, Transportation Research Board,
Arasan, V. T., and Vedagiri, P. (2006). “Estimation of saturation flow Washington, DC.
of heterogeneous traffic using computer simulation.” Proc., 20th Zhang, G., and Chen, J. (2009). “Study on saturation flow rates for signal-
European Conf. on Modelling and Simulation, ECMS, Bonn, ized intersections.” Proc., Int. Conf. on Measuring Technology and
Germany. Mechatronics Automation, IEEE, New York, 598–601.

JOURNAL OF TRANSPORTATION ENGINEERING © ASCE / MAY 2013 / 457

J. Transp. Eng. 2013.139:448-457.