International Journal of Science and Technology Education Research Vol. 1(6), pp. xxx - xxx, November 2011
Available online http://www.academicjournals.org/IJSTER
ISSN 2141-6559 ©2010 Academic Journals
Full Length Research Paper
Estimation of Roadway Capacity of Eight-lane Divided
Urban Expressways under Heterogeneous Traffic
Through Microscopic Simulation Models
Errampalli Madhu* and S. Velmurugan
Traffic Engineering and Transportation Planning Area, Central Road Research Institute (CRRI)
Mathura Road, CRRI (P.O.), New Delhi - 110020, India.
Accepted 3 November, 2011
The aim of the study is to develop a microscopic simulation model by considering eight-lane divided
urban expressway. For this purpose, the partially access controlled Delhi - Gurgaon Expressway which
handles about 220,000 vehicles/day has been considered. From the collected data, speed - flow
equations of different vehicle types on eight-lane divided urban expressways have been developed
using PARAMICS simulation software. The developed simulation model has replicated the ground
conditions realistically with the mean percentage error in speed ranging about 2 to 15% across all
vehicle types considering the heterogeneous traffic mix prevalent on Indian urban roads. The study
revealed that the roadway capacity of eight-lane divided urban expressways is about 11,435 passenger
car units/hour/direction. Further, the formation of virtual lanes by vehicles (that is, formation of eight
virtual lanes in just the one direction of eight-lane divided carriageway) has been extensively examined
and its impact on roadway capacity has been estimated. It was observed from this study that the
roadway capacity of eight-lane divided urban expressways under no virtual lanes condition (that is,
traffic plying on the demarcated four lane in one direction of eight-lane divided carriageway) would
obviously reduce the roadway capacity by about 15%. At the same time, it is to be noted that though the
imposition of restriction on virtual lanes decreases the average free speeds by 7%, it would enhance
the safety situation due to less vehicular interactions.
Key words: Free speed, speed-flow equations, roadway capacity, lane change behaviour, urban expressways.
INTRODUCTION
The sustained economic growth in India in recent years
has brought opportunities and challenges to the planning
and management of the Indian transportation system.
Like in other developing countries, the transportation
system in India is characterized by limited roadway
infrastructure and the lack of operation and management
experience. Among the most critical issues in highway
planning and management, is to determine the roadway
capacities of any inter-city highway/ urban expressway.
Though India has one of the largest road networks in the
world, the available road infrastructure and public
transport services in the metropolitan cities are highly
*Corresponding author. E-mail: errampallim711@yahoo.com.
inadequate for achieving faster movement of commuters
in comparison with the situation prevalent in the cities of
the developed world. The major urban arterials
connecting important location within metropolitan cities
are in need of capacity augmentation, pavement
strengthening, rehabilitation of bridges, improvement of
riding quality, provision of traffic safety measures, etc.
The aforementioned gross inadequacies in the existing
inter-city and urban road network coupled with
congestion caused by heterogeneity in traffic mix is
contributing to huge economic losses in terms of high
road user cost (RUC) which is also contributing to high
rate of road accidents. Realising the present
shortcomings in the transport sector, the Government of
India has initiated massive construction programmes of
highways linking major cities / activity centres and urban
expressways. Moreover, the automobile industry and
road design standards in India have also undergone
tremendous changes in the recent decade. Therefore, it
was considered necessary to take a look at the changing
trends of prevailing speed - flow characteristics
considering the emerging urban multi-lane highways/
expressways in the metropolitan cities.
In this paper, an attempt has been made for the first
time to explicitly study the free speed and speed - flow
characteristics on eight-lane divided urban expressways
in plain terrain. In order to assess these characteristics,
speed under free flow and congested conditions coupled
with traffic flow data was extensively collected on Delhi Gurgoan Expressway. From the collected data, speed flow equations have been evolved for different vehicle
types based on microscopic simulation models developed
in PARAMICS software environment. Subsequently,
roadway capacity for eight-lane divided urban
expressways is evolved with reasonable degree of
authenticity under the prevailing heterogeneous traffic
conditions. The impact of typical Indian driving behavior,
that is, how the lane discipline (formation of virtual lanes)
and lane change behavior affects roadway capacity on
eight-lane expressways has been assessed through
microscopic simulation approach.
CAPACITY OF MULTI-LANE HIGHWAYS
Earlier studies
The studies accomplished for multi-lane highways are
discussed here. As such, the determination of highway
capacity is one of the most important applications of any
traffic theory (Kerner, 2004). Some previous theories and
empirical researches focused on the interrelationships
among the influence of capacity, traffic features,
geometric elements, environmental conditions and
temporal weather factors on interrupted multi-lane
highways (Hoban, 1987; Iwasaki, 1991; Ibrahim and Hall,
1994; Shankar and Mannering, 1998). Many years of
research has led to the development of theories and
methodologies in roadway capacity analysis in the
developed countries. For example, the highway capacity
manual (HCM) developed in the United States of America
describes roadway capacity under ideal conditions and
then estimates practical capacities under prevailing
conditions in the field. US-HCM 2000 (TRB, 2000)
suggested that a maximum flow rate that can be
achieved on a multilane highway is 2200 Passenger Car
Units (PCU)/hour/lane. The Danish method is also a
modification of U.S. HCM to suit Danish conditions. The
adjustment factors in the Danish method cause a steeper
capacity reduction than in US-HCM 2000 as the
conditions become less ideal and therefore, the capacity
under ideal conditions on a four-lane highway is 2300
PCU/hour/lane on Denmark highways (Nielsen and
Jorgensen, 2008). Similarly, in Finland and Norway too,
US-HCM 2000 (TRB, 2000) was followed with minor
modifications to suit the local conditions and the roadway
capacities obtained by the Finnish and Norwegian
methods for multi-lane highways is 2000 PCU/hour/lane.
The structure of the Swedish method is similar to the USHCM (1995) and it uses the 1995 HCM adjustment
factors for the roadway width, whereas other adjustments
factors are mostly omitted. Consequently, the Swedish
method yielded higher capacity estimates and the
estimated capacity of four-lane divided highways was
4200 PCUs/h per direction (Luttinen and Innamaa, 2000).
The Australian method for analysis of capacity was
basically same as that of US-HCM method with the basic
difference being additional modification has been
suggested for specific problems. Under ideal conditions,
the average minimum headway of 1.8 s was considered
and maximum flow of 2000 vehicles per hour per lane
was assumed. The succeeding paragraph focuses on the
roadway capacity evolved in Asian countries like
Indonesia and China for multi-lane highways wherein
largely heterogeneous traffic conditions as experienced
on Indian highways is witnessed.
