PhD Thesis Full Version - Jezan Md Diah

FIGURE 3.1
Methodology Flow Chart

Site Selection

Data Collection

Field Data Roundabout Traffic Flow Gap
Geometric Design (using Video Recorder) (using Video Recorder)

Wardrop (1957) Calibration and Validation Data Reduction
adapted and modified
by Arahan Teknik (Jalan) Field Model Development
(using Empirical Data)
11/87 (1987)
RESULTS from
RESULTS from Field Model Development
Arahan Teknik (Jalan)

11/87 (1987)

Sensitivity Analysis of Comparison of the results and analysis
Model Development Finding and Discussion
Development of LOS
Conclusion

3.3 SITE SELECTION

Technically, the research methodology begins with site selection, after
extensive literature search on the subject matter and formulation of research focus.
The sites were first identified based on three criterions (see Table 3.1 – 3.3): physical
entity, control mechanism and traffic operation. The following Tables 3.1-3.3 provide
a list of initial potential sites for the study that have been identified within the study
area (Shah Alam). Further elaborations on these criterions were represented in section
3.3.1.

32

Name TABLE 3.1
No. (based on Site Selection – Physical Entity

Map) First Characteristic
Physical Entity

Number of lanes Number of Number Radius of Shape of Type of
per approach lanes at of Arm centre island centre islands pavement at
@ Leg roundabout
circulation in the
approaches Bitumenious

1 A - Bulatan - - -- - Interlocking
Damai Block

2 B - Bulatan 3 arm - 2 lanes 2 4 Dc > 25m Circular Interlocking
Bistari 1 arm - 3 lanes Di > 50m Block

3 C - Bulatan 2 lanes for 2 4 Dc > 25m Circular Interlocking
Mestika every arm Di > 50m Block

4 D - Bulatan 3 arm - 2 lanes 2 4 Dc > 25m Circular Bitumenious
Permai 1 arm - 3 lanes Di > 50m
Bitumenious
5 E - Bulatan 2 lanes for 2 4 - Parabolic
Setiajasa every arm with Flyover Bitumenious

6 F - Bulatan 2 lanes for 2 4 - Parabolic Interlocking
Melawati every arm with Flyover Block

7 G - Bulatan 3 lanes for 2 4 Dc > 25m Circular Bitumenious
Kayangan every arm Di > 50m
Interlocking
8 H - Bulatan 3 arm - 2 lanes 2 4 Dc > 25m Circular Block
Sejahtera 1 arm - 3 lanes Di > 50m
Interlocking
9 I - Bulatan 2 lanes for 2 4 Dc < 4m Circular Block
Darul Ehsan every arm Di < 20m
Interlocking
10 J - Bulatan 2 lanes for 2 4 Dc > 25m Circular with Block
Selangor every arm Di > 50m Flyover
Bitumenious
11 K - Bulatan 2 lanes for 2 4 - Parabolic
Perusahaan every arm Bitumenious

12 L - Bulatan 2 lanes for 2 4 Dc > 25m Circular Bitumenious
Kemajuan every arm Di > 50m Dc = 85m
Interlocking
13 M - Bulatan 2 lanes for 2 3 Dc > 25m Circular Block
Andalas every arm Di > 50m
Interlocking
14 N - Bulatan 2 lanes for 2 4 Dc > 25m Circular Block
Jubli Perak every arm Di > 50m
Bitumenious
15 O - Bulatan 3 arm - 3 lanes 2 4 Dc > 25m Circular with
Megawati 1 arm - 2 lanes Di > 50m Flyover

16 P - Bulatan 2 lanes for 2 4 Dc < 25m Circular
Budiman every arm Di < 50m

17 Q - Bulatan 2 lanes for 2 4 Dc < 25m Circular
Aman every arm 3 Di < 50m

18 R - Bulatan 3 arm - 3 lanes 4 Dc > 25m Circular
Stadium 1 arm - 2 lanes + Di > 50m

Special lane

33

TABLE 3.2
Site Selection – Control Mechanism

Second Characteristic
Control Mechanism

No. Name (based on Map) Priority rules – Lane division: Lane division
yield signs directional lines, single,
Speed limit arrows & double, spaced
signing
at entry = -
1 A - Bulatan Damai at entry = - at entry = - at entry = - at circular = X
at circular = X at circular = X at circular = X

2 B - Bulatan Bistari at entry =  at entry =  at entry = X at entry = single
at circular =  at circular = X at circular = X at circular = X

3 C - Bulatan Mestika at entry =  at entry = X at entry = X at entry = single
at circular =  at circular = X at circular = X at circular = X

4 D - Bulatan Permai at entry =  at entry = X at entry = X at entry = single
at circular =  at circular = X at circular = X at circular = X
at entry = single
5 E - Bulatan Setiajasa at entry =  at entry = X at entry = X at circular =
at circular =  at circular = X at circular = X single
at entry = single
6 F - Bulatan Melawati at entry =  at entry = X at entry = X at circular =
at circular =  at circular = X at circular = X single
at entry = single
7 G - Bulatan Kayangan at entry =  at entry = X at entry = X at circular =
at circular =  at circular = X at circular = X single

8 H - Bulatan Sejahtera at entry =  at entry = X at entry = X at entry = single
at circular =  at circular = X at circular = X at circular = X

9 I - Bulatan Darul Ehsan at entry = X at entry = X at entry =  at entry = single
at circular = X at circular = X at circular = X at circular = X

10 J - Bulatan Selangor at entry =  at entry = X at entry = X at entry = X
at circular =  at circular = X at circular = X at circular = X

11 K - Bulatan Perusahaan at entry =  at entry = X at entry = X at entry = X
at circular =  at circular = X at circular = X at circular = X
at entry = X at entry = X
12 L - Bulatan Kemajuan at entry =  at circular = X at circular = X at entry = single
at circular = X at circular = X
at entry = X at entry = X at entry = single
13 M - Bulatan Andalas at entry =  at circular = X at circular = X at circular =
at circular = X single
at entry = X at entry = X
14 N - Bulatan Jubli Perak at entry =  at circular = X at circular = X at entry = single
at circular =  at circular = X
at entry = X at entry = X
15 O - Bulatan Megawati at entry =  at circular = X at circular = X at entry = single
at circular = X at circular =
at entry = X at entry = X single
16 P - Bulatan Budiman at entry =  at circular = X at circular = X
at circular = X at entry = single
at circular = X

17 Q - Bulatan Aman at entry =  at entry = X at entry = X at entry = single
at circular = X at circular = X at circular = X at circular = X

18 R - Bulatan Stadium at entry =  at entry = X at entry = X at entry = single
at circular = X at circular = X at circular = X at circular =
single

34

TABLE 3.3
Site Selection – Traffic Operation

Third Characteristic

Traffic Operation

No. Name (based Traffic flow, Queuing Remarks
on Map) composition, condition in
distribution %
approach
(number of

queuing
vehicles)

1 A - Bulatan at entry = - at entry = - it is T-Junction NOT roundabout

Damai at circular = X at circular = X

2 B - Bulatan at entry =  at entry =  Conventional Roundabout, Surrounding by
Institutional, Residential & Commercial Area
Bistari at circular =  at circular = 
(speed limit at entry - 60km/hr)
3 C - Bulatan at entry = X at entry = X
Mestika at circular = X at circular = X Conventional Roundabout, Surrounding by
Residential, Commercial & School Area
4 D - Bulatan at entry =  at entry =  Traffic flows is too low

Permai at circular =  at circular =  Conventional Roundabout, Surrounding by
Residential, Commercial, School Area and
Recretional Area (similar like Bulatan Bistari)

5 E - Bulatan at entry =  at entry =  Surrounding by Residential, Recreational,
Commercial & Religious Area
Setiajasa at circular =  at circular = 

