Effect of Road Width and Traf Fic Volume On Vehicular Interactions in Heterogeneous Traf Fic

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

JOURNAL OF ADVANCED TRANSPORTATION

J. Adv. Transp. 2014; 48:1–14


Published online 2 March 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/atr.196

Effect of road width and traffic volume on vehicular interactions in


heterogeneous traffic

Karuppanagounder Krishnamurthy1* and Venkatachalam Thamizh Arasan2


1
Department of Civil Engineering, National Institute of Technology Calicut, Calicut, Kerala 673601, India
2
Department of Civil Engineering, IIT Madras, Chennai, Tamil Nadu, India

SUMMARY
Highway traffic flow phenomena involve several complex and stochastic variables with high interdependen-
cies. The variations in roadway, traffic and environmental factors influence the traffic flow quality signifi-
cantly. Capacity analysis of road sections under different traffic and geometric conditions need to quantify
the vehicles of widely varying characteristics to a common and universally acceptable unit. Passenger car unit
(PCU) is the universally adopted unit of traffic volume, keeping the passenger car as the ‘standard vehicle’
with reference to its static and dynamic characteristics; other vehicles are expressed to its equivalent number
in terms of PCUs. The studies carried out in this aspect represent the dynamic nature of impedance caused by
a vehicle while moving through a traffic stream. The PCU values recommended by the Highway Capacity
Manual are widely applied in many countries; however, their applicability is highly under debate because
of the variations in prevailing local traffic conditions. There are several factors that influence the PCU values
such as traffic, roadway, vehicle, environmental and control conditions, etc. Apart from vehicular character-
istics, the other two major factors that influence the PCU of vehicles are the following: (i) road width and
(ii) traffic volume. In this study, estimation of PCU values for the different types of vehicles of a highly
heterogeneous traffic on 7.5- and 11.0-m-wide roads, using micro-simulation technique, has been dealt with.
It has been found that the PCU value of a vehicle type varies significantly with variation in road width and
traffic volume. The results of the study indicate that the PCU values are significantly influenced by the said
two factors. Copyright © 2012 John Wiley & Sons, Ltd.

KEY WORDS: flow; headway; traffic safety capacity; transport planning; transportation analysis

1. INTRODUCTION

Road capacity, in general, can be understood as the maximum possible volume of traffic that can pass
through a given road section. Thus, the knowledge of traffic volume is an important basic input
required for accurate estimation of road capacity. Expressing traffic volume as the number of vehicles
passing a given section of road or traffic lane per unit time will be inappropriate when several types of
vehicles with widely varying static and dynamic characteristics are comprised in the traffic. The prob-
lem of measuring the volume of such heterogeneous traffic has been addressed by converting the dif-
ferent types of vehicles into equivalent passenger cars and expressing the volume in terms of passenger
car unit (PCU) per hour. PCU is the universally adopted unit of measurement of traffic volume derived
by keeping the passenger car as the ‘standard vehicle’. Under fairly homogeneous traffic conditions,
the vehicles follow traffic lane, and the volume or capacity under such conditions is expressed in terms
of passenger cars per hour per lane. The vehicles of highly heterogeneous traffic with widely varying
physical and operational characteristics, such as the one prevailing on Indian roads, occupy on the
basis of the availability of space any convenient lateral position on the road without any lane disci-
pline. Under such conditions, expressing traffic volume in terms of PCU per hour per lane is irrelevant,

*Correspondence to: Karuppanagounder Krishnamurthy, Department of Civil Engineering National Institute of Technology
Calicut, Calicut, Kerala 673601, India. E-mail: kk@nitc.ac.in

Copyright © 2012 John Wiley & Sons, Ltd.


2 K. KRISHNAMURTHY AND V. T. ARASAN

and the volume of traffic has to be expressed by taking the whole width of roadway into consideration.
The interaction between moving vehicles under such heterogeneous traffic condition is highly
complex.
The present study is aimed at quantifying the vehicular interactions in heterogeneous traffic under
different roadway and traffic conditions. A micro-simulation technique has been used to model the
heterogeneous traffic flow observed on Indian roads, and the vehicular interaction has been quantified
in terms of PCU.

2. LITERATURE REVIEW

With the aim of the study being the quantification of the interaction of different vehicle types with other
vehicles in heterogeneous road traffic, it was first decided to identify the major factors influencing the
extent of interaction so that the relevant literature can be identified and reviewed. Apart from vehicular
factors, the other major factors influencing the extent of the interaction are road geometry and volume of
traffic (the environmental factors and driver behaviour, which may also influence the extent of interac-
tion, are held as invariants for the purpose of this study). As this study is focused on vehicular interac-
tion at the micro level, as the first step, a straight stretch of road on level terrain is considered. Thus, the
width of the road way is the only geometric aspect considered for this study. The literature review is thus
focused on aspects related to the effect of traffic volume and road width on vehicular interaction. The
search for literature on effect of traffic volume on vehicular interaction revealed that most of the studies
conducted in this regard are related to lane-based and fairly homogeneous traffic (e.g. [1,3] and Srinivas
et al. 2004). A few studies conducted under heterogeneous traffic conditions (e.g. [4–6]) deal only with
macro-level aspects of vehicular interactions, either on the basis of field observed data or with the use of
mathematical techniques. The literature review on impact of road width on vehicular interaction
revealed that the studies are all related to lane-based traffic conditions (e.g. [7–10]), and no studies per-
taining to lane-less heterogeneous traffic condition, dealing with this aspect, have been reported. Thus,
the available literature provided only a limited base for the present study.