Bang et al. (1997) in their study for establishing
Indonesia HCM mentioned travel speed (synonymous
with journey speed) as the main measure of performance
of road segments, since it is easy to understand and to
measure, and is an essential input to road user costs in
economic analysis. In this study, the capacity of multilane highways has been estimated as 2300 light vehicles
(LVs)/hour/lane for Indonesian multi-lane highways. In
the case of Chinese conditions, based on the field data
collected, VTI highway simulation model was calibrated
and validated and this model was used for the
determination of passenger car equivalents (PCE) and
speed-flow relationships for different terrain types in
parallel with multiple regression analysis of empirical
speed-flow data. The results showed that the free-flow
speeds of vehicles were substantially low and that the
roadway capacity was also marginally lower (2100 PCEs
per lane on four-lane divided carriageways) under
Chinese conditions as compared with the values obtained
for Indonesian multi-lane highways. Further, Yang and
Zhang (2005) have established based on their extensive
field survey of traffic flow on multi-lane highways in
Beijing and subsequent empirical model development,
that the average roadway capacity per hour per lane on
four-lane, six-lane and eight-lane divided carriageways is
2104, 1973 and 1848 PCUs respectively. This is unlike
HCM results obtained for many developed countries
which prescribe that average capacity per lane on
different highways is equal as they assume that highway
capacity is constantly proportional to the number of lanes
on multi-lane divided carriageways.
Based on the review of mentioned studies in both
developed and developing countries, it is obvious that the
roadway design and traffic control practices are mostly
country or region specific and hence cannot be simply
transferred to any country for direct applications. In this
context, it is to be noted that the roadway capacity and
the conditions for adjustment are vastly different on
Indian roadways as the local roadway design (that is,
lane width, curves and grades), vehicle size and more
importantly, traffic mix and behaviour of a driver
especially lane changing and lane discipline phenomenon
are entirely different. Further, since there is no systematic
approach to this problem, coupled by a lack of
fundamental data, the adjustment factors from say, the
US HCM 2000 (TRB, 2000) cannot be easily revised and
applied to Indian highways. This is because adherence to
lane discipline characterizes homogeneous traffic in the
developed nations whereas very loose lane discipline
describes heterogeneous traffic which is very much an
integral part of all roadways in India including multi-lane
highways. This is due to fast moving vehicle cars, goods
vehicles, motorized two wheelers sharing the same road
space with bicycles, farm tractors, tractor trailers and
other types of slow moving vehicles (like cycle rickshaw,
animal drawn vehicles, etc.) on the Indian traffic scene
accounting for varying proportion on multi-lane highways
depending on its geographical location. Ironically, most of
the
models
discussed
earlier
developed
for
homogeneous condition are not applicable for the
heterogeneous traffic prevalent on Indian roads.
Eventually, the first major research effort in India in this
direction was done as part of the RUCS-1982 (CRRI,
1982) and this was followed by URUCS-1992 (Kadiyali
and Associates, 1992) and URUCS-2001 (CRRI, 2001).
IRC-64 (1990) suggested the tentative design service
volume (DSV) of 40,000 PCUs for the four-lane divided
carriageway in plain terrain which is significantly lesser
than the values evolved in most of the developed and
developing countries and therefore the need was felt for
revisiting the DSV values evolved under IRC-64.
Consequently, many research studies (Kadiyali et al.,
1991; Tiwari et al., 2000; Velmurugan et al., 2002, 2004,
2009, 2010; Chandra and Kumar, 2003; Reddy et al.,
2003; Chandra, 2004; Errampalli et al., 2004, 2009; Dey,
2007) aimed at assessing the roadway capacity for
varying carriageway widths including single lane,
intermediate lane, two-lane bi-directional and four-lane
divided carriageway widths covering different terrains
have been carried out during the last two decades.
URUCS-2001 (CRRI, 2001) recommended tentative
roadway capacity of 70,000 to 90,000 PCUs/day for a
four-lane divided carriageway in plain terrain (1750 to
2250 PCU/hr/lane considering 10% peak hour flow).
Chandra and Kumar (2003) studied the effect of roadway
width on capacity under different volume capacity ratios
and varying proportions of vehicles. Shukla (2008)
studied the mixed traffic flow behavior on four-lane
divided highway for varying conditions of traffic volume
and shoulder and developed a simulation model for the
observed traffic flow to estimate roadway capacity under
these conditions. To understand the traffic flow behavior
on four-lane divided highways under mixed traffic
condition, the arrival pattern of vehicles, speed
characteristics, lateral placement of vehicles and
overtaking behavior was analyzed. Shukla (2008) further
reported that the roadway capacity of four-lane divided
carriageways as 4770 vehicles/hour in each direction is
estimated for ‘all cars’ situation. This exhaustive look at
the literature indicates that no substantial work has been
carried out for establishing the roadway capacity for
varying carriageway widths on multi-lane highways and
urban roads, and urban expressways for the
heterogeneous traffic mix prevalent on Indian highways
with reasonable degree of confidence; hence, this
research endeavor can be termed as a significant attempt
in this direction.
Need for microscopic simulation
The traditional capacity estimation methods assume
homogeneous conditions and lane discipline; however, it
is not applicable for Indian conditions. In such
circumstances, roadway capacities could be either
underestimated or overestimated. Capacity estimation
primarily depends on vehicular movements on the road
stretch and in this regard, lane change behaviour and
lane discipline can severely affect the movements. On
Indian roads, vehicles seldom observe lane discipline and
make their own virtual lanes instead of the demarcated
physical lanes. The conventional methods ignore vehicle
movements and interactions and these behaviours
cannot be explained which has great impact on speed flow relationships and capacity estimation. In the absence
of accounting for such situations, the output might be far
from reality. As described earlier, microscopic simulation
considers each and every vehicle movement on a
roadway and hence such a lane change behaviour and
vehicle interactions can be described. More realistic
estimation of speed - flow relationships can be achieved
through microscopic simulation system which can lead to
the estimation of capacity with reasonable degree of
accuracy. Such microscopic simulations are able to
model individual vehicles and pedestrians in a large area
and it is possible to estimate realistic speed - flow
characteristics and capacity considering all possible lane
change behaviour even under heterogeneous traffic
conditions. Further, these techniques are highly useful in
estimating the traffic characteristics under different traffic
flow and driver behaviour conditions which cannot be
observed on the field. However, it is to be borne in mind
that the data collection task needed to develop the
microscopic simulation can be a bit tedious and
cumbersome. To arrive at speed - flow characteristics
and establish capacity norms through microscopic
simulation, one has to model the flow of individual
vehicles in a detailed manner for which established
Table 1. Details of selected test sections for traffic data collection on Delhi - Gurgaon expressway.
