See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336666913

Traffic Simulation Modeling for Major Intersection (Sakarya University

Journal of Science)

Article in Sakarya University Journal of Science · February 2020

DOI: 10.16984/saufenbilder.490984

CITATIONS READS

7 639

1 author:

Ahmed A. Bakhsh
King Abdulaziz University
38 PUBLICATIONS 167 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ahmed A. Bakhsh on 26 November 2021.

The user has requested enhancement of the downloaded file.

 Sakarya University Journal of Science
ISSN 1301-4048 | e-ISSN 2147-835X | Period Bimonthly | Founded: 1997 | Publisher Sakarya University |
http://www.saujs.sakarya.edu.tr/

Title: Traffic Simulation Modeling for Major Intersection

Authors: Ahmed Bakhsh

Recieved: 2018-11-30 22:50:35
Accepted: 2019-09-24 15:07:10

Article Type: Research Article

Volume: 24
Issue: 1
Month: February
Year: 2020
Pages: 37-44

How to cite
Ahmed Bakhsh; (2020), Traffic Simulation Modeling for Major Intersection.
Sakarya University Journal of Science, 24(1), 37-44, DOI:
10.16984/saufenbilder.490984
Access link
http://www.saujs.sakarya.edu.tr/tr/issue/49430//490984

New submission to SAUJS

http://dergipark.gov.tr/journal/1115/submission/start
 Sakarya University Journal of Science 24(1), 37-44, 2020

Traffic Simulation Modeling for a Major Intersection

Ahmed Atef Bakhsh1

Abstract

Nowadays, computer simulation is widely used in industries to support decision-making and performance
enhancement. Real situations often require optimal solutions in the presence of multiple constraints. Present study
aims to solve a transportation problem with the help of simulation modeling. Different kind of models are investigated
using multiple iterations to reduce vehicle cycle and waiting time at roundabout intersection. Discrete Event
Simulation (DES) is done to compare base model against two alternatives, while, ARENA soft ware is used for data
analysis and computations. Results indicated that the alternative 2 of round about only is the best option considering
cycle time, waiting time and number of vehicles transferred out of the system. Findings are beneficial to improve
traffic flow at similar intersections, where commuters face long waiting time during peak hours.
Keywords: Traffic control, Traffic lights, Roundabout, Urban area, DES, Arena, Decision-making

1. INTRODUCTION vehicles. Fig. 1 shows the four main streets intersecting

at the Aldarrajah roundabout. Each road leading to the
In the second half of 20th century, most of the major roundabout is composed of four lanes with one extra
cities in Saudi Arabia were not crowded with people lane dedicated for right turn making a total of five per
and vehicles. However, with the passage of time, road.
number of people and cars were multiplied due to N
urbanization. As a result, Jeddah faces frequent traffic
jams, creating lot of undue stress and safety concerns
among drivers. Some people blame police officers for
their lack of training, while others, shift this
responsibility to the ministry of transportation and the
municipality [1].
In order to improve safety and reduce traffic crowding,
the city management made a plan to smooth traffic flow
at intersections having both traffic lights and a Figure 1. Illustration of the aldarrajah roundabout
roundabout structure. The system under study is a
roundabout with four traffic lights. A roundabout is a The next section presents the simulation modelling
one-way road around a checked focal island to improve practice. In Section 3, problem description is explained.
traffic flow with typical flared ways to deal numerous The proposed input data, data sources and data

1
King Abdulaziz University, Faculty of Engineering, Industrial Engineering Department, Jeddah, Saudi Arabia -
[email protected]
 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