6 F - Bulatan at entry =  at entry =  -

Melawati at circular =  at circular = 

7 G - Bulatan at entry =  at entry =  There is traffic signal in this roundabout

Kayangan at circular =  at circular =  Conventional Roundabout, Surrounding by
Industrial & Commercial Area. More heavy
8 H - Bulatan at entry =  at entry = 
vehicles in this roundabout
Sejahtera at circular =  at circular = 
Mini Roundabout
9 I - Bulatan at entry =  at entry = 

Darul Ehsan at circular =  at circular = 

10 J - Bulatan at entry =  at entry =  Surrounding by Industrial & Residential Area

Selangor at circular =  at circular =  Surrounding by Industrial, Commercial &
Residential Area
11 K - Bulatan at entry =  at entry = 
Surrounding by Industrial Area,
Perusahaan at circular =  at circular =  Not too many vehicles

12 L - Bulatan at entry =  at entry = 

Kemajuan at circular =  at circular = 

13 M - Bulatan at entry =  at entry =  Surrounding by Industrial, Commercial &
Residential Area
Andalas at circular =  at circular = 
Propose Flyover for future purpose
14 N - Bulatan at entry =  at entry = 
Had special lane
Jubli Perak at circular =  at circular =  Surrounding by Industrial & Commercial Area

15 O - Bulatan at entry =  at entry =  Small Roundabout

Megawati at circular =  at circular =  Small Roundabout

16 P - Bulatan at entry =  at entry = 

Budiman at circular =  at circular = 

17 Q - Bulatan at entry =  at entry = 

Aman at circular =  at circular = 

18 R - Bulatan at entry =  at entry =  Surrounding by Commercial & Residential Area

Stadium at circular =  at circular = 

35

LEGEND FOR TABLES 3.1 - 3.3:-

:- Not justified
:- Justified to be selected
:- Reason for not be selected
:- Similar like others or exceptional reasons
X :- Generally not covered / not represent the factor for roundabout
 :- Generally covered / represent the factor for roundabout
- :- No information or data

3.3.1 Site Selection Process

Figure 3.2 shows the key plan of the study area of Shah Alam, where eighteen
(18) roundabouts have been identified in Shah Alam. The criterions used in the
selection of a representative typical roundabout for this study are as shown in Table
3.1 – 3.3, and being defined as follows;

i. physical entity (see Table 3.1):
Physical entity is an identifiable of the physical resource / existence related to
the geometric design (e.g. a number of lane per approach, number of lane at
circulation, number of arm or leg, radius of the centre island, shape of centre
islands in the approaches and type of pavement).

ii. control mechanism (see Table 3.2):
Control mechanism is defined as the priority traffic law / rules (e.g. yield signs,
speed limit, lane division: directional arrows & signing, lane division: lines,
single, double, spaced).

iii. traffic operation (see Table 3.3):
Traffic operation is normally defined as the movement of the vehicles at a
certain period / time (e.g. traffic flow, composition, distribution %, queuing
condition in approach number of queuing vehicles).

36

Few other roundabouts in Shah Alam were not considered because of the
following factors: presence of flyover, traffic light and the shape and size of the
roundabout which is not representative for conventional roundabout.

FIGURE 3.2
Map of Roundabout in Shah Alam

Bulatan Bistari

Bulatan Sejahtera

3.3.2 Site Identification
As for this study, two sites were required for model development and model

validation. Through the process of elimination based on the site criterion set,
subsequently, Bulatan Bistari and Bulatan Sejahtera were selected for fieldwork data
collections (see Plate 3.1 and 3.2). Both roundabouts are seems to have same
characteristics and fulfilled the specification set for a typical roundabout in justifying
for their selection (see Table 3.1 – 3.3). As this study is concerning the development
of a capacity model of a weaving section of conventional roundabout, thus only one of
the weaving sections from each of the sites was selected for data collection. Traffic
data were collected continuously (2 hours in the morning and 2 hours in the evening)
to cover both peak and non-peak flow over which stable flow duration was identified
and used for model development and analyses.

37

PLATE 3.1
Satellite Image for Roundabout at Bulatan Bistari

PLATE 3.2
Satellite Image for Roundabout at Bulatan Sejahtera

Although, study at roundabout entry is quite common and had been studied by
many researchers like Kimber (1980), Akcelik et al. (1997), Chik et al. (2004), Khatib
(2006) and Wu (2006). Study at the weaving section of conventional roundabout is
crucial in Malaysia context especially under two conditions; the practical capacity is
being calculated solely based on geometric parameters and secondly, Malaysian traffic
behaviour (manoeuvrability at weaving area) are much to be desired. Thus,
effectiveness of movement at the weaving section would have some influence on the
effectiveness of the whole system, hence the capacity of roundabout. This
phenomenon is illustrated in Plate 3.3 of Persiaran Bulatan Sejahtera, Shah Alam. It
can be seen that weaving section of conventional roundabout is one of the locations
where traffic failure frequently takes place. Hence, the issues such the design and
operational analysis need to be investigated in depth. Plate 3.4 demonstrates the likely

38

activities (decision making and lane changing movements) that may occur
simultaneously in the weaving area of a roundabout. As a result, this may affect the
traffic at the entry of roundabout. Hence, in this study, the focus is to investigate
traffic movement during weaving and diverging condition at conventional roundabout
which not cover the situation of gap at entry and an impedance of upstream traffic.

PLATE 3.3
Traffic Flow at Weaving Section of Bulatan Sejahtera during Morning Peak Hours

PLATE 3.4
Typical Pattern Movements at Weaving Section of Bulatan Bistari

at Entry
at Weaving Section

at Exit

39

3.4 DATA COLLECTION

It is crucial for any traffic engineering study to ensure that accurate and
adequate traffic information (data) being collected and analyzed accordingly to the
specified study objectives. Basically, there are three methods of traffic data collection
from fieldwork which can be manual, semi-auto or automatic method. Previously, the
manual approach had been used widely but it is labour intensive (e.g. enumerators use
traffic counter for counting the total number of vehicles and radar gun for traffic speed
of vehicles).

Nowadays, mech-electronic and it-electronic technologies have changed from
manual to becoming semi-auto and automatic method for traffic data collection and
reduction, which is considered more efficient and economical. Video recording
technique used in this study is helpful especially where vehicle movements are
complicated such as at roundabout, and to get information regarding gap acceptance at
a weaving section is quite impossible to do manually. However, sample of manual
count is required for calibration purposes.

In this study, data collected were related to; roundabout geometric design
parameters, traffic flow and gap. As for roundabout geometric design parameters,
what being examined were the effect of the entry width, width and length of weaving
section of the conventional roundabout in detail. For traffic flow and gap, a video
recording technique was used. The data collection was done during typical weekdays
(Tuesday, Wednesday or Thursday). The reason of selecting any of these days because
the same traffic flow pattern is expected as compared with the other days which may
cause by holiday (Saturday and Sunday), starting day to work (Monday) and ending
day to work (Friday). The duration covers about four (4) hours for each site. This is
because during data retrievable and further analysis consideration are being made for
the stable flow condition where there is no ‘stop’ and ‘go’ traffic flow movements.

In general, studies on roundabout (Garber and Hoel, 2002) and Akcelik (2002)
have stated that the condition of negotiation speed is normally between 48-50km/h
which is based on safety considerations. Therefore, speed is not the main concern by
the following authors, like Kimber (1980), Tan (1997), Akcelik et al. (1997), TRB
(2000), Chik et al. (2004), Kabit et al. (2006), Tracz and Chodur (2006), Khatib
(2006), Wu (2006) and Macioszek (2007) as compared to parameters such as flow and
gap.