Test section
number
1
2
3
Location
Direction
Km 26.0 (near
IFFCO Intersection)
Km 26.0 (near
IFFCO Intersection)
Km 15.0 (near
Mahipalpur)
Gurgaon to
Delhi
Delhi to
Gurgaon
Gurgaon to
Delhi
Number of lanes
in each direction
Carriageway width
(m) in one direction
Time (h)
Trap length
(m)
4
16.5*
06:40 - 10:00
130
4
16.5*
06:40 - 10:00
130
4
16.5*
15:40 - 19:00
130
* including 2.5 m wide paved shoulder.
Speed (v)
Vf
Uncongested
Capacity
VCap
Congested/ Queuing
FCap
Flow (F)
Figure 1. Uncongested and congested parts of speed-flow curve.
simulation packages can be used. The data collection
and methodology followed for this phase is explained
further.
MATERIALS AND METHODS
Data collection
As mentioned earlier, the primary objective of the present study is
to develop realistic speed - flow equations for estimating the
roadway capacity of eight-lane divided urban expressway. In order
to achieve the envisaged objective, speed - flow studies were
conducted on fully access controlled Delhi-Gurgaon Expressway.
The videography survey was conducted by capturing the traffic
plying during the morning and evening time periods on both
directions of travel. The details of the test sections are given in
Table 1.
shown in Figure 1 (Yao et al., 2009). The two segments encompass
the following:
i. Uncongested (Upper Part): Traffic related to uncongested and
queue discharge states
ii. Congested (Lower Part): Traffic related to queuing state (Stop
and Go)
It is proposed to analyze these two parts separately and determine
the speed - flow relationships accordingly in the present study.
Different models including linear, exponential, polynomial,
logarithmic, power, Akcelik and Bureau of Public Roads (BPR) are
available for speed-flow analysis as given in Table 2. However,
linear models are adopted for the sake of simplicity in
understanding these relationships between speed and flow for both
upper part (uncongested) and lower part (congested) of the curve in
the present study. Using this procedure, the speed - flow equations
for different vehicle types on eight-lane divided urban expressway
are developed. In this study, the roadway capacity is considered as
the intersecting point of best speed - flow models developed for the
upper and lower part as shown in Figure 2.
Adopted methodology
Speed-flow curves
Framework for microscopic simulation model
In the present study, the traffic flow data was proposed to be
analyzed by typically dividing the traffic volume into two segments
corresponding to congested and uncongested traffic conditions as
One aspect kept in mind while applying traffic simulation technique
is that there are no universally accepted procedures for conducting
the calibration and validation of any road network. The
Table 2. Functional form of candidate models for speed-flow curves.
Name of the Equation
Linear
Logarithmic
Exponential
Power
Polynomial
Bureau of Public Roads (BPR)
Akcelik
Functional Form
v = -α x + β
v = -α ln x+β
v = α vf exp(-βx)
v = α /xβ
2
v = -αx –βx + γ
β
v = vf/(1+α(x) )
2
V=L/[L/vf+0.25{(x-1) + SQRT{(x-1) +αx}}]
Comments
Not always advisable; Reaches zero speed at high F/Fcap
Not always advisable; Has no value at x = 0.
Has all the required traits for equilibrium assignment
Not always advisable; It goes to infinity at F/Fcap at x = 0.
Not always advisable; It reaches zero speed at high F/Fcap
Has all the required traits for equilibrium assignment
Has all the required traits for equilibrium assignment.
Speed (Kmph)
v = speed; α, β and γ = global parameters for equation; x = F/Fcap ratio; vf = free - flow speed; F = flow; Fcap = flow at capacity; L = link length.
Flow (PCU/h)
Figure 2. Capacity estimation from speed-flow curves.
responsibility lies with the modeler to implement a suitable
procedure which provides an acceptable level of confidence in the
model results. The methodology followed for the microscopic
simulation is shown in the form of flow chart in Figure 3.
From Figure 3, it can be observed that the data collection is the
first and foremost requirement for understanding speed-flow
characteristics on urban expressways. To capture lane change
behavior on these high speed corridors, videography method was
adopted for data collection. The input parameters considered,
adopted network design along with traffic demand data used for the
development of simulation model coupled with calibration and
validation processes are described in detail in the subsequent
sections. During data collection at Mahipalpur section of Delhi Gurgaon Expressway, it was observed that vehicles are plying on
seven virtual traffic lanes though the demarcated lanes available
are only four lanes in each direction of travel.
Due to this reason, the flow levels were abnormally high during
the peak hours hovering around 17,000 vehicles/h. Accordingly, in
the simulation run also, it was decided to consider seven virtual
lanes for traffic movement matching with the observed ground
conditions despite the fact that the available road width is only 14 m
which can cater ideally to four lanes (measuring 3.75 m lane width)
in each direction of travel. However, the observed pattern of traffic
movement (as witnessed during video data decoding) deteriorated
even further during the evening peak hours as the traffic movement
was observed to ply on eight virtual lanes as shown in Figure 4.
Consequently, the observed ground conditions were replicated in
the simulation process by resorting to eight virtual lanes for the last
one hour of simulation. To analyze speed against flow and develop
equations, estimated speed values were considered for all the
vehicle types under congested flow conditions to plot the scatter
diagrams of the speed of vehicles and the corresponding flows.
Input parameters
In this study, the first step in the calibration and validation process
involved choosing of suitable model parameters like vehicle
characteristics, aggressiveness, awareness, target headways and
reaction times that provided realistic results. The range of suitable
values for these parameters has been established through
calibration of the PARAMICS model. Most of these were chosen
based on previous experience in using the model in similar urban
conditions. Much emphasis was laid on comparison of modeled and
observed flows as well as classified vehicle speeds as described
further.
Data Collection
Video Recording
Method
Data Extraction and
Analysis
i.
ii.
Traffic Flow
Vehicle Speed
iii. Driver Characteristics
iv. Vehicle Characteristics
v. Road Network
Preparation of Input Parameters for
Microscopic Simulation
Development of Microscopic
Traffic Simulation Model in
PARAMICS Software
Lane Change Behaviour
Calibration and Validation of
Simulation model
Estimate Output: Speed and Flow
Development of Speed Flow Model
Estimation of Capacity under different
Scenarios of lane discipline and lane change
behaviour
Figure 3. Methodology for estimating speed-flow equations and capacity of urban expressways.