collection methods are described in Section 4. The Traffic Responsive Plan Selection (TRPS) was studied
model building is presented in Section 5. The proposed with the help of actual data to improve intersection’s
alternative system designs are shown in section 6. The traffic control system and reduce the green time
output analysis is presented in section 7. Conclusions wastage. Researches developed the SimTraffic
and future suggestions are discussed in Section 8. simulation model to evaluate traffic timings in different
scenarios. Subsequently, the traffic response control
2. SIMULATION MODELING PRACTICE mode was found to improve the performance of traffic
network [8]. A sudy conducted in China reduced the
load of traffic using a simulation model, which used
Computer simulation is a powerful tool that enables
congestion indicators to show traffic situations in
decision-makers in business industry to enhance their different timings [9].
operational readiness and objectivity [2]. Both
objectives and specificity are better controlled in Another study used simulation modelling to balance
simulation modeling, thus, providing sufficient support the flow of traffic on double-lane roundabouts, where
to the decision-makers. Different types of simulation traffic congestion increased due to increase in the flow
models are available that transforms inputs into outputs of heavy vehicles that also impact the flow of
using multiple iterations. Discrete event simulation passenger cars [10]. DES model was developed in
(DES) is a common method, which simulates the Indonesia optimizing the flow of traffic on a major road
behavior and performance of a real-life process, facility intersection. Arena used all different traffic light
or system. DES models the system as a series of events options against multiple traffic scenarios to reduce the
and support the decision-making process in queue of traffic during red lights [11].
organizations. It supports companies in better
utilization of their resources and capabilities for IT experts frequently used DES and Arena for the
competitive advantage. A surgical center in Belgium simulation of diversified scenarios [12]. A study was
works on DES model. Real data is used for resource conducted to improve operations at airport using
optimization and technical decision making by simulation modeling. Arena was used to reduce
comparing different surgical procedures in atrial congestion at the airport by optimizing passenger time
fibrillation. Major portion of the cost is consumed in between his entery and departure. Iterations were done
managing the resources of operation room, besides for check-in procedures time, security search time and
choosing the best surgical procedure among different parking lot time. The software generated multiple
options. [3]. combinations that were checked and verified in real
time situations to reach the best possible combination
Airport operations used DES to model the landside [13]. Moreover, production of the rubber smoked
processes of baggage drop-off, check-in, security sheets were enhanced using Arena simulation program.
check, security access, and immigration processes. The A Systematic Layout Planning (SLP) theory was used
processes are optimized to allow greater number of to eliminate redundant steps in the process of rubber
passengers utilizing the services per hour [4]. sheet production. Simulation modelling is used to
Environmental impact of the hot in-place asphalt was control air flow parameter, while the plant layout is
studied using multiple recycling approaches with the optimized for better time and distance [14]. Likewise,
help of DES model. Different iterations were done to traffic flow Simulation Model is developed using
compute fuel consumption, construction time, and air Arena for the waterway intersection seaport in China.
emissions for optimized settings [5]. A simulation Different inputs included the tidal window, traffic
model was developed using DES to optimize the rules, different feet types and weather conditions.
assembly line of a television printed circuit board. The Results indicated that the capacity of the waterway was
simulation model improved the workload balance and improved by reducing the waiting time [15].
increase the throughput of the assembly line [6].
Present paper considers the DES modelling by Arena
An event-driven stochastic process was modeled to solve the traffic intersection problem [16]. The
through conventional simulation for the analysis of simulation modelling and analysis are done for the
unmanaged intersections. The analysis was done to current and proposed scenarios in order to support the
improve the flow of traffic by reducing traffic delays decision-making process.
and temporal gaps between two vehicles [7]. The