40

Prior to data collection, all apparatus or equipment need to be properly
prepared and checked, at least a day before field data collection operation (e.g. make
sure that the battery is fully charge while using the traffic equipment). The following
paragraphs descript some of the equipment used in this study.

3.4.1 Roundabout Geometric Design

Data on roundabout geometric design parameters such as the entry width, the
circulating roadway width, the diameter of inscribed circle and centre circle are used
for data input in model development. The equipment used was Walking Measure
Meter (see Figure 3.3) and the unit is in meter.

FIGURE 3.3
Walking Measure

A survey form was developed for this study for gathering data on roundabout
geometric is as shown in Table 3.4.

TABLE 3.4
Data of Roundabout Geometric Design

Roundabout Geometric Design

Length Entry Entry Capacity of
of weaving
Location / Per Width of width at width at Average section
Description lane weaving entry (veh/hr)
No. Width section weaving circular, roundabout
of (m) width, e
Lanes (m) section, e1 entry, e2 (m)

W (m) (m) (m)

41

3.4.2 Traffic Flow

As for the traffic flow data, Figure 3.4 shows the processes and sequence of video-
captured technique employed in one of the field data collection undertaken at one of
the conventional roundabout being studied. Once, a suitable site had been identified
through site selection process (Section 3.3). The equipments that were used in the
study were video camera with its accessories, computer for data recording and
retrieval, and data reduction software.

FIGURE 3.4
Method of Data Collection using Video-Capture Technique

Fieldwork Video-Captured

Roundabout Video-capture in AVI format
Output
entry flow (pcu) 1800 1707 249 259 1437
1600 1495 116 144 1129 1119 Using TRAIS or SAVA
1400 1257 451
1200 Non Peak Flow
1000 818 period Evening (Peak Flow )
800
600
400
200

0

Morning (Peak Flow )

Pers.RajaMuda(N)-Qin Pers.Masjid-Qin Pers.RajaMuda(S)-Qin Pers.Inst.-Qin

vehicle trajectory

15
10
5
0
-5

Time period (sec)
head vehicle lead vehicle target vehicle lag vehicle

Result and Analysis
Vehicle position (m)
00000000000000000000000000:::::::::::::00000000000000000000000000:::::::::::::00000000000002103658420076:::::::::::::040428408662800000000000000000000000000

The Portable Vision Based Traffic Analyzer (PVBTA) equipment and its
components consist of automated portable camera pole, lifting motor, automated
pan/tilt mount decoder (with video camera), central controller (cables included) and
battery charger. Figure 3.5 shows the set-up of PVBTA during the fieldwork.

42

FIGURE 3.5
The Set-up of PVBTA

Automated pan/tilt mount decoder with video camera

Automated portable camera pole
Computer (laptop)
Central controller

Road section

The important things that need to be considered when setting-up the PVBTA
are:
i. Position

Suitable location that is safe and easy to work on the equipment must be
decided beforehand. Ensure that the pole stands on stable ground. Placed and secured
the pan/tilt mount decoder together with video camera onto the top of the pole.
Connect the control cable from the video camera unit to the central controller, the
local area network (LAN) cable to the laptop/computer and the power cable to the
battery charger. Then connect the two cables from the motor to the central controller.
The pole can then be extended to the required height (maximum height is 5 meter). To
start lifting the pole, press one of the buttons on the remote control. Push the button
again after the pole has reached the desired height. Using the controller, the pan/tilt
mount decoder is adjusted until the vantage point can captured traffic picture on the
road section. The camera is now ready to capture picture of the traffic movements at
target section.

43

ii. Video Recording

The recording duration is then set (timer) and it would depend on the battery
life. Switch on the battery charger by pressing the “on” button for 5 seconds. Switch
on the laptop/computer and open the data reduction software. Make sure that the LAN
cable from video camera unit to the laptop/computer is in good connection/condition.
Make adjustment by focusing and the level of brightness from the view of data
reduction software. The system is now ready to be used. Figure 3.6 shows the section
of roundabout (red line coverage) where the PVBTA was positioned.

FIGURE 3.6
The Location of Setting up the PVBTA

Arm 1

Arm 4 Arm 2

- PVBTA located

Arm 3

The data survey form that was developed to be used for traffic flow data
collection using video- capture technique is shown in Table 3.5. Generally, the traffic
flow was categorised into stable and unstable flow. The stable flow was the flow when
the demand is less than the capacity and unstable flow was vice versa.

44

TABLE 3.5
Data of Traffic Flow

Point Total Flow Total Flow Conflicting Conflicting Total Max. Total Total Non-
(minute) at Weaving at Weaving Flow Flow Flow at Conflicting Conflicting
Weaving
– – (Pcu/hr) (Pcu/hr) (Pcu/hr) Flow Flow
Inner Lane Outer Lane from Inner from Outer (Pcu/hr) (Pcu/hr)
to Outer
(Pcu/hr) (Pcu/hr) to Inner Qwsf Qcf Qncf
Lane Lane
q11 q22
q12 q21

AASHTO (2004), defined unstable flow as when the level of service (LOS) is
E (see Table 3.6). There are many traffic parameters that may be used as indicator for
LOS (e.g. headway, critical gap, delay, flow) based on different types of road.
Hagring et al. (2003) had considered that the circulating flow (particularly at large
circulating flow rates) of roundabout has a considerable influence on capacity which
can cause an unstable condition. Where else, TRB (2000) had pointed out delay as
indicator of LOS (range between 35 – 50 sec/veh as LOS E) for roundabout entry.

In this study, the model was developed under stable flow condition at the
weaving section of the Malaysian conventional roundabout using one minute time
interval where the movement of 0.04 sec per frame of vehicles had been made through
data reduction. In the weaving section, traffic flows are divided into non-conflicting
flow (Qncf), conflicting flow (Qcf) and weaving section flow (Qwsf). In the non-
conflicting flow (Qncf,), the flow consists of flows at the inner lane (q11) and at the
outer lane (q22). Where as, conflicting flow (Qcf), consists of the flows from inner to
outer lane (q12) and flows from outer to inner lane (q21). While as for weaving section
flow (Qwsf ) is the sum of q11, q22, q12 and q21.

TABLE 3.6
General Definition for Level of Service (LOS)

Level of Service General Operating Conditions
A Free flow
B Reasonably free flow
C Stable flow
D Approaching unstable flow
E Unstable flow
F Forced or breakdown flow

(Source : AASHTO 2004)

45

3.4.3 Gap

As for Gap data reduction may be considered more complicated as the process
would involve two vehicles at a time and identification of ‘exact’ spot for weaving
section. Generally, gap is based on units of time (time gap) or units of distance (space
gap). In this study gap based on units of time is used. Studies done by Lertworawanich
and Elefteriadou (2003 and 2007), used the concept of ideal safe gap (Tisg) which is
defined as an adequate time for merging vehicles to change lanes without making or
causing any harmful disruptions to the main traffic streams. Thus, the concept of ideal
safe gap is adopted with slight modification to suite estimating weaving section flow
of conventional roundabout for this study.

The Tisg was used as input data and the rationale of using it (Tisg) had been
explained in Chapter 2 (Section 2.4.2). The Tisg data was extracted from video
captured using SAVA program. Table 3.7 shows the format of the table used for data
analysis in deriving at the ideal safe gap value where further explaination is given in
Section 3.6.

