E1A119089 - Wa Ode Ainun Jariah

NUMBER
№ 4/2(36), 2017
NAME
Dolya C., Botsman A.,

Kozhyna V.
RESEARCH TITLE
Investigation of approaches
to modeling of intercity
passenger transportation
system
PURPOSE
1. To establish the scientific

approaches proposed by
contemporaries for the calculation
of passenger correspondence
between cities, taking into
account their geographical
calculation, social and economic
indicators. 2. To define methods for
predicting the distribution of
correspondence between modes
of transport.
METHOD
To establish the scientific approaches proposed by contemporaries on the

calculation of passenger correspondence between cities, taking into account
their geographical calculation, social and economic indicators, it is suggested
to use methods of system analysis for studying contemporary scientific thought
on the topic of research. Determine methods for predicting the distribution of
correspondence between modes of transport, using mathematical modeling
techniques to determine the parameters of the technological process of
transporting passengers. Choose methods of field studies to determine the
actual parameters of the technological process of transport and
mathematical statistics to assess the adequacy of the developed models.
CONCLUSION
1. Scientific approaches to the calculation of passenger
correspondence between cities, which rely on the modeling of
transport processes using gravitational modeling techniques, are
investigated. Accounting for geographical location of transport
nodes, their social and economic characteristics are used as
factors of attractiveness or resistance to passenger
correspondence. At the same time, at this stage of the
development of scientific approaches to the formalization of the
parameters of passenger transport correspondence, the
integrated accounting of the flow of technical and economic
parameters of the carriage for satisfying the demand is
determined at an insufficient level. 2. The modern methods for
predicting the distribution of correspondence between modes of
transport are based on the probability of choosing the mode of
transportation from the proposed options. At the same time, the
definition of the parameters of such probability is proposed to be
implemented from the approach of concrete efficiency of the
transportation.
JURNAL KE-
NUMBER
doi:10.4271/11-01-02-0004
NAME
Kaveh Bakhsh Kelarestaghi and Kevin

Heaslip,
RESERCH TITLE
Intelligent Transportation System

Security: Hacked Message Signs
PURPOSE
This study attempts to examine the impacts

of VMS security vulnerabilities, and
contributes to the literature of ITS security
and resiliency, by exploring possible
consequences of an adversary hacking a
VMS. To this aim, we synthesize current
literature and real-world VMS hacking
events’ evidence that has been described
in the current literature and news reports, to
pinpoint possible impacts that those
attacks can impose on the transportation
network.
METHOD
The methodology of risk assessment (Figure 1) comprises of (1)

identification of threat sources and event, (2) identification of
the system vulnerabilities and predisposing conditions, and (3)
determination of the adverse impacts and magnitude of
impact. To this aim, physical and cyber-attacks will
be represented in a tree structure [22] (i.e., attack tree), with
the goal of tampering VMS content and different ways of
achieving that goal. The attack tree, proposed by Schneier
[22], specifies a range of actions that an adversary undertakes
to exploit the vulnerabilities of the VMS
CONCLUSION
The risk-based approach of this study delivers the

impact-likelihood matrix, which maps the adverse
impacts of the threat events onto a meaningful, visual,
matrix. The result provides insights for system operators
and decision-makers to prioritize the risk of a VMS
hacking event. • VMS hacking adverse impacts can
be categorized mainly as high-risk and medium-risk
clusters. The safety, operational (i.e., monetary losses)
and behavioral impacts are associated with a high-risk
cluster. While the security, reliability, efficiency, and
operational (i.e., congestion) impacts are associated
with the medium-risk cluster. • The VMS security
vulnerabilities and predisposing conditions allow
adversaries to compromise the ITS functionality. System
operators should consider adequate physical and
cyber security measures to improve transportation
critical infrastructure security and resiliency. Moreover,
at the same time, since threats on ITS are inevitable,
impact assessment could guide decision-makers on
the adverse impact of a security breach. I
JURNAL KE-3
NUMBER
DOI:
10.1080/23302674.2018
NAME
Reza Sayyadi & Anjali