Road network design
The link-node diagram is the blueprint for constructing the
microscopic simulation model. The diagram identifies which streets
and highways will be included in the model and how they will be
represented. The link-node diagram was created through the
PARAMICS Modeller suite. Nodes are placed in the model using xy coordinates. Links were used as connectors between these
nodes. A test link of 300 m has been created. This is to create a
buffer distance for the vehicles before entering the marked section
(130 m). This has been done to ensure that generated vehicles
reach a stabilized flow state before entering the section. Figure 5
shows the schematic representation of the test network created in
PARAMICS. The physical and operational characteristics of the
links are given as input into the model which included category of
the road (major highway), number of lanes (4), lane width (3.75 m)
and lane speed (90 kmph).
130 m are placed in the link in both directions which replicate the
trap length in real case as shown in Figure 5. These detectors
deduce the number of classified vehicles and speeds of different
classes of vehicles are used for the validation purpose.
Traffic demand data
Traffic demands are defined as the number of vehicles and the
percentage of vehicles of each type that traverse the study area
during the simulation time period. Furthermore, it may be necessary
to reflect the variation in demand throughout the simulation time
period. As discussed earlier, nine vehicle types were considered for
simulation modelling. The configuration of these vehicles like
physical attributes, acceleration/deceleration rate, proportion, etc.,
severely affects the capacity of the roadway.
Model calibration
Traffic operations data
Traffic operations and management data for links consist of warning
data (incidents, lane drops, etc.), regulatory data (speed limits,
variable speed limits, high-occupancy vehicles (HOVs), highoccupancy toll (HOT), lane use, etc., information (guidance) data
(dynamic message signs and roadside beacons) and surveillance
detectors (type and location). In this study, loop detectors of length
Calibration is the adjustment of model parameters to improve the
model’s ability to reproduce local driver behavior and traffic
performance characteristics. Calibration is performed on various
components of the overall model. Calibration is necessary because
no single model can be expected to be equally accurate for all
possible traffic conditions. Even the most detailed microscopic
simulation model still contains only a portion of all of the variables
1
2
3
4 5 6 7
8
Figure 4. Vehicles travelling in eight virtual lanes on the study section during peak hours.
Figure 5. Designed network in PARAMICS with two zones and one link with two directions.
that affect real-world traffic conditions. Since no single model can
include the whole universe of variables, every model must be
adapted to local conditions. Every microscopic simulation software
program comes with a set of user-adjustable parameters for the
purpose of calibrating the model to local conditions. Therefore, the
objective of calibration is to find the set of parameter values for the
model that best reproduces local traffic conditions. During the
calibration phase, the global capacity-related parameters in the
simulation model were adjusted to best replicate local field
measurements such as speed and volume and thus arrive at the
possible capacity using the calibrated model after subjecting the
model to validation process.
Table 3. PCU factors adopted based on IRC specifications (IRC: 641990).
Time
Section 3
2000
18:50 - 18:55
18:40 - 18:45
18:30 - 18:35
18:20 - 18:25
18:10 - 18:15
18:00 - 18:05
17:50 - 17:55
17:40 - 17:45
17:30 - 17:35
17:20 - 17:25
0
17:10 - 17:15
09:50 - 09:55
09:40 - 09:45
09:30 - 09:35
09:20 - 09:25
09:10 - 09:15
08:50 - 08:55
08:40 - 08:45
08:30 - 08:35
08:20 - 08:25
08:10 - 08:15
08:00 - 08:05
07:50 - 07:55
07:40 - 07:45
07:30 - 07:35
07:20 - 07:25
07:10 - 07:15
07:00 - 07:05
06:50 - 06:55
09:00 - 09:05
Section 2
0
4000
17:00 - 17:05
2000
6000
16:50 - 16:55
Section 1
8000
16:40 - 16:45
4000
10000
16:30 - 16:35
6000
12000
16:20 - 16:25
8000
14000
16:10 - 16:15
10000
16000
16:00 - 16:05
12000
18000
15:50 - 15:55
14000
PCU factor
0.5
1.0
1.5
1.0
0.5
3.0
3.0
1.0
3.0
3.0
15:40 - 15:45
16000
06:40 - 06:45
Traffic volume (PCU/h)
18000
Traffic volume (PCU/h)
Vehicle type
Motor cycles (MC) and scooters (SC)
Auto rickshaws (A)
Cycle rick and other slow vehicles (OT)
Small cars (≤1400 cc) (CS) and big cars (CB)
Cycles (CY)
Buses (B) and mini buses (MB)
Tractors and tractor trailers (TT)
Light commercial vehicles (LCV)
Two-axle commercial vehicles (HCV)
Multi-axle commercial vehicles (MCV)
Time
Figure 6. Variation of traffic volume on different test sections.
RESULTS AND DISCUSSION
Traffic data analysis
The speed and flow data was analyzed by classifying the
vehicles into the following categories namely, cars which is further sub-classified into two categories, small
cars (engine capacity ≤ 1400 cc) and big cars (engine
capacity > 1400 cc); two wheelers (TW); auto rickshaws
(Autos); buses, light commercial vehicles (LCV); two-axle
heavy commercial vehicles (HCV) and multi-axle heavy
commercial vehicles (MCV). As mentioned earlier, the
recorded video was replayed in the laboratory for
decoding purposes and the required data were decoded
through manual method. The vehicles were categorized
into the afore said categories and the primary data
extracted from video recording were volume and space
mean speed of individual vehicles during every five
minute time interval. The extracted data was analyzed to
get the classified traffic volume count and was used for
deriving speed - flow relationship for each category of
vehicle and thus estimate the capacity. The passenger
car unit (PCU) values as per IRC:64-1990 as given in
Table 3 were applied to convert traffic volume of different
vehicle types into single unit. The traffic volume,
composition and average speed for all the three sections
are shown in Figures 6, 7 and 8 respectively. A close look
at Figure 6 illustrates that the traffic has substantially
increased to reach morning peak at 08:30 h as the traffic
flow is hovering between 7,000 and 10,000 PCU/h at
Section 1 and 2 respectively on Delhi - Gurgaon
Expressway whereas it caters to the evening peak traffic
of about 17,000 PCU/h at 18:35 h on Section 3 of the
Delhi - Gurgaon Expressway. Figure 7 demonstrates that
the average speed on Section 1 and 2 is almost varying
between 50 and 60 kmph whereas it is mostly varying
between 25 and 50 kmph on Section 3 during the data
collection period which reduces drastically to less than 20
kmph during the evening peak hour as depicted in Figure
7. Figure 8 reveals that the cars and two wheelers
obviously dominate the traffic flow ranging from 92 to
97% of the total volume on both the directions. This
scenario clearly describes the dominance of personalised
cars and two wheelers on this road section thus
Buses
2%
Autos
0.20%
Section 1
LCV
1%
Two wheelers
21%
Buses
2%
Autos
2%
highlighting the urbanized nature of the test sections.