3. OBJECTIVES

The general objective of the work reported here is to study the effect of road width and traffic volume,
on the level of interaction between vehicles under heterogeneous traffic conditions prevailing on
Indian roads. The specific objectives of this study are the following: (i) to quantify the vehicular inter-
actions in heterogeneous traffic in terms of PCU, using micro-simulation technique, and (ii) to study
the effect of road width and traffic volume on PCU values of vehicles.

4. SIMULATION FRAMEWORK

On Indian roads, as mentioned earlier, because of the high level of heterogeneity of traffic, the vehi-
cles, while manoeuvring, occupy any lateral position on the road, on the basis of space availability.
In view of this, while modelling the heterogeneous traffic flow, appropriate vehicle movement logics
need to be developed to simulate the stated conditions of traffic flow. As per the adopted methodology
for this study, the entire road space is considered as a single unit. The road space is then considered to
be a surface made of small imaginary squares (cells), thus transforming the entire space into a matrix.
The vehicles will be represented, with dimensions, as rectangular blocks occupying a specified number
of cells, whose coordinates are defined in advance with reference to a fixed origin. This technique will
facilitate identification of the type and position of vehicles on the road stretch at any instance of time
during the simulation process [11]. For the purpose of simulation, the time-scan procedure is adopted.
The scan interval chosen for the simulation is 0.5 s. To address the problem of arrival of more than one
vehicle within the scan interval of 0.5 s at higher volume levels, we have adopted a 0.05-s precision for
generation of time headway. This 0.05-s precision facilitates generation of a maximum of 20 vehicles
per second when the traffic volume is very high. Accordingly, a check will be made for every 0.05 s for
vehicle arrivals and the arrived vehicles, if any, will be put on to the entry point of the study stretch of
the road, on a first-come-first-served basis. The simulation process, which is intended to model traffic

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 3

flow through mid-block sections of urban roads, basically, consists of the following three modules:
(i) vehicle generation; (ii) vehicle placement; and (iii) vehicle movement. The flow chart shown in
Figure1 depicts the major logical steps involved in the overall simulation process.

4.1. Vehicle generation


In a stochastic traffic simulation process, the vehicles arrive randomly, and they may have varying
characteristics (e.g. speed, vehicle type, etc.). Traffic simulation models therefore require randomness
to be incorporated to take care of the stochasticity. This is easily performed by generating a sequence
of random numbers. For generation of headways, free speed, etc., the model uses several random num-
ber streams, which are generated by specifying seed value. Whenever a vehicle is generated, the asso-
ciated headway is added to the sum of all the previous headways generated to obtain the cumulative
headway. The arrival of a generated vehicle occurs at the start of the warm-up road stretch when the
cumulative headway equals the simulation clock time. At this point of time, after updating the posi-
tions of all the vehicles on the road stretch, the vehicle placement module is invoked.

4.2. Vehicle placement


Any generated vehicle is placed at the beginning of the simulation stretch, considering the safe head-
way, which is based on the free speed assigned to the entering vehicle, lateral gap and the overall width
of the vehicle with lateral clearances. If the longitudinal gap in front is less than the minimum required
safe gap, the entering vehicle is assigned the speed of the leading vehicle, and once again the check for
safe gap is made. If the gap is still insufficient to match the reduced speed of the entering vehicle, it is

Start

Inputs, Initialization and Generate first vehicle headway

Yes
Is cumulative Move all
precision step = scan vehicles

No

No
Is current time =

Current time = current Yes


time + precision Add vehicle

Headway to output

Generate headway for next vehicle

No
Is simulation time
over?

Yes

End

Figure 1. Simulation framework.

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
4 K. KRISHNAMURTHY AND V. T. ARASAN

kept as backlog, and its entry is shifted to the next scan interval. During every scan interval, the vehi-
cles remaining in the backlog will be admitted first, before allowing the entry of a newly generated
vehicle.

4.3. Vehicle movement


This module of the program deals with updating the positions of all the vehicles in the study road
stretch sequentially, beginning with the exit end, using the formulated movement logic. Each vehicle
is assumed to accelerate to its free speed or to the speed limit specified for the road stretch, whichever
is minimum, if there is no slow-moving vehicle immediately ahead. If there is a slow-moving vehicle
in front, the possibility for overtaking the slow-moving vehicle is explored. If possible, the fast-moving
vehicle will overtake the slow-moving vehicle. If overtaking is not possible, the fast-moving vehicle
decelerates to the speed of the slow-moving vehicle in front and follows it.
The model is also capable of showing the animation of simulated traffic flow over the road stretch
for better understanding of the system.