Awasthi
RESEARCH TITLE
An integrated approach based on

system dynamics and ANP for
evaluating sustainable
transportation policies
PURPOSE
The results of our study reveal that

trip sharing based policies perform
better in comparison to the other
policies for achieving sustainability
in a transportation system.
METHOD
The proposed methodology

involves two steps. In the first step,
sustainable transportation policies
aresimulated. The second step
involves evaluating policy
evaluation.
CONCLUSION
Extending the current study to integrate large size

networks. • Testing the proposed model with real
data and comparing the results with other
comparable studies in the literature • Extending
the study to integrated modelling of passenger
and freight transport policies
JURNAL KE-4
NUMBER
VOL.3,NO 9
NAME
Yu Shen, Hongmou Zhang

and Jinhua Zhao
RESEARCH TITLE
Integrating Shared Autonomous

Vehicle in Public Transportation
System: A Supply-Side Simulation
of the First-Mile Service in
Singapore
PURPOSE
An agent-based supply-
side simulation is built to
assess the performance of
the proposed service in
fifty-two scenarios with
different fleet sizes and
ridesharing preferences.
METHOD
Agent-based simulation of passengers,

buses and AVs (rounded rectangles refer to
the agents, ellipses refer to locations,
rectangles with solid line refer to the
movements of agents, rectangles with
dashed line refer to the interaction
between agents, and diamonds refer to
the decisions of agents.
CONCLUSION
Finally, the case study presented in the paper is

only one of many possible ways to integrate the
shared AVs into the PT system from a planning
perspective. As in Figure 1, there are many
alternative schemata from the viewpoints of
operation, governance, technology, and
consumption, e.g., pricing or ticketing integration,
information sharing, institutional design, etc.
Private companies can also be potential AV fleet
owners and operators.
JURNAL KE-5
NUMBER
VOL.7,NO 8
NAME
Bouraima Mouhamed
Bayane, Qiu Yanjun * and
Yusupov Bekhzad
RESEARCH TITLE
A review and analysis of

railway transportation
system in the economic
community of West African
States: Towards the
development of sustainable
regional goal
PURPOSE
These strategies are envisaged to

contribute to the development of
an efficient and interoperable
railway transportation system in the
Economic Community of the West
African States which in turn will
improve the interconnectivity and
enhance the economic growth
and trade in West Africa
METHOD
The methodology of the study includes the

use of major sources from a practical survey,
analysis of evidence from the experimental
study as well as analysis of data from different
published sources. The list of selected
countries consists of four English speaking
countries (Ghana, Liberia, Nigeria, and Sierra
Leone) and seven French-speaking countries
(Benin Republic, Burkina Faso, Guinea, Ivory
Coast, Mali, Senegal, and Togo)
CONCLUSION
A review and analysis of the railway

transportation system in the Economic
Community of West African States,
including four English speaking
countries and seven French-speaking
countries, is conducted in this paper.
The review of the current situation of
the regional railway transportation
system certifies the global vision of its
assets inherited from colonial powers
with few restorations since then and
single-track lines with light rails, low
axles loads and restricted speeds.
JURNAL KE-6
NUMBER
VOLUME 9,NO. 2
NAME
Jihane LARIOUI 1 ,
Abdeltif EL BYED 2
RESEARCH TITLE
Towards a Semantic Layer Design for an

Advanced Intelligent Multimodal
Transportation System
PURPOSE
The global purpose of the advanced

multimodal information system is to propose
the optimal route that corresponds to the
different criteria of preferences chosen by
the passenger according to their needs in
terms of: travel time, number of mode
changes, cost and safety. This system is
based on a multi-agent system architecture
coupled with semantic layer approach for
a better management of the multimodal
transport network. The purpose of the
Semantic layer is to characterize existing
information in order to facilitate the
automation of services, to discover, link and
infer similar knowledge
METHOD
The structure of this paper is as follows:

Section 2 discusses some related works
and the role of the ontology in a multi-
agent system. The approach used for
ontology development and design are
stated in section 3. Section 4 describes
the Implementation part and discusses
the results. Finally, we conclude the
document with some remarks and
perspectives.
CONCLUSION
This ontology has been developed by

understanding the common terms used in
the transport system and by looking at the
different preference criteria chosen by the
user such as cost, duration, number of
connections and safety. To build this
ontology, we opted for a hybrid
architecture, which consists in developing a
global ontology from local ontologies. For
the present case, we are taking into
account only two information resources:
Tramway and Bus Data Bases. In a future
work, we propose to extend our ontology in
order group all the sources so that these
data are comprehensible and accessible for
the agents of the system. Moreover, we will
detail, the negotiation and communication
part between the agents themselves and
between the different layers of the system.
The ontology can be used with a data
integration technique to build a multimodal
transport system with an agent based
semantic approach.
JURNAL KE-7
NUMBER
VOL.4,NO 1
NAME
, Danman WU1 , Qiuwei WU2 ,

Miadreza SHAFIE-KHAH3,4,
Joaõ P. S. CATALA˜ O
RESEARCH TITLE
Interdependence
between transportation
system and power
distribution system: a
comprehensive review on
models and applications
PURPOSE
The penetration of EVs in modern

smart cities is growing fast, and the
interaction between transportation
and power distribution systems are
becoming more evident. This
paper provides an up-to-date
survey on modelling and
applications of such
interdependent infrastructures, and
envisions prospective research
directions in the near future
METHOD
With the increasing

penetration of EVs, the
system-wide
interdependence across
transportation and power
distribution infrastructures will
become more prominent.
The topic reviewed in this
paper is a very young and
active research field. Authors
believe the following
directions deserve more
attentions in the future.
CONCLUSION
We believe the
distributed mode and
network equilibrium are
more practical in the
future. In conclusion, this
direction is still in its
infancy stage. With
further proliferation of
EVs and the advent of
sharing economy, the
benefits gained from
traffic-power
coordination for
enhancing system
reliability, efficiency and
resilience will be
prominent.
JURNAL KE-8
NUMBER
Vol. 19, no. 1, 2018
NAME
Artūras Petraška1 ,
Kristina Čižiūnienė2 ,
Olegas Prentkovskis3 ,
Aldona Jarašūnienė
RESEARCH TITLE
METHODOLOGY OF
SELECTION OF HEAVY AND
OVERSIZED FREIGHT
TRANSPORTATION SYSTEM
PURPOSE
There are following tasks to achieve the

purpose: • to perform the analysis of
theoretical aspects concerning transport
routing, when carrying HOF; • to create
HOF transportation methodology, after
evaluating the routing principles; • to
perform hypothetical inspection of
created methodology applicability using
an example of Lithuania
METHOD
In scientific literature is a lack of

methodology concerning the
performance of calculations and
choice of elements, what must be
assessed, during HOF transportation
CONCLUSION
HOF routes and processes evaluation

criteria can be grouped into two
groups, in one group all criteria have
the time dimension, and in another –
the monetary dimension. This allows
the formation of the evaluation system
ensuring an objective assessment of
the HOF transportation processes by
comparing different types of
transportation, route segments,
technologies of transportation and
freight transhipment. RHOF routes and
processes evaluation criteria can be
grouped into two groups, in one group
all criteria have the time dimension,
and in another – the monetary
dimension. This allows the formation of
the evaluation system ensuring an
objective assessment of the HOF
transportation processes by comparing
different types of transportation, route
segments, technologies of
transportation and freight
transhipment.
JURNAL KE-9
NUMBER
Vol. 18 No. 3, September

2018
NAME
Nurul Qamar and Aan Aswari

Faculty of Law, University of
Muslim Indonesi
RESEARCH TITLE
HEALING OR HURTING:
DEVELOPMENT OF
HIGHWAY PUBLIC
TRANSPORTATION
TECHNOLOGY
PUPOSE
This paper is not intended to reject

the collaboration of technology.
However, it may fail to cope with
human needs but not in conjunction
with new technology. In conclusion,
the law should easily materialize its
goal of public transportation
innovation and becomes a healer
from previous product failures.
METHOD
Presence of Healing Innovative