TABLE 3.7
Data of Ideal Safe Gap

Time Pcu/min Pcu/min Total Ideal Safe Total Ideal Safe Ideal Safe Gap
Inner to Outer Outer to Inner Gap (sec) from Gap (sec) from (sec) for Both
Inner to Outer Outer to Inner
- VL9 - VL10 Lane - VL10 Lane, Tisg
Lane - VL9 (sec/pcu)

3.5 DATA REDUCTION

Data of traffic flow and gap from video capture technique were extracted using
Semi-Automatic Video-Analyser (SAVA) program for movement of 0.04 sec per
frame were tabulated in Table 3.5 and 3.7. According to Archer (2003), the program
has been designed to interpret the information from digital films recorded in digital
Audio Video Interleaved (*.avi) and Motion Pictures Experts Group (*mpeg) format.

The SAVA program provides a basis for analysing traffic film data. The basic
functionality of the program includes being able to step forwards through the film one
frame at a time using the media player controls, arrow keys or the mouse wheel. The
program also has a timer or clock that keeps track of the relative position of the

46

current frame in the film sequence in terms of time, thus each time the user moves one
step forward the clock advances 40 milliseconds.

The program also provides the possibility to change the timer in order that it
can be synchronized with other sources of logged data. To start the SAVA-program it
is necessary to click on the SAVA.EXE program icon. The following window will
then appear (see Figure 3.7).

FIGURE 3.7
The SAVA-Program Window at Start-up

3.5.1 Assigning a Film File
The first step in the process of analysing a film is to assign a suitable film file.

Principally, the file menu is open automatically after click on the SAVA.EXE
program icon. When the film file has been assigned, the film will be loaded into the
film display area and the first frame of the film will be shown with the timer at the
first frame. This is illustrated in Figure 3.8.

47

3.5.2 Loading an Existing Output File
The second and optional stage in the set-up phase is to reload an existing

output file. This is only necessary if the user has saved earlier unfinished work and
now wished to re-continue the input of road-user events in the same output file.

FIGURE 3.8:
The Opened Film File

1

3.5.3 Input of Road-User Events
Before the user begins to input event data, it is necessary to create one or

several virtual line using the V-Lines menu. V-Lines are also drawn on the film display
area of the program when the user assigns an existing output file. Once the v-loops
have been created the user can begin to input road-user events (see Figure 3.9).

When the user clicks on an icon, the road-user type e.g. “Car” and an
identification number is shown below the list of icons under the text “Current Input
Object (Type + Identify No)”.

48

Figure 3.9
Input Road User Events

3.5.4 Drawing Virtual-Lines
The key function of this program is the editing of virtual lines. Editing the

lines is carried out from the “V-Lines No” dialog-box is shown a virtual line number
must be selected in the pull-down box above the text “Current Virtual Line” as shown
in Figure 3.10. The user can then draw a line on the film display area that represents a
virtual line as shown in Figure 3.8.

FIGURE 3.10
Creating a Virtual Line

3.5.5 Output File Window and Associated Functions
The output window is shown in the output menu and represents all road-user

event input information (see Figure 3.11). Each event recorded by the user represents
one line in the window. When the file is saved using the “Save All Input to Current
Output File” button is pressed the information that exists in the window is written to

49

an output file together with information concerning time synchronization and details
of the numbers and positions of virtual lines.

FIGURE 3.11
The Output Window

3.6 MODEL DEVELOPMENT

The developed model flow pattern at the weaving area of conventional
roundabout can be represented in Figure 3.12. In the weaving area, basically there are
two types of movement: Non-Conflicting Flow (Qncf) and Conflicting Flow (Qcf). For
condition of Qncf, is when the vehicles are at inner lane, and traverse along the lane
(B-D), and is categorized as Non-Conflicting Flow in Inner lane (q11). Similarly, if
when the vehicles are at outer lane either continue at the circulating roadway or want
to exit (A-C), is being categorized as Non-Conflicting Flow in Outer lane (q22). For
condition of Qcf, is when the vehicles are from inner to outer lane either continue at
the circulating roadway or want to exit (B-C), and is categorize as Conflicting flow
from Inner to Outer lane (q12). Similarly when the vehicles are from outer to inner lane
(A-D), it categorize as Conflicting flow from Outer to Inner lane (q21). In the weaving
area, the weaving section flow of conventional roundabout (WSFCR), the maximum
number of vehicles that travel is depending on non-weaving (non-conflicting flow)
and weaving (conflicting flow) traffic streams. Figure 3.12 shows the method on how
data being retrieved for weaving section flow (Qwsf) and conflicting flow (Qcf) at
weaving section of roundabout. In simple explanation, referring to Figure 3.12, the
non-weaving traffic streams or non-conflicting flow are traffic travelling from point A
to point C and from point B to point D, while the weaving traffic streams or

50

conflicting flow are those traffic travelling from point A to point D and from point B
to point C. High weaving traffic streams can cause restricted weaving area and vice
versa. Both point A and C are in outer lane and for point B and D are in inner lane.
The Weaving Section Flow (Qwsf) can be mathematically expressed as Equation 3.1:-

Qwsf = MAX (q11 + q22 + q12 + q21) (3.1)

FIGURE 3.12
Weaving Area and Weaving Diagram

q22 q21 q22

A q11 q12 q11 C

Outer Outer

Lane D Lane

B Inner
Lane
Inner

Lane

AC

B D - Weaving Area

Where: = Non-Conflicting Flow in Inner lane (pcu/h)
q11 = Non-Conflicting Flow in Outer lane (pcu/hr)
q22 = Conflicting flow from Inner to Outer lane (pcu/hr)
q12 = Conflicting flow from Outer to Inner lane (pcu/hr)
q21 = Weaving Section flow (pcu/hr)
Qwsf = Non-Conflicting flow of weaving section (pcu/hr)
Qncf = Conflicting flow of weaving section (pcu/hr)
Qcf = Ideal Safe Gap (sec)
Tisg

51

While, the Conflicting Flow (Qcf) and Non-Conflicting Flow (Qncf) are
expressed as Equation 3.2 and 3.3 respectively:-

Qcf = q12 + q21 (3.2)

Qncf = q11 + q22 (3.3)

Figure 3.13 and 3.14 show the method on how data being retrieved for Ideal
Safe Gap (Tisg) at weaving section of roundabout.

FIGURE 3.13
Ideal Safe Gap from Inner to Outer Lane Diagram

B.V Conflict point I.V Outer Weaving area
Lane
Inner S.V
Lane g1
g2 g3

FIGURE 3.14
Ideal Safe Gap from Outer to Inner Lane Diagram

g1 g2 g3

Outer S.V I.V Inner Weaving area
Lane Conflict point Lane

B.V

52

The Ideal Safe Gap from Inner to Outer Lane (ISGIO) can be mathematically
expressed as Equation 3.4 when Subject Vehicle (S.V) hit the conflict point, between
Back Vehicle (B.V) and In front Vehicle (I.V), therefore:-

ISGIO = g1 + g2 + g3 (3.4)

Where;
ISGIO – ideal safe gap from inner to outer lane
g1 – time movement of B.V to back of S.V
g2 – time movement of B.V from back of S.V to infront of S.V
g3 – time movement of B.V from infront of S.V to back of I.V

The ideal safe gap from Outer to Inner Lane (ISGOI) can be mathematically
expressed as Equation 3.5 when Subject Vehicle (S.V) hit the conflict point, between
Back Vehicle (B.V) and In front Vehicle (I.V), therefore:-

ISGOI = g1 + g 2 + g3 (3.5)

Where;
ISGOI – ideal safe gap from outer to inner lane
g1 – time movement of B.V to back of S.V
g2 – time movement of B.V from back of S.V to infront of S.V
g3 – time movement of B.V from infront of S.V to back of I.V

As the result, the Ideal Safe Gap, Tisg (sec) for Both Lane is expressed as
Equation 3.6:

Tisg = ISGIO + ISGOI (3.6)
Total number of S.V for both lane

Most Malaysian conventional roundabout is made without road lane marking
(refer Table 3.2). Therefore, in this study, the assumption of vehicles movement are
made based on ISGIO and ISGOI within the range of maximum weaving width (see
Chapter 2, Figure 2.3) of Malaysian conventional roundabout of Arahan Teknik

53

(Jalan) 11/87 (1987). While in Figure 3.12, the dot line at the centre of the weaving
area is just to indicate the jurisdiction between the inner and outer lane.