5. MODEL VALIDATION

The validation problem arises because various approximations to reality are made in creating the model.
The goal of the validation process is to produce a model that represents true system behaviour closely
enough for the model to be used as a substitute for the actual system for the purpose of experimenting
with the system. Validation is usually achieved through the calibration of the model—an iterative
process of comparing the model with actual system behaviour and using the discrepancies between
the two and the insights gained to improve the model. This process is repeated until model accuracy
is judged acceptable. Accordingly, the model of heterogeneous traffic flow was validated using field
observed traffic data.

5.1. Model input


The complexity of heterogeneous traffic flow increases because of the presence of several categories of
vehicles with widely varying performance characteristics in the traffic stream. The different types of
vehicles, commonly observed on urban roads of India, can be grouped into eight different categories:
(1) buses; (2) trucks; (3) cars including jeeps and small vans; (4) light commercial vehicles comprising
large passenger vans and small four-wheeled goods vehicles; (5) motorised three-wheelers, which
includes auto-rickshaws, which are three-wheeled motorised vehicles that carry a maximum of three
passengers and three-wheeled motorised vehicles that carry small quantities of goods; (6) motorised
two-wheelers, which include motor cycles, scooters and mopeds; (7) Bicycles; and (8) tricycles, which
include cycle-rickshaw, which are three-wheeled pedal-type paratransit vehicles that carry a maximum
of two passengers and three-wheeled pedal-type vehicles that carry a small amount of goods over short
distances. The motorised three-wheelers constitute a significant proportion in the total traffic because
of their advantages in higher maneuverability at congested traffic conditions [12].
The operational characteristics of vehicles are influenced by their physical dimensions and mechan-
ical capability. The ability to respond to the traffic stream environment depends upon the manoeuvring
capabilities of the vehicles, governed by the engine power, acceleration/deceleration and turning capa-
bilities, etc. The average overall dimensions of the different vehicle types and the field observed lateral
clearances between vehicles are shown in Table I. The lateral clearance share values are used to calcu-
late the actual lateral clearance between vehicles on the basis of the type of the subject vehicle and the
vehicle by the side of it. For example, at zero speed, if a bus is beside a motorised three-wheeler, then
the minimum clearance between the two vehicles will be 0.3 + 0.2 = 0.5 m. The longitudinal clearance
at zero speed, based on field observations, is taken as 0.5 m for all combinations of vehicles.
The acceleration rate of a vehicle varies over operating speed and between vehicle types. The
field-observed acceleration rates of the different types of vehicles, for three different speed ranges,
are given in Table II.

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 5

Table I. Vehicular dimensions and field-observed lateral clearances.

Average overall dimension (m) Minimum lateral clearance share (m)


Vehicle type Length Breadth At zero speed At a speed of 60 km/h
Buses 10.3 2.5 0.3 0.6
Trucks 7.5 2.5 0.3 0.6
Light commercial vehicles 5.0 1.9 0.3 0.5
Cars 4.0 1.6 0.3 0.5
Motorised three-wheelers 2.6 1.4 0.2 0.4
Motorised two-wheelers 1.8 0.6 0.1 0.3
Bicycles 1.9 0.5 0.1 0.3
Tricycles 2.5 1.3 0.1 0.3

Table II. Acceleration characteristics of the different categories of vehicles.

Rate of acceleration at various speed ranges (m/s2)


Vehicle type 0–20 (km/h) 20–40 (km/h) Above (40 km/h)
Buses 0.89 0.75 0.67
Trucks 0.79 0.50 0.43
Light commercial vehicles 0.82 0.45 0.35
Cars 1.50 1.10 0.95
Motorised three-wheelers 1.01 0.45 0.30
Motorised two-wheelers 1.35 0.80 0.60
Bicycles 0.10 0.10 0.10
Tricycles 0.07 0.07 0.07

5.2. Study stretch of road


The credibility of the simulation model needs to be established through a comparison of simulated
values of traffic flow characteristics with its corresponding values observed in the field. Hence, a road
section, on a straight and level stretch of a four-lane divided road with raised curbs, in Chennai city,
India, was considered for model validation. The width of the road space available for each direction
of traffic flow on the road is 7.5 m. Out of the total width of 7.5 m, a 1.5-m-wide road space, adjacent
to the curb, is reserved for bicycles by making paint markings on the pavement surface. The road sur-
face intended for bicycles, when found free, is also used by motorised vehicles if required. The traffic
flow on the stretch was measured for 1 h by video capturing the traffic and making classified count of
vehicles after transferring the video data to a computing work station. A total of 3704 vehicles were
observed to pass in one direction, through the section, during the observation period of 1 h. It was
found during the traffic volume count that the number of trucks and tricycles was very small, consti-
tuting about 0.4% and 0.2%, respectively, of the total traffic. Hence, these vehicles, for the purpose
of analysis, were treated to be equivalent to suitable other categories of vehicles with similar charac-
teristics. With the static and dynamic characteristics of trucks being more or less the same as that of
buses, the trucks were considered to be equivalent to buses. Also, each tricycle was treated to be equiv-
alent to two bicycles. The observed composition of the traffic, after the said modifications in compo-
sition, is depicted in Figure 2.
The simulation model requires the free-speed parameters of vehicles to fix limiting speeds for vehi-
cles in the simulation process. The free-speed values of the different vehicle types, obtained through
field measurement on the study stretch, during lean traffic periods are shown in Table III.