Sakarya University Journal of Science 24(1), 37-44, 2020 38

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

3. PROBLEM DESCRIPTION Right at

Light 364.50 0.41 74.39
Number of vehicles waiting in queue at each traffic Straight 477.50 0.53 97.45
light is high, creating stress and loss of time among Left 571.00 0.63 116.53
commuters. An optimum road design must be found in
U-Turn 660.75 0.73 134.85
order to reduce the signal waiting time and total cycle
time at the intersection. Table 1 indicates the
3.1. Study Objectives
calculations for travel time and road capacity.
Present study aims to reduce the total waiting time of
Assumptions and descriptions, which are designed for
vehicles in an urban intersection, thus allowing
system behavior, can be listed as follows. maximum number of vehicles to utilize the roundabout
at any given time. For this purpose, traffic control
 Because, each vehicle has to adopt a departure lane parameters need to be designed in an optimized way to
depending upon its destination. These destinations improve the performance measures. It will be done by
are randomly selected from a set using a uniform analyzing different configurations of traffic lights and
and normal distribution. Therefore, these vehicles use of roundabout. In addition, the factors affecting the
are supposed to take corresponding lanes before vehicle queue time and total time consumed in the
time. system will be studied.
 A vehicle can enter the roundabout only, if light is
green. If the light assigned to specific lane is red, 3.2. Propsed Performance Measures
vehicle has to stop.
 At the point of start, where the number of vehicles Below are the performance measures that will be used
are entered into the system for counting, it is in the comparison of different alternatives.
assumed that each vehicle will not change its lane
with any another one.
1. Time in system (minutes): Since there is a single
 For alternative design 2, a vehicle’s wait time
intersection under consideration, so, vehicles are
before a stop sign is distributed from UNIF (0.0333
simulated only, when they enter in one of the
to 0.0833) minutes after consultaions and
joining streets or when they leave the system. In
interviewing sessions with concerned experts.
between these stages, they are considered for
 The roundabout is assumed as a perfect circle. alternative structures.
 Vehicles can enter into system at the point, where 2. Queue waiting time (minutes): Vehicles have to
there is a possibility of only right turn. wait, until they pass through the intersection.
 In order to calculate the road capacity, it is assumed Vehicle waiting is due to traffic light or volume of
that average length of a vehicle is 4.9 meters, which the traffic. This measure can be further split into
is approximately 16 feet. statistics such as “waiting time in queue of line A
 The capacity of lanes is assumed to hold certain of street Y”.
number of vehicles based on defined road length 3. Number of vehicles in system (vehicles): Number
and average vehicle length. of vehicles in system can show the efficiency of the
 Travel time for vehicles is defined on the basis of proposed system at the intersection.
average road length along with average speed,
which is assumed to be 15 km/h. Apart from the performance measures, some response
variables are also defined for the validation of model
Table 1. Calculations for travel time and road capacity that includes average number of vehicles and average
time of vehicles that are randomly selected in the
Average following way:
Average Average road
distance time in capacity  Average number of vehicles turning towards
travelled minutes at assuming a ‘x’ direction from street i (i=1,2,3,4)
(meters) 15Km/h vehicle is  Average number of vehicles waiting in lanes of
4.9 meters street ‘i’
Right before
Light 351.00 0.39 71.63

Sakarya University Journal of Science 24(1), 37-44, 2020 39

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

The simulation model will be validated by comparing Number of vehicles in system

the maximum and minimum values of these measures 1000
with the observed data.
800
4. PROPOSED INPUT DATA, DATA SOURCES,
600
AND DATA COLLECTION METHODS
400
The data for the number of vehicles entered the
intersection was collected by a team of students from 200
the Civil engineering department of King Abdulaziz
University (KAU) Jeddah, Saudi Arabia. The number 0

1045
1155
1305
1415
1525
1635
1745
1855
2005
2115
2225
2335
605
715
825
935
of vehicles passing through the intersection were
calculated for all four streets and their respective lane,
at fixed intervals of 5 minutes for 18-hours, starting at
6:00 am. Arena’s Input Analyzer is used for the Figure 3. Histogram of data (number of vehicles in the
system versus time of the day)
analysis. Resultantly, a graph of the number of vehicles
entered the system for all the intersections is plotted as
seen in Fig. 2. From the hypothesis results, enough evidence was not
available to support the Beta distribution of data. Due
to multi-mode nature of data set, it was decided to use
discrete distribution of number of vehicles by dividing
the data into 18 parts, which corresponds to hourly
readings of data. More so, the probability of each
vehicle to choose one of the possible directions was
Figure 2. Distribution of number of vehicles entered the computed for each hour. Table 2 shows the average
system for 18 hours number of vehicles along with a probability of
choosing a specific lane.
The Input Analyzer is used to perform the chi-square
and Kolmogorov-Smirnov Tests. Similarly, the Table 2. Example data analysis
distribution with smallest MSE that fit into the data was
Right Right
considered as Beta distribution. Also, the MSE results 6:00 am
after before Forw U-
for each distribution are tested. to 7:00
traffic traffic ard
Left
Turn
Total
am
light light
However, after plotting the total number of vehicles Total
versus diffrent day timings, it is realized that the arrival number 4 28 60 143 43 278
of cars
of vehicles are dependent on the time of the day. Fig. 3
Average
shows the relationship of the number of vehicles in the number
system versus the time of the day. Peak loads are of cars
0.34 2.34 5 11.92 3.58 23.17
noticed around noon as well as late afternoon, due to per 5
school and office timings respectively. mins
interval
Probability
of
choosing 0.02 0.10 0.22 0.51 0.15 1.00
lane