Researchers such as Cheng (2004), Ip et al. (2004) and Osborne (2002) used
the method of data transformation in their application for mathematical modification
to the values of a variable. As such, this research adopted the method used by Osborne
(2002) in data transformation to the values of a variable but with slight modification in
its application to estimate weaving section flow of conventional roundabout. Hence,
the analytical results, multiple regression of data transformation analysis were used to
initiate the model development for weaving section flow Qwsf (pcu/hr). The statistical
analysis process using MiniTab, is explained in Figure 3.15 as follows;

54

FIGURE 3.15
Statistical Methodology

STEP 1: DATA REDUCTION FROM FIELDWORK
ON DATA COLLECTION

STEP 2: HOW GOOD THE DATA SAMPLE

Graph • Histogram Checking
• Boxplot the outliers
• Stem & Leaf

STEP 3: DESCRIPTIVE STATISTICS
- Mean
- Median
- Maximum
- Minimum
- Standard Deviation
- Skewness and Kurtosis

STEP 4: IDENTIFYING DEPENDENT AND INDEPENDENT VARIABLES

(a) Graph → scatterplot → with regression (visual observation)

(b) Stat → regression fitted line plot (graph show the value of S, R2 and adj. R2)
n to find the regression analysis (respond vs predictor)

STEP 5: DEPENDENT AND INDEPENDENT VARIABLES DECISION
Select from (a) and (b) in order to clarify several potential useful subsets of the variables as
dependent and independent variables and justifying the data transformation stepwise search

STEP 6: DEVELOP MODEL BASED ON SELECTED VARIABLES
- Stepwise Regression: Stat → regression → to find the regression analysis (respond vs predictor)

- Descriptive Statistic base on Respond/Dependent/y-value
[Data transformation is required to perform well based on P, T and R2 value]

STEP 7: VALIDITY CHECKS No
P Value →  < 0.05 = accepted H and No

Normality Test:
Anderson Darling and Kolmogorov Smirnov > 0.150

Durbin Watson between 0 to 4.

Yes

STEP 8: MODEL VALIDATION
New fieldwork = half of the model development sample

Stat → regression → fitted line
Paired t-test > 0.05

Yes

FINAL MULTIPLE REGRESSION OF DATA TRANSFORMATION MODEL

SENSITIVITY ANALYSIS

55

Generally, there are eight steps involved before the final model developed:

Step 1: Retrieve the data from fieldwork (either from primary or secondary

data source). For this study, the fieldwork data obtained from the process of data

reduction by video capture technique.

Step 2: At this stage, after the data reduction was made, the outliers of data

reduction need to be checked thoroughly in MiniTab. Basically there are many ways

to check whether by graph of Histogram, Boxplot or Stem and Leaf (see Figure 3.16).

Normally the outliers occur when there exist device or human errors during process of

step 1.

FIGURE 3.16
Checking the Outliers

Step 3: After there is no outliers, the data need to be defined through
descriptive statistic (see Figure 3.17) in order to define its mean, median, maximum
and etc. It is important to know the limitation of the minimum and maximum for each
data variables. Basically, the range of data variables can be determined from traffic
theory, studies from previous researchers or in technical guideline. Otherwise, the data
reduction from fieldwork data collection needs to be screened thoroughly in order to
know exact range of each data variables.

FIGURE 3.17
Checking Descriptive Statistic

56

Step 4: This stage needs to identify which variables are dependent or
independent. Normally, the scatter plot or fitted line plot is used to determine this
dependent or independent variables (see Figure 3.18 and 3.19). The dependent and
independent variables is normally determined based on visual observation or the value
of standard deviation S and coefficient of determination R2. The decision for
dependent and independent is made for the purpose of research studies.

FIGURE 3.18
Checking Dependent or Independent Variables based on Scatter Plot

Step 5: Next, the decision on significant relationship of each parameters of
data reduction are identified through scatter plot and regression analysis process. In
determining the dependent and independent variables, meaningfully, multiple
regression of data transformation is applied in order to make it easier to be visualized
and to improve interpretability.

FIGURE 3.19
Checking Dependent or Independent Variables based on Fitted Line Plot

57

Step 6: After the final selected dependent and independent variable had been
decided, in order to perform the multiple regression of data transformation model (see
Figure 3.20), the data transformation is required to perform well based on P, T and R2
value.

FIGURE 3.20
Develop Model based on Selected Variables

Step 7: There are two types of validity that need to be checked, which are

model development and model validation. At this stage, after the multiple regression

of data transformation model had been made, then the model needs to be calibrated

and analysed to make sure that the model development for weaving section flow at

roundabout is valid. In model development, this validity can be checked well through

normality test: if P value less than 0.05 (means accept Ho), Anderson Darling and

Kolmogorov Smirnov test > 0.150 and Durbin Watson test value in the range 0 to 4

(see Figure 3.21). If one of test is not valid, the model needs to be re-checked again

through from Step 2.

FIGURE 3.21
Validity Checks for Develop Model

58

Step 8: In this stage, the model needs to be validated with other new fieldwork
data. Normally the minimum data is half of the model development sample (Ryan,
2007). The comparison result of new fieldwork data needs to be in similarity with
model development data through t-paired test. The t-paired test is thus, an indicator
for this model validation where the value should be more than 0.05 (see Figure 3.22).
If the value is less than 0.05, then the data need to be re-checked again in Step1
especially the range of minimum and maximum value should not be less or more than
model development value. That is why many model equations had their limitation or
range values based on their fieldwork conditions.

FIGURE 3.22
Model Validation with other New Fieldwork

Finally, after all steps had been accepted, the Final Multiple Regression of
Data Transformation Model can become as successful model. The result and analysis
of the model developed is discussed in Chapter 4.

3.7 MODEL VALIDATION

In this study, model validation was performed by comparing the developed
model output with new fieldwork data. Rakha et al. (1996) had explained that
validation is the process whereby model outputs (e.g other fieldwork or simulation)
are compared to actual field data of model developed to determine how well the model
replicates real-world conditions. Model validation results are explained and discussed
in Chapter 5.

59

3.7.1 Validation from Other Fieldwork

The developed empirical model is compared with other fieldwork data which
come from other location. The other fieldwork needed to be analyzed must has the
same procedure as of the developed model. Then, the model validation can be done by
comparing the results between the developed empirical model and other fieldwork.
Further investigation was made if the goodness of fit test between both the model and
field validation data is not satisfactory based on correlation factor and paired t-Test
value.

3.8 SENSITIVITY ANALYSIS OF MODEL DEVELOPMENT

Once the model had been calibrated and validated, the final model need to
undergo sensitivity analysis. The sensitivity analysis was performed to determine the
relationship of the parameters (e.g. weaving section flow, non-conflicting flow,
conflicting flow and ideal safe gap) through some plots in order to specify the
sensitiveness of measures of effectiveness (MOE). Once the sensitivity analyses were
satisfied, a chart of weaving section flow (Qwsf) was developed for measuring the level
of service (LOS) at weaving section of conventional roundabout. The LOS criterion
for weaving section is made based on ideal safe gap. To-date studies on LOS for
roundabout were done at the entry and can be found through TRB (2000), where the
parameters used were flow and delay. There is no record on LOS being prescribed
based on studies at weaving section of roundabout. This study has able to come up
with the LOS chart at the weaving section, whose parameters are flow and gap. With
current data acquisiance technology, such as the video capture technique of vehicles
movement, LOS measured using ideal safe gap is no more complicated in the field.
The discussion on sensitivity analysis results and development of Qwsf application
chart is explained in Chapter 6.