5.3. Representation of input variables


A number of probability distributions are available to represent the occurrence of events in modelling
stochastic systems. In simulation models, the governing variables pertaining to various input

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
6 K. KRISHNAMURTHY AND V. T. ARASAN

Bus and
Bicycle Truck 2.5%
L.C.V.
9.6% 3.3%

Car
M.T.W. 27.8%
40.9%

M.Th.W.
15.9%
Figure 2. Field-observed traffic composition. L.C.V., light commercial vehicle; M.T.W., motorised two-wheelers;
M.Th.W., motorised three-wheelers.

Table III. Free-speed parameters of the different types of vehicles.

Free-speed parameters (km/h)


Vehicle type Mean Minimum Maximum Standard deviation
Buses and trucks 53 38 68 7
Light commercial vehicles 50 37 65 7
Cars 60 40 90 14
Motorised three-wheelers 45 30 55 7
Motorised two-wheelers 48 25 60 12
Bicycles 14 10 18 3

parameters, which are stochastic in nature, are represented through appropriate probability distribu-
tions. In the traffic flow phenomenon, vehicle arrivals and headways are random in nature and hence
are to be represented using probability distributions. The arrival and headway data were obtained by
running the video of the traffic flow at a slower speed ( one-eighth of the actual speed) to enable
recording of all the vehicle arrivals manually by observing the details displayed on the monitor of
the computer. The number of vehicle arrivals, in successive 5-s intervals, was recorded, covering
the whole of the hourly volume of traffic. The data, thus obtained, after grouping into classes, were
used to represent the vehicle arrivals through suitable distributions. In this study, it was found, through
chi-squared test, that the observed arrival pattern has a significant fit with Poisson distribution (calcu-
lated value of chi-squared was 12.13 against the critical value of 16.92).
Similarly, for the observed traffic volume of 3704 vehicle per hour, the inter-arrival time between
successive vehicles was measured by noting down the time gap between successive vehicle arrivals
by playing the video of the traffic flow at one-eighth of the original speed. Because the measurement
pertains to the total width of road space and because the traffic had a considerable number of smaller
vehicles like motorised two-wheelers, a significant proportion of the observed headways was very
small, resulting in a mean headway, t, of 1.03 s. The headway data, classified over a time interval of
0.8 s, were found to fit into the negative exponential distribution. The goodness of fit was tested using
a chi-squared distribution. The calculated chi-squared value is 13.19 against the critical value from a
chi-squared table, for seven degrees of freedom at a 5% level of significance, of 14.07. Thus, the
observed headway data fitted well into the assumed negative exponential distribution. For a depiction
of the goodness of fit of the headway data, cumulative frequency distribution of the observed and
theoretical headways (inter-arrival time) were plotted on the same set of axes, as shown in Figure 3.
It can be seen that the distribution of observed and theoretical headways match with each other to a
large extent, corroborating the inference obtained through the chi-squared test.
The vehicles generated during the simulation process will be placed at the start of the simulation
stretch of road after assigning free speeds corresponding to the vehicle types. The free speeds of the

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 7

Observed Headway Theoretical Headway

Cumulative % Frequency
120
100
80
60
40
20
0
0 0.8 1.6 2.4 3.2 4 4.8 5.6 6.4 7.2 8
Headway in Seconds

Figure 3. Observed and theoretical headways.

different categories of vehicles were considered to follow normal distribution [13]. The average reac-
tion time of drivers was taken as 0.7 s [14].

5.4. Comparison of observed and simulated values


The model was validated by simulating the field-observed traffic flow and comparing the characteristics
of the simulated traffic with those of the observed traffic. For the purpose of simulation, a 7.5-m-wide,
1400-m-long road stretch was considered. The middle 1000 m was the observation stretch, the initial
200-m length at the entry point was used as a warm-up zone and a 200-m length at the exit point was
also excluded from the analysis to ensure a fairly uniform flow of traffic being observed. To ensure sta-
ble flow condition during the measurement of flow characteristics, we set the simulation clock to start
only after the first 50 vehicles crossed the exit end of the road stretch. During the simulation process,
the time taken by each vehicle to traverse the specified simulation stretch is observed to estimate the
speed maintained by each vehicle. Road geometry and vehicle characteristics were the fixed inputs to
the simulation model. Traffic volume and composition were the variable inputs. It was decided to
validate the model by comparing a derived output of the simulation model with the corresponding
field-observed value. Accordingly, the traffic flow was simulated for 1 h, and the average speed main-
tained by the different categories of vehicles was obtained as the output of the model. The simulation
was run with three random number seeds, and the average of the three values was taken as the final out-
put of the model. The average speed of different vehicle types was observed using video survey data
collected on the stretch. A comparison of the simulated and observed speeds is given in Table IV. It
can be seen that the two values match with each other on the basis of t-test, to a significant extent,
implying the satisfactory level of validation of the model.