5. MODEL BUILDING

Simulation model was developed on the basis of

vehicles entering each street and going in different
directions. The schedule of the route that a vehicle
would travel was defined on the basis of data analysis,
starting from 6:00 am until 12:00 pm. Each vehicle was

Sakarya University Journal of Science 24(1), 37-44, 2020 40

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

assigned a path from the discrete distribution.

Probabilities of these paths are derived from the data
analysis and defined as a variable in the model.

The model logic and traffic light logic are built with the
help of Arena. All models are run for 30 replications.
The resulting confidence intervals are found to be
satisfactory after their comparisons with mean values,
in whichhalf width was much smaller than its mean
value.

a. Model Verification

Animation of the base model was prepared to check the

visual behavior of the model. Picture entities were Figure 5. Animation during the run
assigned in the model to show the directions of
vehicles. Be doing so, it became clear that a vehicle was Another method was used to generate only one entity
choosing the right lane and following the correct path with an objective to trace its path from beginning till
to cross the intersection. Fig. 4 and Fig. 5 shows the end. In addition, multiple entities can also be generated
picture entities along with animation run before and to see the response of model in extreme conditions.
after respectively. However, this option cannot not be realized in present
research due to the limit of maximum 150 entities for
inuse academix version of the software. By using these
techniques, it was verified that the model works error
free.

b. Model Validation

For the validation purpose, some output measures from

the model were compared with the observed real-life
data. Actual system’s values were checked to see their
values within the confidence interval (CI) of simulation
model or not.

CI for average number of cars entered into street 1 was

(18940, 20580) in the simulation model consisted of
Figure 4. Animation before the run
18-hour period with 30 replications, whereas it was
recorded as 19,890 in real life scenario for the same
period. While looking at the lane distributions, the CI
of average vehicles that made U-turns from street 4
were reported as (1230, 1850) in the model whereas,
the real-life data reported the number of these vehicles
as 1,450.

Road lengths were calculated with the help of Google

Earth and measurements were made on the map using
the “path” tool. For the alternative design 1, which has
no roundabout, the measurements were made
according to virtual paths that were designed on the
assumption that no roundabout existed in the map.

Sakarya University Journal of Science 24(1), 37-44, 2020 41

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

The average waiting time in queue was determined

using an expert opinion. In the model, the durations for
street 1, street 2, street 3 and street 4 was around 3
minutes, whereas, it was around 3 to 4 minutes in the
real-life scenario. After comparing model outputs with
real-life observations in the presence of experts, a high
level of confidence on the validity of model was
established.

6. PROPOSED ALTERNATIVE SYSTEM

DESIGNS

Roundabouts are designed everywhere to streamline

the flow of traffic at intersections without any
Figure 6. Group 1 direction
stoppage. However, the current intersection uder study
combined both traffic lights and a roundabout. After
examining the intersection and doing the simulation
modelling, two alternative solutions are considered for
optimizing the flow of vehicles.

a. Propsed Design 1

The first alternative is to simulate the intersection on

the assumption that only lights are present. This design
requires the removal of roundabout as shown in Fig. 6
& Fig. 7, which also includes the cost of road
modifications. Different sequence of the light’s timings
are used to find optimal settings that will streamline
traffic flow.
Figure 7. Group 3 direction