60

3.9 SUMMARY

This empirical study on the development of weaving flow model of
conventional roundabout requires a systematic, well planned and organized/structured
data collection procedure. The experimental process of the study had been explained
through a simplified flow chart process. Field data comprises both the physical
geometry measurements and traffic flow parameters that were collected from the
selected identified typical conventional roundabouts (two numbers of site selection)
were statistically verified before further analyses being done. Traffic data were
collected using an elevated video recording technique to capture traffic movement
along the weaving section of the roundabout. This technique has few advantages,
among others no interruption to traffic operation/movement, safety and convenience
with less man-power required, continuous data recorded which can be retrieved any
time, and most important able to visualize or study traffic operation/interaction during
the recording periods. Field works were performed during a typical weekday over a
period of four (4) hours. SAVA software was used to assists during data
retrieval/image processing, and statistical tools such as Excell and MiniTab make
analyses much easier and efficient. The field raw data were treated and flow trends
were identified, from which a sample of typical stable flow condition was selected for
further analyses. Details statistical analyses and tests were performed for data
verification, model development and sensitivity analyses. Details of the processes
involve have been deliberated in the relevant and related sections of the chapter. It is
important to stress that traffic data collected using video recording at the weaving
section provides very useful and fundamental information as regard to the way traffic
weaving behaviors/interactions during decision making and changing lanes process.
As with the case of Malaysia scenario (for this case study) mixed traffic with quite
high proportion of motorcycles and attitude of drivers make the study very interesting
and challenging. The following chapter will discuss on the results and data analysis of
this study.

61

CHAPTER FOUR
EXPERIMENTAL DATA, DATA ANALYSIS AND RESULTS

4.1 INTRODUCTION

This chapter presents the data collected from videotaping, that were retrieved
using Portable Vision Based Traffic Analyser (PVBTA) and data reduction using
Semi-Automatic Video Analyser (SAVA). The raw data were then treated statistically,
and the results were used to develop the model (weaving section flow model of
conventional roundabout). For the purpose of this study, the flow movement at the
weaving section of conventional roundabout was ‘captured’ from traffic volume data
such as non-conflicting flow, conflicting flow and ideal safe gap. The reason or
relevance of these three parameters to develop the model had been explained in
Chapter 3 (Section 3.4). Microsoft Excel and Statistical Software MINITAB are
statistical tools used in this project.

It is important to emphases that the data captured/collected should cover the
full range of peak and non-peak traffic volume and exhibits a normal distribution
pattern (typical flow) in order to justify for model development in the statistical
analysis. In Figure 4.1, Soper (2011) explains that in order to justify for sufficient data
requirement, minimum sample size is required and should be based on anticipated
effect size (r2), desired statistical power level, number of predictors and probability
level. Since the number of predictor that used for this study in develops the model are
three parameters only (e.g. Qncf, Qcf and Tisg for develop the model of Qwsf), therefore
the requirement of the sample is adequate which should be at least 76 for the
minimum required of sample size (see Figure 4.1).

62

FIGURE 4.1
The Minimum Required Sample Size for Multiple Regression Study

4.2 TRAFFIC VOLUME AND TIME GAP IN WEAVING SECTION OF
CONVENTIONAL ROUNDABOUT

Weaving section (weaving area) is the area where vehicles perform through
and weave movement in the area, and parameters such as the non-conflicting flow,
conflicting flow and ideal safe gap are captured. Table 4.1 shows Bulatan Bistari
geometric dimension measured at site, and Figure 4.2 shows a sketch of Bulatan
Bistari. The weaving section of Persiaran Raja Muda (North) in Bulatan Bistari is
selected as its observed capacity (=2205 veh/hr) is close to that calculated based on
Arahan Teknik (Jalan) 11/87 (1987) (=2218 veh/hr), as compared to the other
weaving sections of the roundabout. An example of weaving section capacity (e.g.
Persiaran Raja Muda (North) in Bulatan Bistari) calculation is shown below:

160W (1 + e ) 160(9.5)(1 + 7 )
W = 9.5
Qp = 1+ W = 2218 veh/hr
1 + 9.5
L 50

63

TABLE 4.1
Geometrical Configuration of Bulatan Bistari (Base on Roundabout Geometric Design)

Roundabout Geometric Design

Location / Per Length Width Entry Entry Average Capacity
Description lane entry of
of Weaving No. of Width of of width at width at
Entry (m) width, e weaving
Section Lanes weaving weaving circular, roundabout (m) section
3.5 (veh/hr)
Persiaran 2 section section, e1 entry, e2
Raja Muda 3.5
3 (m) W (m) (m) (m)
(South 3.5
Approach) 2 60 7 7 7 7 2006
Persiaran 3.5
2 65 8 10.5 7.5 9 2422
Masjid
(East 50 9.5 7 7 7 2218
Approach)
Persiaran 45 8 9 7 8 2174
Raja Muda TOTAL AVERAGE CAPACITY = 2205
(North
Approach)
Persiaran
Institut
(West
Approach)

The flow patterns at the weaving area of conventional roundabout can be
modelled to consist of two types of movement: Non-Conflicting Flow (Qncf) and
Conflicting Flow (Qcf). As for two circulating lanes conventional roundabout (see
Chapter 3 in Figure 3.12), the non-conflicting flow consists of q11 (number denotes
lane number i.e 1-1 move in same lane direction) and q22 similarly for conflicting flow
may consists of q12 and q21 (number denotes lane number i.e 1 to 2). From Arahan
Teknik (Jalan) 11/87 (1987), the conventional roundabout is normally based on the
width and length of weaving section and not considers the lane of the weaving section.
The width of weaving section sometimes can be bigger at the outer lane compare with
inner lane because of give way to the next exit leg of roundabout. Basically, in stable
flow condition, the behaviour of driver is normally following the two types of
movement: Non-Conflicting Flow (Qncf) and Conflicting Flow (Qcf).

64

FIGURE 4.2
Geometric Layout of Weaving Section at Bulatan Bistari

Persiaran Raja Muda (North)

Persiaran Institut (West) Persiaran Masjid (East)

Persiaran Raja Muda (South)

4.2.1 Traffic Volume for Non-Conflicting Flow (q11 and q22)

In this study, the data were recorded in 1-minute interval (which represent

traffic volume in pcu per hourly interval) over a duration of two hours (morning and

afternoon sessions) on Wednesday which is taken to represent a typical weekday.

Non-conflicting flow is where vehicles maintain in its lane which traverse through the

weaving section. Tables 4.2 and 4.3 show the total number and different types of

vehicles at weaving section (Persiaran Raja Muda (North)) of Bulatan Bistari on

Wednesday. Volumes on both lanes are the non-conflicting flows at inner lane and

outer lane and symbol as q11 and q22.

TABLE 4.2
Total Vehicle at Inner Lane of Bulatan Bistari, q11

Vehicle/Time Car Mcy Mpv/Van Lgv Hgv Bus
5
Morning session 1249 658 205 26 10 4
(7.00am – 9.00am) 1350 741 214 33 0

Evening session
(4.00pm – 6.00pm)

TABLE 4.3

Total Vehicle at Outer Lane of Bulatan Bistari, q22

Vehicle/Time Car Mcy Mpv/Van Lgv Hgv Bus
70
Morning session 196 28 0 103
(7.00am – 9.00am) 1421 503

Evening session 334 46 2
(4.00pm – 6.00pm) 2324 1317

65

Figures 4.3 and 4.4 show the pie chart of the percentages of the vehicle passed
through the weaving section under non-conflicting movement (q11 and q22).