LCV
3%
Sma ll Cars
38%
HCV
0.27% MCV
0.41%
Big Cars
38%
HCV
0.35%
Figure 8. Observed traffic composition on different test sections.
Two wheelers
32%
Big Cars
29%
Autos Buses
0.17% 3%
Section 3
MCV
0.25% Cycles
1%
LCV HCV
3% 1%
Two wheelers
18%
Big Ca rs
34%
Small Cars
31%
Time
Figure 7. Variation of average speeds on different test sections.
Section 2
MCV
1%
Small Ca rs
40%
18:50 - 18:55
18:40 - 18:45
18:30 - 18:35
18:20 - 18:25
18:10 - 18:15
18:00 - 18:05
90
80
70
60
50
40
30
20
10
0
17:50 - 17:55
17:40 - 17:45
17:30 - 17:35
17:20 - 17:25
17:10 - 17:15
17:00 - 17:05
16:50 - 16:55
16:40 - 16:45
16:30 - 16:35
16:20 - 16:25
16:10 - 16:15
16:00 - 16:05
Average Speed (kmph)
Section 1
Section 2
15:50 - 15:55
15:40 - 15:45
09:50 - 09:55
09:40 - 09:45
09:30 - 09:35
09:20 - 09:25
09:10 - 09:15
09:00 - 09:05
08:50 - 08:55
08:40 - 08:45
08:30 - 08:35
08:20 - 08:25
08:10 - 08:15
08:00 - 08:05
07:50 - 07:55
07:40 - 07:45
07:30 - 07:35
07:20 - 07:25
07:10 - 07:15
07:00 - 07:05
06:50 - 06:55
06:40 - 06:45
Average speed (kmph)
90
80
70
60
50
40
30
20
10
0
Section 3
1400
Observed
Average Error =4%
RMSE Value = 23 Veh/5min
1200
Simulated
Observed
Average Error =10%
RMSE Value = 116 Veh/5min
Simulated
1000
1000
800
600
400
200
800
600
400
200
Time
09:50 - 09:55
09:40 - 09:45
09:30 - 09:35
09:20 - 09:25
09:10 - 09:15
09:00 - 09:05
08:50 - 08:55
08:40 - 08:45
08:30 - 08:35
08:20 - 08:25
Time
Section 3
1600
Traffic Volume (Veh/5min)
08:10 - 08:15
08:00 - 08:05
07:50 - 07:55
07:40 - 07:45
07:30 - 07:35
07:00 - 07:05
09:50 - 09:55
09:40 - 09:45
09:30 - 09:35
09:20 - 09:25
09:10 - 09:15
09:00 - 09:05
08:50 - 08:55
08:40 - 08:45
08:30 - 08:35
08:20 - 08:25
08:10 - 08:15
08:00 - 08:05
07:50 - 07:55
07:40 - 07:45
07:30 - 07:35
07:20 - 07:25
07:10 - 07:15
07:00 - 07:05
07:20 - 07:25
0
0
07:10 - 07:15
1200
Traffic Volume (Veh/5min)
Traffic Volume (Veh/5min)
1400
Section 2
1600
Section 1
1600
Observed
Average Error =5%
RMSE Value = 98 Veh/5min
1400
1200
Simula ted
1000
800
600
400
200
06:50 - 06:55
06:40 - 06:45
06:30 - 06:35
06:20 - 06:25
06:10 - 06:15
06:00 - 06:05
05:50 - 05:55
05:40 - 05:45
05:30 - 05:35
05:20 - 05:25
05:10 - 05:15
05:00 - 05:05
04:50 - 04:55
04:40 - 04:45
04:30 - 04:35
04:20 - 04:25
04:10 - 04:15
04:00 - 04:05
0
Time
Figure 9. Comparison of observed and simulated traffic flows on the study sections.
Validation results of simulation model
The calibrated microscopic simulation model is validated
by comparing the observed and simulated values. Figure
9 shows the comparison between observed and
simulated volumes for Sections 1, 2 and 3 on Delhi Gurgaon Urban Expressway. The average percentage of
error and Root Mean Square Error (RMSE) were
determined by comparing the observed and simulated
flows and shown in Figure 9.
From Figure 9, it can be seen that developed
simulation model is reasonably representing the ground
conditions in terms of number of vehicles simulated
during each five minute time period as the average error
is ranging between 4 and 10% and RMSE value is
hovering between 23 to 116 Vehicels/5 min. To validate
the developed simulation model, the average speeds of
the different vehicles were also compared for different
sections of Delhi - Gurgaon Urban Expressway as given
in Table 4. From Table 4, it can be observed that error
between observed and simulated speeds of all the
vehicles is ranging between 19 to 27%. From these
results, it can be inferred that the mean percentage error
and RMSE of different vehicle types for both speed and
volume are quite reasonable indicating that the
developed simulation model is able to predict vehicular
behavior with fair degree of accuracy.
Speed - flow equations and roadway capacity
through simulation model
Using the developed simulation model, the speed data for
different vehicle is estimated under different traffic
volume conditions for eight-lane divided urban
expressway and used to arrive at the speed-flow
Table 4. Error between average observed and simulated speeds on the study sections.
relationships for different classes of vehicles. The speed
estimation from the data of IFFCO intersection (Sections
1 and 2) and Mahipalpur (Section 3) of Delhi - Gurgaon
Expressway has been done using PARAMICS and the
scatter diagrams of the speed of vehicles and the
corresponding flows were plotted. After critical
observation of the data, it was decided to split these data
into two parts: congested and uncongested parts of the
speed - flow relationships. The data of IFFCO
intersection (Sections 1 and 2) was taken for the upper
part of the speed-flow relationship area (that is,
uncongested) since the data observed during the offpeak morning hours while the Mahipalpur data (Section
3) was taken for the lower part (that is, congested). On
the basis of this, these two data are considered
separately and separate speed - flow equations were
developed for uncongested and congested areas. In case
of congested conditions, the stop and go scenarios were
introduced to get the data spread into all the regions of
congested. The estimated speed in relation with the flow
conditions are pictorially shown in Figure 10 for different
vehicle types. The developed speed-flow relationships for
different vehicle types plying on eight-lane divided urban
expressway are shown in Table 5.