5.5. Study of speed–flow relationship


As an additional validation of the simulation model, it was decided to develop speed–flow relationship
for the heterogeneous traffic and check whether the relationship follows the well-established trend. For
developing the speed–flow relationship, we considered a heterogeneous traffic stream on a 7.5-m-wide
road space (for the simulation), with the field-observed traffic composition (depicted in Figure 2). The
simulation experiments were carried out starting from a very low traffic volume-to-capacity level. The
traffic volume, at different traffic flow levels, was measured in terms of the number of vehicles. The traf-
fic speed is the mean traffic stream speed. To get the stream speed, we first obtained the exit volume of
each category of vehicles, with its average speed, for all volume levels. Then, the traffic stream speed
was obtained as the weighted average of the speeds of the different categories of vehicles in the traffic
steam. Also, the model was used to simulate the heterogeneous traffic flow on an 11.0-m-wide road
space over a wide range of traffic volume, and measurement of flow and speed was made following
the procedure adopted for the 7.5-m-wide road. The two speed–flow curves, plotted on the same set
of axes, are shown in Figure 4. It can be seen that the relationship follows the well-established trend
in both the cases, thus confirming the validity of the model to simulate heterogeneous traffic flow under

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
8 K. KRISHNAMURTHY AND V. T. ARASAN

Table IV. Model validation based on observed and simulated speeds.

Observed average speed Simulated average speed Difference Squared


Vehicle type (km/h) (km/h) (deviation) deviation
Bus and truck 31.47 30.39 1.08 1.17
Light commercial vehicle 30.86 30.50 0.36 0.13
Car 33.40 32.31 1.09 1.19
Motorised three-wheeler 31.18 32.31 1.13 1.28
Motorised two-wheeler 33.59 35.12 1.53 2.34
Bicycles 13.63 15.73 2.10 4.41
Total 7.29 10.51
dmean = mean of observed difference = 7.29/6 = 1.21.
t statistic, to = dmean/(Sd/√K), where K = number of data sets = 6.
Sd2 = 10.51/5 = 2.102, where Sd is the standard deviation; Sd = 1.449.
to = 1.21/(1.449/√6) = 2.06.
The critical value of t statistic for 0.05 level of significance and five degrees of freedom, obtained from standard t-distribution
table, is 2.57. It can be seen that the value of t statistic calculated from the observed data (to) is less than the corresponding value
from the t table. Therefore, it is corroborated that the formulated null hypothesis H0: ms mo = 0 (there is no significant difference
between the simulated and observed means speeds) may be accepted.

7.5 m road space 11.0 m road space


50
45
Stream speed in km/h

40
35
30
25
20
15
10
5
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500
Volume in number of vehicles per hour

Figure 4. Speed–flow relationship for the heterogeneous traffic.

varying roadway and traffic conditions. It can also be seen that the capacity values for the traffic com-
position considered are about 4300 and 6800 vehicles per hour for 7.5- and 11.0-m-wide road spaces.

6. ESTIMATION OF PASSENGER CAR UNIT

After a careful study of the various approaches adopted for the estimation of PCU of vehicles, it was
found that the methodology of approach of the Transport and Road Research Laboratory [15], London,
UK, may be appropriate for the heterogeneous traffic being dealt with. The PCU has been defined by
TRRL as follows:

on any particular section of road under particular traffic conditions, if the addition of one
vehicle of a particular type per hour will reduce the average speed of the remaining vehi-
cles by the same amount as the addition of, say x cars of average size per hour, then one
vehicle of this type is equivalent to x PCU.This definition has been taken as the basis
for derivation of PCU values in this study.

The PCU value for the different types of vehicles, at various volume levels, was estimated by taking
the average stream speed as the measure of performance. Accordingly, the stream speed of the

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 9

heterogeneous traffic of chosen composition (Figure 2), for a chosen volume, was first determined.
Then, a certain percentage (50%) of cars was replaced by the reference vehicle type (for which the
PCU value is to be estimated) in the traffic stream, such that the average stream speed remained the
same as before. This was achieved by varying the number of the reference vehicles introduced to sub-
stitute the removed cars until the original speed of the traffic was obtained by simulation. Then, the
number of cars removed divided by the number of reference vehicle type introduced will give the
PCU value of that vehicle type. To account for the possible variation due to randomness, three random
number seeds (three seeds were found to be optimal in this case after trying with more number of
seeds) were used for the simulation, and the average of the three values was taken as the PCU value.
This procedure was repeated for different volume levels, falling over a wide range. The trends of var-
iation of PCU values for the different types of vehicles, obtained as per the said procedure, for 7.5- and
11.0-m-wide road spaces are shown in Figures 5–9. It may be noted that the traffic volume has been

7.5 m road space 11.0 m road space


4
Bus PCU Value

3.5

2.5

1.5
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

Figure 5. Passenger car unit value for buses and trucks.

7.5 m road space 11.0 m road space


2.4
LCV PCU Value

2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

Figure 6. Passenger car unit (PCU) value for light commercial vehicles (LCV).