FIGURE 4.3
The Total Vehicle, q11 in Percentage at Bulatan Bistari

Total vehicle in percentage at Bulatan Bistari, q11

Car Mcy Mpv/Van Lgv
Mcy 31% 9% 1%
Mpv/Van
Lgv Car Hgv
Hgv 59% 0%
Bus

Bus
0%

FIGURE 4.4
The Total Vehicle, q22 in Percentage at Bulatan Bistari

Total vehicle in percentage at Bulatan Bistari, q22

Lgv Hgv Bus
Mpv/Van 1% 0% 3%

Car 8%
Mcy
Mpv/Van

Lgv

Hgv Mcy Car
Bus 29% 59%

From the figures, it can be seen that car constitutes the highest percentage with
an average of 59% for both q11 and q22. This followed by motorcycle (30%),
multipurpose vehicles (mpv) or van (8.5%) and others (e.g. Lgv, Hgv and Bus) with
less than 2.5%. This could be likely due to the type of the surrounding land-use areas
which are institutional area, residential area and commercial area.

Figure 4.5 shows the time series plot for the non-conflicitng flows (q11 and q22)
in pcu per hourly interval at Bulatan Bistari in the morning (range between 7.00am –
9.00am). The plot indicates at certain point of time there were more flow in the inner

66

lane as compared to those at the outer lane, and most other times the numbers almost
equal between q11 and q22 respectively. Similiarity, the pattern for both non-conflicting
flow (q11 and q22) shows that the flows are almost equal from non-peak to peak hours
and had almost the same flow pattern. From the plot it can be seen that for the first 40
minutes (7.05am to 7.45am) flow is increasing to the highest point, could be because
of the critical time (peak hour) which may lead to congestion. For q11 and q22 at
minute 26 (7.31am), the point is decrease, which different from minute 25 (7.30am).
This might be of an extra ordinary event or the equipment error during data collection.
After minute 40 (>7.45am), both flows are decreasing. Time series plot is good to
have as it recorded and highlighted the occurrence of non-peak to peak and to non-
peak flow periods, this assists in the selection of stable flow condition to be used in
model development.

FIGURE 4.5
Total Non-Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Morning)

Variable

It is quite common, in most traffic engineering studies, to convert the total
number of vehicles to an equivalent passenger car unit (pcu) as to take into
consideration the effect of the various vehicle types on road space. The average
passenger car unit per hour (pcu / hr) factor for this study had been computed to be
1.0595. Sample calculation of pcu for Persiaran Raja Muda (North) of Bulatan
Bistari) is shown in the last page of Appendix A.

67

Similarly, Figure 4.6 shows the time series plot of non-conflicting flow (q11
and q22) in pcu per hourly interval at Bulatan Bistari during the evening (4.00pm –
6.00pm) session. From the graph it shows that more vehicles were travelling on the
outer lane as compared to the inner lane. This most probably due to most of the
vehicles were the one just enter the roundabout from the immediate arm to the exit.
The figure shows that at minute 103 (5.42pm) and 105 (5.44pm) for q22, the values are
the same which is 6166 pcu/hr and seem abnormal (too high). This indicates that
outliers exist within the data. Hair et al. (1998) explained that “all outliers point
should be identified and differentiate from the results adopted. The classifications of
outliers are based on:

i) Procedural like data entry errors where it should be corrected or treated as
missing values.

ii) Outlier that occurs as the result of an extraordinary event, which then is the
explanation of the uniqueness of the outlier. The researcher must decide
whether the extraordinary event should be represented in the sample.

iii) Extraordinary observations for which the researcher has no explanation. These
are the outliers most likely to be omitted but they may be retained if the
researcher feels that they are a valid segment of the population.

iv) Observations that fall within the ordinary range of values on each of the
variables but are unique in their combination of values across the variables.”

Hence, during model development process, few of the data is likely to be
omitted (especially outliers) in order to represent the sample adequately. Figure 4.6,
indicates that during the evening the outer lane flow (q22) is higher as compared to the
inner lane flow (q11). Comparing the two plots (Figure 4.5 and Figure 4.6), it seems
that the total flows does not varies very much between the morning and evening
sessions.

68

FIGURE 4.6
Total Non-Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Evening)

4.2.2 Traffic Volume for Conflicting Flow (q12 and q21)

Conflicting flow is where vehicles change lane (manuouveur) either from inner
lane to outer lane or vice versa (q12 or q21). Tables 4.4 and 4.5 show the total number
and different type of vehicles at weaving section Persiaran Raja Muda (North) of
Bulatan Bistari for both lanes which are inner lane and outer lane (q12 and q21). All
data is recorded in 1 minute time interval.

TABLE 4.4 Bus
The Total Vehicle at Inner to Outer Lane of Bulatan Bistari, q12
1
Vehicle/Time Car Mcy Mpv/Van Lgv Hgv 2
Morning session 116 25 26 1 0
Evening Session 153 85 28 1 0

TABLE 4.5 Bus
The Total Vehicle at Outer to Inner Lane of Bulatan Bistari, q21
2
Vehicle/Time Car Mcy Mpv/Van Lgv Hgv 0
Morning session 58 45 18 3 0
Evening Session 75 84 21 3 0

Figures 4.7 and 4.8 showed the percentage pie chart of vehicle that passed
through the weaving section under conflicting movement (q12 and q21). From the
charts the highest percentage of vehicle that performed lane changing was car with an
average of 52% for q12 and q21. This is followed by motorcycle (33.5%), multipurpose

69

vehicles (mpv) or van (12.5%) and others (e.g. Lgv, Hgv and Bus) with less than 2%.
The numbers or amount between non-conflicting flow (q11, q22) and conflicting flow
(q12, q21) activities at the weaving section, it seems that the amount of non-conflicting
flow is greater than conflicting flow.

FIGURE 4.7
The Total Vehicle, q12 in Percentage at Bulatan Bistari

The total vehicle in percentage at Bulatan Bistari at q12

Car Lgv Hgv Bus
Mcy 0% 0% 1%
Mpv/Van Mpv/Van
Lgv 12% Car
Hgv 62%
Bus Mcy
25%

FIGURE 4.8
The Total Vehicle, q21 in Percentage at Bulatan Bistari

The total Vehicle in percentage at Bulatan Bistari at q21

Lgv Hgv Bus
0% 1%
2%
Car Mpv/Van
Mcy
Mpv/Van 13%
Lgv
Hgv Car
Bus 42%

Mcy
42%

Figure 4.9 shows the plot of conflicitng flow (q12 and q21) in pcu per hourly
interval with time at Bulatan Bistari in the morning. The plotted values have been
converted to passenger car units (pcu/hr), as to consider the effect of the various type
of vehicles on the road. The pattern indicates that there were more movement or lane

70

changing from inner lane to outer (q12) as compared from outer to inner (q21), and this
is most probably that vehicles were exiting at the end of the weaving section. From the
graph it shows that the first 30 minutes, ther is an increase lane changing activities to
the highest point because of the critical time (peak hour) which may contribute to
congestion. At minute 11 (7.15am), q21 shows drastic increase, which is quite
different and suspected due to an extraordinary event (Hair et al., 1998) or the
equipment error during taking the data. It is important to empheses that equipment
calibration and validation had been done prior to field data collection as to ensure that
all of the data being observed in this study is not subjected to the equipment error.
After minute 30, both flows are decreasing and showing normal expected trend.

FIGURE 4.9
Total Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Morning)

Figure 4.10 highlights the pattern of conflicitng flow (q12 and q21) at Bulatan
Bistari during the evening session. The graph shows the pattern vehicles move with
weave action at the weaving section of conventional roundabout. It seems that there
are greater movement of changing lane from inner lane to outer lane during this period
of time. In other words there are many who possibily want to exit than
obiting/circulating the roundabout. The graph also shows that at minute 6 (4.05pm) for

71

q12, the value is 528pcu/hr. This show the outlier point which similar with Figure 4.9
that has an extraordinary event (Hair et al., 1998) such as forced flow situation occurs.