The same set of data collected for the base model was Table 3. Direction groups
further used to find arrival times and direction
assignments. Alternative 1 used direction groups,
Street Green light for
which were different from the base model to signal
green light for a given period of time. Referring to Fig. 1&3 Forward and Right
6, the vehicles in lane 1, 2, 9 and 10, which were 2&4 Forward and Right
grouped in 1 direction, started crossing the intersection 1&3 Left and U-Turn
at the same time. In Fig. 7, the direction of group 3, 2&4 Left and U-Turn
consisted of lane 3, 4, 11 and 12 is shown. The same
discussion is valid for street 2 and 4 as shown in Table b. Propsed Design 2
3.
A study of 23 Roundabouts conducted by the Insurance
Institute for Highway Safety (IIHS) in 2001 in the USA
found that roundabouts not only reduced crashes by
40% but also decreased injury crashes by 80% as
compared to intersections having traffic lights [17].
Therefore, the second alternative model simulated the
roundabout without traffic lights. The modifications for
alternative 2 are based on removal of traffic lights as
shown in Fig. 8. The results are compared with the

Sakarya University Journal of Science 24(1), 37-44, 2020 42

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

results of both alternative 1 and base model to ascertain Table 4. Average number of vehicles left the system
the effectiveness of roundabouts without traffic lights.
Base
Alternative 1 Alternative 2
Model
Average
Number 115,660 115,790 116,200
Out

b. Comparisons with Different Trafic

Light Durations on Base Model and
Alternative 1

Another comparison is done between base model and

alternative 1, because traffic lights are only employed
in these two designs. Different combinations of time
durations for green signal were tested as (2, 1, 2, 1),
Figure 8. Alternative model 2 design
while the real life conditions are reflected as (1, 2, 1,
2); (1, 0.5, 1, 0.5); (2.5, 1.5, 2.5, 1.5) for street 1 to 4 in
7. OUTPUT ANALYSIS the base model and direction groups 1 to 4 in alternative
1 respectively.
ARENA’s output analyzer was used to compare
different models. Statistics of cycle time and waiting From the results, it is observed that alternative 1
times of the vehicles entering each street has been reduced the cycle time of vehicles, whereas base model
determined as a performance measure. Hypothesis performed better regarding the waiting time of
testing was performed at α=0.05 level. For simplicity, vehicles. Also, it is observed that (1-0.5-1-0.5)
performance measures were numbered as 1; 2; 3; 4 combination gave the best improvement in waiting
corresponding to Street 1; Street 2; Street 3; Street 4. times and cycle times for both models. These
Also, CRN is used in the model as a variance reduction comparisons were based upon vehicle cycle time and
option. vehicle waiting time with the help of ARENA output
analyzer.
a. Comparison of Three Models
8. CONCLUSION AND FUTURE WORK
During comparison of the three models, it is revealed
that base model has the poorest cycle time, whereas The alternative 2 of round about only is found to be the
alternative 1 has the best cycle time. Regarding waiting best solution among all options considering the cycle
times, alternative 2 performs better than the base model time, waiting time and number of vehicles transferred
whereas alternative 1 is slightly poorer than the base out of the system. On the other hand, the results
model. All results indicated that there is not enough indicated that the use of round abouts along with traffic
evidence to support the respective means as they are lights was the last option in terms of performance
same between base model - alternative 1 and base measures. Two variables of moving time and waiting
model- alternative 2. time significantly affecte the results.

For further testing, waiting time and cycle time of Similarly, the cycle time of vehicles was reduced in
alternative 1 and alternative 2 are compared with each alternative 1 for different traffic light durations.
other. It is found that Alternative 2 performed better However, the average waiting time was increased.
than alternative 1 in all measures. Therefore, it is concluded that removal of roundabout
was the most critical factor in reducing the cycle time
Table 4 shows the average number of vehicles moving of vehicles at intersection crossing. It is found to have
out of the system at the end of each replication. more impact on cycle time than traffic light durations.