7.5 m road space 11.0 m road space

2.4
Auto PCU Value

2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

Figure 7. Passenger car unit (PCU) value for motorised three-wheelers.

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
10 K. KRISHNAMURTHY AND V. T. ARASAN

7.5 m road space 11.0 m road space

Two-Wheeler PCU Value


0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

Figure 8. Passenger car unit (PCU) value for motorised two-wheelers.

7.5 m road space 11.0 m road space


0.8
Bicycle PCU Value

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

Figure 9. Passenger car unit (PCU) value for bicycles.

indicated as the ratio over capacity (capacity values taken from Figure 4) for ease of perception of the
traffic volume levels considered for making the plots.

7. RESULTS AND DISCUSSIONS

7.1. Effect of traffic volume


It is clear from Figures 5–9 that the PCU values of the different types of vehicles change significantly
with change in traffic volume. Hence, for the heterogeneous traffic conditions, it is appropriate to treat
the PCU of a vehicle type as a dynamic quantity rather than a constant. The variation of PCU values
over volume for the different types of vehicles (Figures 5–9) shows an increasing trend at low volume
levels and then a decreasing trend at higher volume levels to reach its lowest value at capacity level.
The reason for the trend can be explained as follows. At low volume levels, with the spacing (both
longitudinal and lateral spacing) between vehicles being more, cars (which are the reference vehicles)
are able to manoeuvre through the gaps easily, which facilitates fast movement. Hence, at low volume
levels, the presence of other vehicles with low free speeds may not become a severe deterrent in reduc-
ing the speed of the cars. The longitudinal and lateral space available at low volume levels facilitate the
cars to have high maneuvering freedom and maintain higher speeds. Although a significant speed
difference exists between cars and other vehicles at low volume levels, cars can perform overtaking
manoeuvres without much reduction in their speeds. An increase in traffic volume at this stage signif-
icantly reduces the spacing between vehicles, resulting in a steep rate of reduction in the speed of cars.
Also, on multilane roads, an increase in traffic volume increases the lateral friction, and overtaking
manoeuvres become difficult. This trend continues up to a certain volume level, at which the speed
of the traffic as a whole gets reduced and, consequently, the speed difference between cars and other
vehicle types gets reduced. When the traffic volume becomes high enough to bring down the speed of
cars equal to the average stream speed, the relative speed differences tend to vanish, and the physical
dimensions of vehicles play a major role in determining PCUs. At this stage, a further increase in
volume results in a lesser rate of change (decreases) in the speed of cars, resulting in lesser impact

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 11

due to the introduction of the subject vehicle. When the traffic volume levels are near the capacity level
of the given roadway section, the vehicles are forced to move with smaller relative speed differences
between a car and other subject vehicles. At this condition, the static characteristics (physical dimen-
sions) of the subject vehicle play a major role in the impedance offered by the vehicle, when compared
with their dynamic characteristics, which play a significant role at lower volume levels. This results in
the decreasing trend of the PCU value of the subject vehicle at higher volume levels. The lateral dis-
tribution of vehicles on the given roadway section is found to be influenced by the traffic composition
and volume [16–18].

7.2. Effect of road width


It is observed from Figures 5 to 9 that it is clear that the PCU value of all the vehicles in the heteroge-
neous traffic is higher on the 11.0-m-wide road when compared with the values on the 7.5-m-wide road.
The variation in the width of pavement, which governs the lateral clearance available between two vehi-
cles, significantly influences the capacity, speed and safety of the roadway sections [19,20]. It would be
useful to know the reason for the higher PCU values on wider roads. As the first step in this regard, plots
connecting the volume (volume-to-capacity ratio) and the speed of the different types of vehicles were
made for the 7.5- and 11.0-m-wide roads. It was found from the plots that in all the cases, for a given
volume-to-capacity ratio, the speed of a vehicle type is higher in the case of the 11.0-m-wide road
when compared with the 7.5-m-wide road. The plots for cars and buses and trucks are shown in
Figures 10 and 11, respectively, as examples. The reason for this may be attributed to the fact that when
vehicles do not follow traffic lanes and occupy any lateral position on the road space, while manoeuvring
forward, the manoeuvring process becomes easier on wider roads, facilitating faster movement of vehi-
cles. As all the vehicles are able to move faster on wider roads, the reason for higher PCU values on wider
roads cannot be explained using the speed data in its raw form. Hence, it was decided to calculate the
percentage increase in speeds of all types of vehicles so that the increase in car speed can be compared
with the increase in the speed of other vehicles. The details of the calculation are shown in Table V. It
can be seen from the table that at all the chosen volume-to-capacity levels, the percentage increase in
the speed of cars is higher than the percentage increase in the speeds of all the other categories of vehicles.
This implies that for a given volume-to-capacity ratio, the speed difference between a car and any subject
vehicle is higher in the case of the 11.0-m-wide roads when compared with the speed differences on the
7.5-m-wide roads. Hence, the PCU values of the vehicles are higher on 11.0-m-wide roads compared with
those on 7.5-m-wide roads.