FIGURE 4.10
Total Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Evening)

4.2.3 Time gap – Ideal Safe Gap
In this study of traffic movement in the weaving section the data collected

comprises of Non-Conflicitng Flow (pcu/hr), Conflicting Flow (pcu/hr) and Ideal Safe
Gap (sec). These data are used to modelled and evaluate the weaving process at the
weaving section of conventional roundabout. Figure 4.11 and Figure 4.12 show data
representation of ideal safe gap with time that had been collected for the morning and
evening sessions respectively.

72

FIGURE 4.11
Ideal Safe Gap at Bulatan Bistari (Morning)

The pattern and trend of both graphs are similar with an average value (Tisg) of
3.000 second. There is an extraordinary event (Hair et al., 1998) exhibit in for both
Figure 4.11 and 4.12 where at minute 33 (7.37am) in the morning and at minute 72
(5.11pm) in the evening. This could be likely due to the occurrence of ‘forced flow’
phenomenon during that interval of time. This is indicated by no conflicting flow
with 1 minute time interval which also represent no ideal safe gap at that time.
Looking critically at the graphs, it seems that the ideal safe gap in the morning and
evening, the data captured mostly between 1.000 – 4.000 second. Thus, this range of
values is used for model development.

73

FIGURE 4.12
Ideal Safe Gap at Bulatan Bistari (Evening)

4.3 ANALYSIS OF TRAFFIC PARAMETERS AT WEAVING SECTION
The following paragraphs explained the process and discusses the effects of

combining the traffic parameters of Non-Conflicting Flow (Qncf = q11 + q22),
Conflicting Flow (Qcf = q12 + q21), in deducing the Weaving Section Flow (Qwsf = Qncf
+ Qcf ) for the morning and evening sessions respectively.
4.3.1 Morning Period – Analysis of Traffic Flows at Weaving Section

Time plot analysis of the Non-Conflicting Flow (Qncf = q11 + q22), Conflicting
Flow (Qcf = q12 + q21) and Weaving Section Flow (Qwsf = Qncf + Qcf ) are shown in
Figures 4.13 – 4.15 respectively. They all showed similar characteristics and trend,
with higher volume initial and tend to decrease and stable toward the end of the
counting periods. Figure 4.13 and Figure 4.14 illustrate the flow pattern when
individual types of movements (non-conflicting and conflicting movements) were
being plotted independently. Whereas, Figure 4.15 shows what is really happening in
the weaving section (existence of both the non-conflicting and conflicting movement).
The interpretation from these graphs is that the initial portion indicates higher volume

74

with congested activities as compared to the later half where with less volume the
flow seems more stable.

FIGURE 4.13
Combination of Non-Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Morning)

FIGURE 4.14
Combination of Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Morning)

75

FIGURE 4.15
Combination of Weaving Section Flow per 1-Minute Interval at Bulatan Bistari (Morning)

4.3.2 Evening Period - Analysis of Traffic Flow at Weaving Section
As for the evening period the flow patterns and trends for movement types for

individual (Qncf and Qcf) and combined (Qwsf) are shown in Figures 4.16 – 4.18. These
Figures 4.16 - 4.18 are similar to those shown in Figures 4.13 - 4.15 of section 4.3.1
earlier, except that the initial values (flows) are lower or less as compared towards the
end of the data collection periods. It starts from lower value and increases slowly until
it reaches minute 100 (5.39pm), 103 (5.42pm) and 105 (5.44pm) for both non-
conflicting flow and weaving section flow respectively. This could be likely due to the
occurrence of ‘forced flow’ phenomenon during that interval of time. Only for
conflicting flow, small differences of flow can be seen. Thus, it can be deduced that at
the beginning, the flow is low and increases with time and this can be due to the
reversal traffic flow pattern related to land-use activity (to-and-from work) of the
study location.

76

FIGURE 4.16
Combination of Non-Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Evening)

FIGURE 4.17
Combination of Conflicting Flow per 1-Minute Interval at Bulatan Bistari (Evening)

77

FIGURE 4.18
Combination of Weaving Section Flow per 1-Minute Interval at Bulatan Bistari (Evening)

4.4 ANALYSIS OF CONFLICTING FLOW VERSUS IDEAL SAFE GAP
AND WEAVING SECTION FLOW
Figure 4.19 shows the scatter plot of relationship between conflicting flow

(Qcf) and ideal safe gap (Tisg). From the figure, the graph shows polynomial trend with
as the ideal safe gap reduces there is a corresponding increase of conflicting flow
Thus, it can be deduced that the availability of ideal safe gap would give significant
impact on conflicting flow.

However looking at Figure 4.20 it shows that most of the points clustered
around the 45º line and within the 5000 pcu/hr (Qwsf). At this stage of analysis, it can
be interpreted that there is a strong correlation between Qwsf and Qcf. It is clear that the
increment of conflicting flow indicate the weaving section flow fall within the range
of 5000 pcu/hr (by ignoring the two points that might give extraordinary event) based
on the results obtained. In addition, Arahan Teknik (Jalan) 11/87 (1987) also indicated
that within the range of 5000 pcu/hr is used for the conventional roundabout.

78

FIGURE 4.19
Ideal Safe Gap vs Conflicting Flow at Bulatan Bistari

FIGURE 4.20
Conflicting Flow vs Weaving Section Flow at Bulatan Bistari

79

4.5 VERIFICATION OF DATA SET – CHECKING THE OUTLIERS OF
THE VARIABLES

Data accuracy and reliability are crucial in empirical modelling, where
statistical tool is normally used for verification. Generally, there are several methods
to identify outliers from the data set. Outlier is data or few data which does not fall or
exhibit the general trend of the data set, and this may be due to observer error,
equipment or even an extraordinary even that occurs during data collection. As
mentioned earlier in Chapter 3, the data sample can be checked through histogram,
box plot and stem & leaf. In this study, a method known as box plot method, as
illustrated in Figure 4.21 is used. Once the outliers are identified, their existence need
to be justified or be omitted, and this has to be based on guidelines from previous
research (i.e. limitation of the variables). More or detail of this subject on verification
will be dealt in the forthcoming related chapters.

FIGURE 4.21
Example of the Outliers using Boxplot Method

Boxplot of q11, q22, Qncf

9000

8000  - Outliers

7000

6000

pcu/hr 5000

4000

3000

2000

1000

0 q22 Qncf
q11

80

4.6 SUMMARY
This chapter had deliberated on data treatment or reduction collected using

videotaping at weaving section of Malaysian conventional roundabout. Basically,
traffic movement were recorded on Wednesday for two sessions (morning and
evening) for two hours (7-9am and 4-6pm). Traffic parameters of concerned are the
traffic volume (non-conflicting and conflicting flow) and ideal safe gap (Tisg). Data is
statistically treated and verified. Initial stage, graphical plots is used to visualised
outliers and where justified can be omitted. Time series plots of individual movement
types for both morning and evening session were presented, interpreted and discussed.
Weaving section flow (Qwsf) is deduced from the combination of individual movement
types (Qwsf = Qncf + Qcf). Graphs of relationship between Qwsf versus time, Tisg versus
Qcf and Qcf versus Qwsf , were plotted interpreted and discuss critically. From these
initial findings (graphs), the driver’s lane changing behaviour (the conflicting flow
and ideal safe gap) has significant impact on the weaving section flow, and more of
this will be explored in the forthcoming chapters related to model development of
weaving section flow of Malaysian conventional roundabout.

81