Sakarya University Journal of Science 24(1), 37-44, 2020 43

 Ahmed Bakhsh
Traffic Simulation Modeling for Major Intersection

Future study can be done at a larger scale for bigger Minhajuddin Aquil, “DESIGN OF TRAFFIC
network of roads in the city. Optimization of vehicular RESPONSIVE PLAN SIGNALS AT
traffic using simulation modeling can be helpful in SIGNALIZED INTERSECTION,” 2018.
better utilization of resources. [9] Z. Li, R. Jia, and J. Huang, “The simulation
model on delay time of road accessibility based
REFERENCES on intelligent traffic control system,” Cluster
Comput., pp. 1–11, Jan. 2018.
[1] “Article Number 88324,” Arab News, 2006. [10] O. Giuffrè, A. Granà, M. L. Tumminello, and A.
[Online]. Available: Sferlazza, “Capacity-based calculation of
http://www.arabnews.com/?page=1&section=0 passenger car equivalents using traffic
&article=88324&d=17&m=10&y=2006&pix= simulation at double-lane roundabouts,” Simul.
kingdom.jpg&category=Kingdom. [Accessed: Model. Pract. Theory, vol. 81, pp. 11–30, Feb.
25-Jan-2009]. 2018.
[2] M. F. Hocaoğlu and I. Kandemir, “Agent-based [11] D. Yuniawan, P. . Aang Fajar, S. Hariyanto, and
Dynamically Reconfigurable Missile R. Setiawan, “Traffic queue modeling using
Modelling,” Sak. Univ. J. Sci., vol. 22, no. 5, pp. arena simulation software (a case study of
1–1, Oct. 2018. Mergan 4-Way intersection in Malang City),”
[3] G. Monnickendam and C. de Asmundis, “Why MATEC Web Conf., vol. 204, p. 2009, Sep.
the distribution matters: Using discrete event 2018.
simulation to demonstrate the impact of the [12] T. J. Schriber, D. T. Brunner, and J. S. Smith,
distribution of procedure times on hospital “Inside discrete-event simulation software: how
operating room utilisation and average IT works and why it matters,” Proceedings of
procedure cost,” Oper. Res. Heal. Care, vol. 16, the 2014 Winter Simulation Conference. IEEE
pp. 20–28, Mar. 2018. Press, pp. 132–146, 2014.
[4] M. Groot, “Optimize Landside Airport [13] D.-U. Kim, M.-S. Jie, and W.-H. Choi, “Airport
Operations Using A Discrete Event Simulation Development Using Arena,” Adv.
Simulation,” Delft University of Technology, Sci. Technol. Lett., vol. 149, pp. 5–8, 2018.
2018. [14] A. Watanapa and W. Wiyaratn, “Improvement
[5] B. Yu, S. Wang, X. Gu, F. Ni, and Q. Liu, of Rubber Smoked Sheet Plant using Arena
“Environmental burden evaluation of hot in- Layout for Increasing Productivity,” in the
place recycling of asphalt pavement based on International MultiConference of Engineers
discrete event simulation,” Transp. Res. Part D and Computer Scientists, 2018.
Transp. Environ., vol. 65, pp. 151–160, Dec. [15] B. R. Habeeb, Z. Guo, and X. Song, “A
2018. Simulation Model to Determine the Capacity of
[6] M. Abdelkhak, S. Salama, and A. B. Eltawil, a Y-Type Waterway Intersection for a Real
“Improving Efficiency of TV PCB Assembly Seaport,” Open Transp. J., vol. 12, no. 1, pp. 8–
Line Using a Discrete Event Simulation 20, Jan. 2018.
Approach,” in Proceedings of the 10th [16] W. D. Kelton, R. P. Sadowski, and D. T.
International Conference on Computer Sturrock, Simulation with Arena, Third edit.
Modeling and Simulation - ICCMS 2018, 2018, McGraw-Hill Higher Education, 2004.
pp. 211–215. [17] “Roadway Systems & Drivers Archives -
[7] C. Liu and M. J. Kochenderfer, “Analyzing AAA Foundation.” [Online]. Available:
Traffic Delay at Unmanaged Intersections,” http://aaafoundation.org/category/roadway-
Cornell Univ. Libr., May 2018. systems-drivers/. [Accessed: 15-Oct-2018].
[8] M. Bin Saleh Mehdi, M. Iqbal Faheem, and M.

Sakarya University Journal of Science 24(1), 37-44, 2020 44

View publication stats