7.3. Check for accuracy of the passenger car unit estimates


The check for the accuracy of the PCU estimates was carried out by simulating homogeneous (cars-
only) traffic and the heterogeneous traffic flows on the same road space. For this purpose, first, the
cars-only traffic flow was simulated on the 7.5-m-wide road space, and the road capacity was obtained
as 3250 cars per hour by making the speed–flow curve. Then, the flows in cars per hour corresponding
to a set of volume–capacity ratios were determined. The capacity of the 7.5-m-wide road under the
heterogeneous traffic (composition as in Figure 2) condition is 4300 vehicles per hour (Figure 4). With

7.0 m Road 11.0 m Road


Cars Speed in km/h

70
60
50
40
30
20
10
0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume-to-Capacity ratio

Figure 10. Speed–flow relationship for cars on the 7.5- and 11.0-m-wide road spaces.

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
12 K. KRISHNAMURTHY AND V. T. ARASAN

7.5 m road space 11.0 m road space

Buses and Trucks speed in


60

50
40

km/h
30
20
10
0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume to capacity ratio

7.0 m Road 11.0 m Road

Buses and Trucks speed in km/h 60

50

40

30

20

10

0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume-to-Capacity ratio

Figure 11. Speed–flow relationship for buses and trucks on the 7.5- and 11.0-m-wide road spaces.

Table V. Comparison of speed variation on 11.0- and 7.5-m-wide roads.

Volume-to- Speed of the vehicle type (km/h) Percentage


capacity increase in
Vehicle type ratio On 7.5-m-wide road On 11.0-m-wide road speed*
Cars 0.25 53.20 56.95 7.0
0.50 41.80 47.90 14.6
0.75 30.95 36.00 16.3
Buses and trucks 0.25 46.50 49.20 5.8
0.50 36.90 42.00 13.8
0.75 27.80 32.15 15.6
Light commercial vehicle 0.25 45.90 48.30 5.2
0.50 38.00 42.75 12.5
0.75 28.50 33.10 16.1
Motorised three-wheeler 0.25 42.80 43.50 4.7
0.50 38.00 42.00 10.5
0.75 30.95 35.50 14.7
Motorised two-wheeler 0.25 45.20 45.70 1.1
0.50 42.00 44.35 5.6
0.75 34.00 38.20 12.4
Bicycle 0.25 13.77 13.80 0.2
0.50 13.76 13.79 0.2
0.75 13.76 13.69 0.5
*Percentage increase in speed is calculated as speed on the 11.0-m-wide road minus speed on the 7.5-m-wide road divided by the
speed on the 7.5-m-wide road and the whole multiplied by 100.

the knowledge of the composition of the heterogeneous traffic, it is possible to know the number of
vehicles of each category present in the traffic stream at the capacity flow level of 4300 vehicles per
hour. The PCU values of the different vehicle categories, at capacity–flow condition, were obtained
from Figures 5 to 9. Then, the number of vehicles in each category multiplied by the corresponding
PCU value gives the PCU equivalents of each category of vehicles. The sum of the equivalent values
then gives the capacity flow of heterogeneous traffic in PCU per hour. On the same lines, the flow in

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
TRAFFIC OPERATIONS AND MANAGEMENT 13

Homogeneous Traffic Heterogeneous Traffic


3500

3000
y = 2816x + 162.5

Volume in PCU
2500 R² = 0.9691
2000

1500

1000

500

0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume-to-Capacity ratio

Figure 12. Traffic volumes in passenger car unit (PCU) on 7.5-m-wide road space under homogeneous and
heterogeneous traffic.

Homogeneous Traffic Heterogeneous Traffic


7000
6000
Volume in PCU

y = 5501.1x + 212.62
5000
R² = 0.9793
4000
3000
2000
1000
0
0.000 0.125 0.250 0.375 0.500 0.625 0.750 0.875 1.000 1.125
Volume-to-Capacity ratio

Figure 13. Traffic volumes in passenger car unit (PCU) on 11.0-m-wide road space under homogeneous and
heterogeneous traffic.

PCU per hour of the heterogeneous traffic, for the selected set of volume–capacity ratios, was esti-
mated. Then, plots relating the set of volume-to-capacity ratios and the corresponding flow were made
for the cars-only and heterogeneous traffic. The two plots, made on the same set of axes, are depicted
in Figure 12. It can be seen that both the plots are closely related to each other, indicating that the PCU
estimates made are fairly accurate at all volume levels. To explain the accuracy of estimates on statis-
tical basis, we performed a t-test by relating the flow in the number of cars per hour for the selected set
of volume-to-capacity ratios and the corresponding heterogeneous traffic flows expressed in PCU per
hour. The calculated value of t (t0) is 2.06 against the critical value (from t table) of 2.30. This implies
that the traffic flow estimates made in PCU per hour significantly represent the simulated flows of the
cars-only traffic. Hence, it can be said that the PCU estimates are fairly accurate.
The results of a similar check performed in the case of the 11.0-m-wide road are depicted in
Figure 13. The results (the calculated value of t is 1.98 against the critical value of 2.17) indicate that
the PCU estimates are fairly accurate in the case of the 11.0-m-wide roads also.