From Figure 10a and b, it can be clearly seen that R2
value is very high for two wheelers, cars, buses, LCV
2
whereas for the remaining vehicle type, R is moderate.
Based on this result, it can be said that the developed
speed-flow equations for the test section, that is, eightlane divided urban expressways are good in terms of
statistical validity and can be applied to predict speeds for
any given traffic volume data with reasonable amount of
Mean error (%)
RMSE (kmph)
12.87
18.9
13.14
12.4
12.3
12.3
19.06
21.85
16.52
Avg. simulated speed
(kmph)
19
30
18
16
17
19
26
29
25
RMSE (kmph)
71.28
60.78
80.00
78.61
67.56
63.65
63.15
57.58
66.49
Mean error (%)
59.90
46.75
67.80
67.77
57.75
53.49
50.12
44.64
53.19
Section 3
Avg. observed speed
(kmph)
10.84
18.95
8.7
7.61
10.67
12.93
19.88
32.9
17.03
Avg. simulated speed
(kmph)
14
32
11
8
14
19
29
49
27
Avg. observed speed
(kmph)
72.49
54.92
79.63
78.46
68.90
63.97
67.51
62.73
71.05
Section 2
RMSE (kmph)
63.59
41.60
71.74
72.64
60.43
53.76
52.33
42.10
55.94
Mean error (%)
Avg. simulated speed
(kmph)
Two wheeler
Autos
Small car
Big car
Bus
LCV
HCV
MCV
All vehicles
Avg. observed speed
(kmph)
Type of vehicle
Section 1
29.63
28.28
30.47
30.71
28.89
31.32
28.43
29.08
30.56
35.82
33.71
38.64
39.15
38.56
40.10
38.16
34.50
37.65
21
19
27
27
33
28
34
19
19
10.57
7.86
12.65
12.74
10.17
10.18
11.60
11.80
6.84
accuracy. Accordingly, the developed speed-flow
equations based on “all cars” were utilized to estimate
roadway capacity for eight-lane divided urban
expressway and the capacity is shown in Figure 11.
From Figure 11, it can be seen that the roadway
capacity was estimated to be 11,544 PCUs/h/direction for
the eight-lane divided urban expressways. To
demonstrate the validity of developed speed - flow
equations through microscopic simulation model, the
estimated free speed which is the intercept of the
equations (at flow almost equal to 0 vehicles/h) are
compared with the observed free speeds and presented
in Figure 12.
Figure 12 illustrated that the error between observed
and simulated free speeds is less than 10% and from
this, it can be concluded that developed speed - flow
equations through simulation model is able to predict the
traffic phenomenon on multi-lane highways more
realistically as the prediction error in free speeds on
eight-lane divided carriageway is less. Thus, the evolved
roadway capacity through simulation approach can be
adjudged to be realistic for the heterogeneous traffic
conditions observed on eight-lane divided urban
expressways.
Impact of virtual traffic lanes on roadway capacity
As discussed earlier, the Indian drivers always tend to
travel on urban corridors in more than designated
physical lanes. On Delhi - Gurgaon Expressway, vehicles
were traveling on 8 virtual lanes vis-à-vis is the
90
80
Speed (kmph)
70
60
Two Wheeler
80
y = -0.00509x + 90.57377
R² = 0.94809
congested
70
uncongested
50
40
30
Speed (kmph)
90
y = 0.00199x + 8.29522
R² = 0.83122
10
60
Uncongested
50
40
30
10
0
0
0
5000
10000
15000
Flow (PCU/h/Dir)
100
0
20000
100
90
Small Cars
y = -0.00609x + 103.39505
R² = 0.96412
uncongested
congested
100
5000
10000
15000
Flow (PCU/h/Dir)
60
uncongested
50
40
30
y = 0.00218x + 8.60328
R² = 0.85823
0
5000
10000
15000
Flow (PCU/h/Dir)
90
uncongested
50
40
30
70
20000
congested
20000
Bus
80
y = -0.00619x + 104.00000
R² = 0.96101
70
60
y = -0.00634x + 105.00000
R² = 0.95469
congested
10
0
All cars
90
80
20000
20
y = 0.00212x + 8.67417
R² = 0.87200
0
5000
10000
15000
Flow (PCU/h/Dir)
Big Cars
80
Speed (kmph)
90
80
70
60
50
40
30
20
10
0
y = -0.0045x + 84.006
R² = 0.7816
70
Speed (kmph)
Speed (kmph)
y = -0.003x + 70.000
R² = 0.412
20
20
Speed (kmph)
Auto
60
Uncongested
50
40
30
20
20
10
10
y = 0.00215x + 8.63710
R² = 0.86478
0
0
5000
10000
15000
Flow (PCU/h/Dir)
0
20000
0
5000
10000
15000
Flow (PCU/h/Dir)
20000
Figure 10a. Developed speed - flow relationships from calibrated simulation models.
availability of only 4 physical lanes in each direction
(Figure 4). This particular driving behavior bound to have
severe impact on speed, flow and roadway capacity and
moreover on the road safety as the interactions among
Table 5. Speed-flow equations for different vehicle types on eight-lane divided
urban expressways.
Vehicle type
Two wheelers
Speed - flow equations
Uncongested: y = -0.005x + 90.57
Congested: y = 0.002x + 8.295
R²
0.948
0.831
Autos
y = -0.00331x + 70.0
0.589
Small cars
Uncongested: y = -0.006x + 103.39
Congested: y = 0.002x + 8.674
0.964
0.872
Big cars
Uncongested: y = -0.006x + 105.00
Congested: y = 0.002x + 8.603
0.955
0.858
All cars
Uncongested: y = -0.00619x + 104.0
Congested: y = 0.00215x + 8.63710
0.961
0.860
Bus
LCV
HCV
MCV
y = -0.004x + 84.00
y = -0.006x + 86.86
y = -0.004x + 80.94
y = -0.004x + 75.00
0.781
0.828
0.547
0.418
All vehicles
Uncongested: y = -0.006x + 99.58
Congested: y = 0.002x + 8.177
0.962
0.872
y = speed (kmph); x = flow (PCU/hr/Dir).
the vehicles increases. To understand the impact of
virtual traffic lane driving resorted by the Indian drivers,
the roadway capacity of eight-lane divided expressway
has been estimated by restricting the vehicles to travel in
the available 4 physical lanes in each direction of travel
during the simulation run. However, lane changes were
allowed within the available 4 lanes. Further, the impact
of restriction of lane changes has been studied as well,
that is, vehicles are allowed to travel only on the 4
available physical lanes without resorting to any lane
change. These two scenarios are considered in the
simulation model and roadway capacities were estimated
under these behaviors. The details can be found
elsewhere (Velmurugan et al., 2010). Figure 13 shows
the typical comparison of roadway capacities and free
speeds of cars on eight-lane divided urban expressways
under varying scenarios.