Products The presence of electronic
system-based product used by many
people is the actual social expectation
of society to improve the efficiency11 in
the utilization as well as covering the
existing weakness as the cause of the
failure in the maximum use of product
CONCLUSION
The evolution of ideal conditions initiated by

legal objectives is often along with the evolution
of threats to keep trying to maintain legal failure
to achieve legal objectives. Through efforts to
produce a concrete picture of the legal
objectives in the implementation of an online
public transportation system, the touch of
technology in terms of innovative products has
given rise to many forms of service from the
development of accessibility, so that smooth
information can influence the decision making.
Indeed, information technology provides a lot of
convenience for the law in realizing its purpose in
this public transportation innovation so that it
becomes a long-time wound healing which does
not recover from previous product failures, due
to its limitations which are actually constructed
by humans themselves.
Downloaded from SAE International by Kaveh Bakhsh Kelarestaghi, Monday, July 02, 2018
ARTICLE INFO
Article ID: 11-01-02-0004
Copyright © 2018
SAE International
doi:10.4271/11-01-02-0004
Intelligent Transportation
System Security: Hacked
Message Signs
Kaveh Bakhsh Kelarestaghi and Kevin Heaslip, Virginia Polytechnic Institute and State University
Volker Fessmann, Turner-Fairbank Highway Research Center
Majid Khalilikhah and Antonio Fuentes, Virginia Polytechnic Institute and State University
History
Abstract Received: 31 Dec 2017
“It cannot happen to us” is one of many common myths regarding cybersecurity in the transportation Revised: 23 Apr 2018
industry. The traditional view that the threats to transportation are low probability and low impact Accepted: 15 May 2018
keep agencies from mitigating security threats to transportation critical infrastructure. Current e-Available: 18 Jun 2018
transportation systems depend on closed proprietary systems, which are enhanced by connected
Keywords
cyber-physical systems. Variable Message Signs (VMS) deliver advisory information to road users
to ensure safe and efficient trips. Since the first VMS physical hacking more than a decade ago, the Intelligent Transportation
Systems, Cybersecurity,
importance of VMS security has been a pressing one. VMS hacks can include physical and remote
Message sign, Vulnerability,
breaches due to the weak protection of the signs and cyber-physical systems. In 2014, multiple
Risk assessment,
cyber-attacks on signs by “Sun Hacker” pushed the Department of Homeland Security (DHS), Attack tree,
which includes the Industrial Control Systems Cyber Emergency Response Team (ICS-CERT), and Cyber-physical systems
the Federal Highway Administration (FHWA) to investigate breaches more seriously. It is known security
that hackers breach cyber systems daily, but white hat hackers have given transportation officials
information to help them rethink Intelligent Transportation Systems (ITS) infrastructure security gaps Citation
to prevent harm to road users and financial losses. This study employs a risk-based approach to Kelarestaghi, K.,
conducting a threat assessment. This threat assessment performs a qualitative vulnerability-oriented Heaslip, K., Fessmann, V.,
threat analysis. The objective is to investigate safety, security, reliability, and operation issues that Khalilikhah, M. et al., “Intelligent
are triggered by compromised VMS. Additionally, countermeasures are proposed to prevent the Transportation System
failure of critical infrastructure. The outcome is anticipated to be of special interest and usefulness to Security: Hacked Message
policymakers and engineers concerned with the potential vulnerabilities of the ITS’s infrastructure. Signs,” SAE Int. J. Transp.
Cyber. & Privacy 1(2):2018,
doi:10.4271/11-01-02-0004.
ISSN: 2572-1046
e-ISSN: 2572-1054
1
2 Kelarestaghi et al. / SAE Int. J. Transp. Cyber. & Privacy / Volume 1, Issue 2, 2018
Introduction
A
ttacks to cyber-physical systems have targeted institutions and infrastructure
around the world, and the frequency and severity of attacks are on the rise [1,
2]. Healthcare manufacturing, financial services, education, government, and
transportation are among the industries that are the most lucrative targets for adversaries