8. CONCLUSIONS

The following are the important conclusions of this study:


(1) The validation results of the simulation model of heterogeneous traffic flow indicate that the model
is capable of replicating the heterogeneous traffic flow on mid-block sections of urban roads to a
highly satisfactory extent. The validity of the model is further confirmed by the speed–flow rela-
tionships developed using the simulation model for 7.5- and 11.0-m-wide road spaces, which are
found to follow the well-established trend of the speed–flow curves.
(2) The PCU estimates, made through simulation, for the different types of vehicles of heterogeneous
traffic, for a wide range of traffic volume levels indicate that the PCU value significantly changes

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr
14 K. KRISHNAMURTHY AND V. T. ARASAN

with changes in traffic volume. Thus, for the traffic condition considered for this study, there is a
reason to treat the PCU value for a vehicle type as a dynamic quantity rather than as a constant.
(3) It is found that, by virtue of the complex nature of interaction between vehicles under the hetero-
geneous traffic condition, at low volume levels, the PCU value of vehicles increases with increases
in traffic volume, whereas under higher volume conditions, the PCU value decreases with an
increase in traffic volume.
(4) The results of the simulation experiment to study the effect of road width on PCU values indicate
that for any vehicle type in heterogeneous traffic, the PCU value increases with increase in the
width of road space.
(5) The check performed to ascertain the accuracy of the PCU estimates by comparing the flow of
cars-only traffic and the PCU equivalent of heterogeneous traffic on 7.5- and 11.0-m-wide road
spaces indicates that the estimates are fairly accurate.

REFERENCES

1. Van Aerde M, Yagar S. Volume effects on speeds of 2-lane highways in Ontario. Transportation Research Part A
1983; 17A(4):301–313.
2. Al-Kaisy AF, Younghan J, Rakha H. Developing passenger car equivalency factors for heavy vehicles during con-
gestion. Journal of Transportation Engineering 2005; 131(7):514–523.
3. Srinivas PZ, Weimin Z, Pengcheng Z. Modelling and mitigation of car–truck interactions on freeways. Transporta-
tion Research Board 2004-Annual meeting. 2004.
4. Kadiyali LR. A study of the problems of single-lane pavements in India and their improvements. Indian Roads
Congress Journal 1973; 296:405–462.
5. Justo CEG, Tuladhar SBS. Passenger car unit values for urban roads. Indian Roads Congress Journal 1984; 362:183–232.
6. Chandra S, Sikdar PK. Factors affecting PCU in mixed traffic situations on urban roads. Road and Transport
Research 2000; 9(3):40–50.
7. Armour M. Effect of road cross section on vehicle lateral placement. Australian Road Research Board 1985;
15(1):30–40.
8. Evans L, Charlton SG. Explicit and implicit process in behavioural adoption to road width. Accident Analysis and
Prevention 2006; 38:610–617.
9. Tang TQ, Wong SC, Huang HJ, Zhang P. Macroscopic modeling of lane-changing for two lane traffic flow. Journal
of Advanced Transportation 2009; 43(3):245–273.
10. Zhang JW, Dai WM, Xiugang Li. Developing passenger car equivalents for China highways based on vehicle mov-
ing space. Transportation Research Board Annual Meeting. 2006.
11. Arasan VT, Kashani SH. Modelling platoon dispersal pattern of heterogeneous road traffic. Transportation Research
Board, Annual Meeting 2003; 1852:175–182.
12. Shimazaki T, Rahman M. Physical characteristics of paratransit in developing countries of Asia. Journal of
Advanced Transportation 1996; 30(2):5–24.
13. Arasan VT, Koshy R. Methodology for modeling highly heterogeneous traffic flow. Journal of Transportation
Engineering 2005; 131(7):544–551.
14. Mukherjee SK, Rao SK, Raichowdhury ML. Fitting a statistical distribution for headways of approach roads at two
street intersections in Calcutta. Journal of Institutions of Engineers (India) 1988; 69:43–48.
15. Transportation and Road Research Laboratory (TRRL), London, Traffic H.M.S.O. Research on Road Traffic. 1965.
16. Chunchu M, Ramachandra Rao K, Satishkumar NV. Analysis of microscopic data under heterogeneous traffic
conditions. Transport 2010; 25(3):262–268.
17. Hidas P. Modelling lane changing and merging in microscopic traffic simulation. Transportation Research Part-C
2002; 10:351–371.
18. Hidas P. Modelling vehicle interactions in microscopic simulation of merging and weaving. Transportation
Research Part-C 2005; 13:37–62.
19. Gunay B. Car following theory with lateral discomfort. Transportation Research Part-B 2007; 41(7):722–735.
20. Gunay B. A methodology on the automatic recognition of poor lane keeping. Journal of Advanced Transportation
2010; 42(2):129–149.

Copyright © 2012 John Wiley & Sons, Ltd. J. Adv. Transp. 2014; 48:1–14
DOI: 10.1002/atr

You might also like