From Figure 13, it can be observed that the restriction
of a road user to travel on available physical lanes on
eight-lane divided urban expressway would lead to a
reduction of roadway capacity (9796 PCU/h/direction) by
about 15%. It can also be observed that this contributes
to a reduction of about 7% in free speeds (cars) as well
because of the restriction of virtual lanes. From the figure,
it can also be observed that the restriction of lane change
would lead to further reduction of 9% in roadway capacity
(8941 PCU/h/direction); however marginal increment in
speed (about 1%) can be observed. Since the formation
of virtual lanes has obviously increased the roadway
capacity as well as free speeds, the road users are
inclined to resort to the formation of virtual lane under the
heterogeneous traffic conditions prevalent on Indian
roads. However, the restriction of such behavior would
certainly enhance the road safety but at the cost of
marginal reduction in roadway capacity.
Conclusions
In this study, free speed profiles and speed - flow
equations of different vehicle types for eight-lane divided
urban expressways has been established for the first time
in the country based on microscopic simulation models
and subsequently roadway capacity has been estimated
with reasonable degree of authenticity for the prevailing
heterogeneous traffic conditions. Further, the lane
change behaviour of different vehicle types has been
extensively studied and its impact on roadway capacity
has been critically evaluated on these eight-lane divided
urban expressways. The conclusions drawn from the
studies are summarized as follows:
i. The maiden attempt to segregating the speed - flow
data into uncongested and congested traffic conditions
90
90
LCV
80
y = -0.005x + 86.868
R² = 0.8286
60
Uncongested
50
40
30
60
Uncongested
50
40
30
20
20
10
10
0
0
0
90
5000
10000
15000
Flow (PCU/h/Dir)
0
20000
100
MCV
y = -0.004x + 75.000
R² = 0.418
60
Uncongested
50
40
30
70
60
congested
50
uncongested
40
30
20
20
10
10
0
y = 0.00211x + 8.17768
R² = 0.87259
0
0
5000
10000
15000
Flow (PCU/h/Dir)
20000
20000
y = -0.00581x + 99.57722
R² = 0.96205
80
Speed (kmph)
70
5000
10000
15000
Flow (PCU/h/Dir)
All Vehicles
90
80
Speed (kmph)
y = -0.0049x + 80.94
R² = 0.5478
70
Speed (kmph)
Speed (kmph)
70
HCV
80
0
5000
10000
15000
Flow (PCU/h/Dir)
20000
Figure 10b. Developed speed - flow relationships from calibrated simulation models.
has been successfully accomplished in the present study
and subsequently roadway capacities has been
estimated through simulation.
ii. The study revealed that the roadway capacities
estimated through microscopic simulation approach
(2859 PCU/Hr/Lane) in this study has replicated ground
conditions more realistically for eight-lane divided urban
expressways.
iii. The present study affirms that the adoption (through
developing adjustment factors) of the roadway capacities
determined for developed world scenarios would not yield
realistic results.
iv. The simulation study carried out in this paper has
substantiated the fact that the lane change behavior
significantly influences the vehicular movements on a
road section and thus affects the roadway capacity.
It can be observed from this study that the restriction of a
road user to travel on available physical lanes on eightlane divided urban expressway would lead to a reduction
of roadway capacity (9796 PCU/Hr/Direction) by about
15%. It can also be observed that there would be a
reduction of about 7% in free speed (cars) as well
because of the restriction of virtual lanes. From these
results, it can also be observed that the restriction of lane
change would lead to further reduction of 9% in roadway
capacity (8941 PCU/Hr/Direction); however, marginal
increment in speed (about 1%) can be observed. Since
the formation of virtual lanes would obviously increase
the roadway capacity as well as free speeds, the road
users are inclined to resort to the formation of virtual lane
under the heterogeneous traffic conditions prevalent on
Indian roads. However, the restriction of such behavior
100
Eight-Lane Divided Urban Expressway
90
Speed (kmph)
80
y = -0.00619x + 104.00000
R² = 0.96101
70
60
50
40
30
Capacity = 11435
PCU/Hr/Dir
20
y = 0.00215x + 8.63710
R² = 0.86478
10
0
0
2000
4000
6000
8000
10000
Flow (PCU/h/Dir)
12000
14000
16000
Figure 11. Roadway capacity of eight-lane divided urban expressways evolved through microscopic simulation
model.
120
Observed Free Speed
Speed (kmph)
100
Simulated Free Speed
80
60
40
20
MCV
HCV
LCV
Bus
All Cars
Big Car
Small car
Auto
Two
Wheeler
0
Figure 12. Comparison between observed and simulated free speeds.
would certainly enhance the road safety but at the cost of
marginal reduction in roadway capacity.
LIMITATIONS AND FUTURE SCOPE
It may be noted that the driver behaviour has been
calibrated for the parameters defined through the
PARAMICS software taking into account the traffic flow
and road conditions prevalent on typical sections of eightlane divided urban expressway located on plain terrain
considering the heterogeneous traffic conditions and the
lane change behaviour as observed in the field. Though
the error seems to be reasonable, it is proposed to
Roadway Capacity
Free Speed
106
11435
12000
104
9796
10000
102
8941
8000
6000
4000
Free Speed (kmph)
Roadway Capacity (PCU/h/Dir)
14000
100
98
96
94
2000
92
90
0
With Virtual Lanes
No Virtual Lanes
No Virtual Lanes and No
Lane Change
With Virtual Lanes
No Virtual Lanes
No Virtual Lanes and
No Lane Change
Figure 13. Typical comparison of roadway capacity and free speed of cars on eight-lane divided urban expressways for different driving
behaviour.
consider additional test sections so as to further improve
the predictive capability of the developed simulation
model. Considering these inherent limitations in this
study, the geographical transferability of the developed
simulation model developed in this study can be tested
through collection of traffic data on other eight-lane
divided urban expressways/ carriageways located on
plain terrain road sections spread over the country.
ACKNOWLEDGEMENTS
Authors are thankful to Dr. S. Gangopadhyay, Director,
CRRI, India, for his kind support and encouragement for
allowing this paper for publication. The technical services
rendered by Mr. Anand Kumar Srivatsava and Mr. Fida
Hussain during the field studies of this study are
gratefully acknowledged.
REFERENCES
Bang KL (1997). “Indonesian Highway Capacity Manual.” Department of
Public Works, Directorate General Highways, Jakarta, Indonesia.
Bang KL, Bergh T, Carlsson A, Hansson A, Ronggui Z (1999). “Highway
Capacity Study, Final Report and Capacity Guidelines.” National
Highway Project of Peoples Republic of China, Hebei and Henan
Provincial Communications Departments.
Bång KL, Heshen A (2000). “Development of capacity guidelines for
road links and intersections for Henan and Hebei Provinces, PRC.”
th
Proc. 4
International Symposium on Highway Capacity,
Transportation Research E-Circular, E-C018, ISSN 0097-8515, pp.
287-298.
Chandra S (2004). “Capacity estimation procedure for two- lane roads
under mixed traffic conditions.” J. Indian. Road. Congress, Vol.65- 1,
pp.139-170.
Chandra S, Kumar U (2003). “Effect of lane width on capacity under
mixed traffic conditions in India.” J. Transportation. Eng., ASCE, pp.
155-160.
CRRI (1982). “Road User Cost in India.” A report submitted to Ministry
of Surface Transport, Central Road Research Institute (CRRI), New
Delhi.
CRRI (2001). “Updation of Road User Cost Data Volume-I and II.” A
report submitted to Ministry of Surface Transport, Central Road
Research Institute (CRRI), New Delhi.
Dey PP, Chandra S, Gangopadhyay S (2006). “Speed distribution
curves under mixed traffic conditions ASCE. J. Transportation.
Eng.,132(6): 475-481.
Dey PP, Chandra S, Gangopadhyay S (2007). “PCU factors for two lane
roads.” HRB, Indian Road Congress, 77: 111-119.
Errampalli M, Reddy TS, Velmurugan S (2004). “Updation of free speed
models through mechanistic principles for Indian conditions. J.
Advanced. Transportation., 38(2): 133-146.
Errampalli M, Velmurugan S, Ravinder K, Gangopadhyay S (2009).
“Speed - flow characteristics of high speed corridors in India.”
th
Presented at 14 International Conference of Hong Kong Society for
Transportation Studies (HKSTS), Hong Kong.
Tiwari G, Fazio J, Pavitravas S (2000). “Passenger car units for
th
heterogeneous traffic using a modified density method.” Proc. 4
International Symposium on Highway Capacity, Transportation
Research E-Circular, E-C018, ISSN 0097-8515, pp. 246-257.
Hoban CJ (1987). “Evaluating traffic capacity and improvements to road
geometry.” World Bank, Washington, DC.
Ibrahim AT, Hall FL (1994). “Effect of adverse weather conditions on
speed-flow-occupancy relationships.” Transportation Research
Record TRB, 1457: 184-191.
IRC (1990). “Guidelines for Capacity of Roads in Rural Areas.” IRC-64,
Indian Roads Congress, New Delhi.
Iwasaki M (1991). “Empirical analysis of congested traffic flow
characteristics and free speed affected by geometric factors on an
intercity expressway.” Transportation Research Record, TRB. 1320:
242-250.
Kadiyali and Associates. (1992). “Study for Updating Road User Cost
Data”, A Report submitted to Ministry of Surface Transport and Asian
Development Bank, Dr. L. R. Kadiyali and Associates, New Delhi.
Kadiyali LR, Lal NB, Satyanarayana M, Swaminathan AK (1991).
“Speed-flow characteristics on Indian highways” J. Indian Roads
Congress, 52(2): 233-251.
Kerner BS (2004). “Three phase traffic theory and highway capacity.”
Physica A, 333: 379 - 440.
Luttinen K, Innamaa H (2000). “Traffic flow simulation for an inter-city
th
freeway corridor.” Proc. 4 International Symposium on Highway
Capacity, Transportation Research E-Circular, E-C018, ISSN 00978515.
Nielsen OA, Jorgensen RM (2008). “Estimation of speed - flow and flow
- density relations on the motorway network in the Greater
Copenhagen Region.” 120/IET Intelligent. doi:10.1049/ ietits:20070024. Transp. Syst., 2: 120-131.
Yi P, Zhang Y, Lu J, Lu H (2004). “Safety - based capacity analysis for
chinese highways - a preliminary study.” IATSS Res., 28(1): 47 - 55.
Reddy TS, Velmurugan S, Errampalli M, Ramalingaiah AN (2003).
“Updation of road user cost and revised software for evaluation of
highway projects.” J. Indian Roads Congress, 64(2): 207-267.
Shukla S (2008). “Traffic flow modelling on multilane divided highways.”
PhD Thesis submitted to IIT, Roorkee.
Shankar V, Mannering F (1998).“Modeling the endogeneity of lanemean speeds and lane-speed deviations: a structural equations
approach.” Transp. Res., 32: 311-322.
rd
TRB (2000). “Highway Capacity Manual 2000.” 3 Ed., National
Research Council, Transportation Research Board, Washington, D.C.
Velmurugan S, Errampalli M, Reddy TS (2002). “Changing operating
speeds on rural highways.” Proc. Emerging Trends in Road Transport
(RORTRAN), , IIT, Kharagpur, India, 2(5): 29-5.40.
Velmurugan S, Errampalli M, Reddy TS (2004). “Evaluation of speed th
flow characteristics on Indian highways.” Proc. 10 World Conference
on Transport Research (WCTR), CD-ROM, Istanbul, Turkey.
Velmurugan S, Errampalli M, Ravinder K, Gangopadhyay S (2009).
“Updation of road user cost for economic evaluation of road projects.”
Indian. J. Traffic. Manage., 33(3): 205-225.
Velmurugan S, Errampalli M, Ravinder K, Sitaramanjeneyulu K,
Gangopadhyay S (2010). “Critical evaluation of roadway capacity of
multi-lane highways through traditional and microscopic simulation
models under heterogenous traffic conditions.” J. Indian Roads
Congress, 33(3): 205-225.
Yang X, Zhang N (2005). “The marginal decrease of lane capacity with
the number of lanes on highway.” Proc. International Conference of
Eastern Asia Society for Transportation Studies (EASTS),5: 739 749.
Yao J, Teng H, Wei H, Hu S (2009). “Estimating roadway capacity using
the simultaneous spline regression model.” J. Transp. Syst. Eng.
Inform. Technol., 9(1): 87-98.