[3, 4, 5]. Hacking is not just about companies, organizations, or banks; it also includes
transportation critical infrastructure. Wireless Sensor Networks (WSN), Vehicle-to-
everything communication (V2X), VMS and Traffic Signal Controllers (TSC) are among
major ITS infrastructure that has already been attacked or remain vulnerable to hacking.
ITS has been deployed with a focus on increasing efficiency and safety in the face of
dramatic increases in travel demand [6]. Any threat to ITS functionality compromises the
system at many levels, including regionally, nationally, and internationally [7]. Although
many studies have been performed and many security primitives have been proposed,
there are significant concerns about maintaining a flawless performance in a dynamic
environment. A holistic security approach [7], in which all infrastructure performs
within a satisfactory level of security, remains undiscovered.
Previously, hacking of road infrastructure was a rare event; however, in recent years,
field devices, such as VMSs, are hacked with higher frequency [8]. The primary reason
that transportation assets are vulnerable to cyber-attacks is their location in the trans-
portation network. Dramatic scenarios occur when hackers attempt to convey tampered
messages to the road users [8]. Conceivable future attacks using wireless communication
could lead to more sophisticated cyber-attacks, similar to the case of WSN, in which the
hacker exemplified a passive attack to compromise a traffic signal by accessing a WSN [9].
There is room for improvement by policymakers and roadway operators when consid-
ering critical infrastructure vulnerabilities. With cybersecurity issues escalating every
day, road users’ safety has been neglected [10]. Although manufacturers have attempted
minimal protection (i.e., isolation, hard-coded password), pranksters have been able to
successfully manipulate critical infrastructure (e.g., VMS, TSC, Vehicles, WSN).
This study attempts to examine the impacts of VMS security vulnerabilities, and
contributes to the literature of ITS security and resiliency, by exploring possible conse-
quences of an adversary hacking a VMS. To this aim, we synthesize current literature
and real-world VMS hacking events’ evidence that has been described in the current
literature and news reports, to pinpoint possible impacts that those attacks can impose
on the transportation network.
Additionally, this study undertakes a risk-based approach to perform a qualitative
vulnerability-oriented threat analysis with aims at identifying issues that are caused by
adversaries exploiting ITS security vulnerabilities. We will present a literature review
of the importance and effectiveness of VMS in the ITS network, followed by identifying
threat sources and surveying threat events which have occurred around the nation;
we then explore safety, operation and security issues ensued by VMS hacking incidents;
finally, we suggest countermeasures to prevent the failure of ITS critical infrastructure.
VMS Importance in ITS Network

VMS or Changeable Message Signs (CMS), also known as Dynamic Message Signs
(DMS), have been used for more than 50 years in the United States [11]. Advanced Traveler
Information Systems (ATIS), such as VMS, offer real time information to enhance drivers’
route choice by giving information on road traffic conditions, avoiding congestion,
selecting better departure times, and increasing network performance by prescribing
diversion decisions to motorists [12].
In contrast to traditional speed regulatory signs that have minimal influence on
drivers’ speed in work zone areas, implementation of VMS caused significant reduction in
vehicle speed [13]. Further analysis revealed changes in VMS’s efficiency corresponding to
message content [13]. Another study [14] comparing different speed control strategies in
© 2018 SAE International. All Rights Reserved.
Kelarestaghi et al. / SAE Int. J. Transp. Cyber. & Privacy / Volume 1, Issue 2, 2018 3
work zone areas indicated that VMS could significantly reduce the speed of an oncoming
vehicle and increase safety for construction crews as well as drivers themselves, thus
maximum efficiency is achieved by locating VMS in highly visible areas [15].
Route choice, speed changes, and braking behavior have been investigated for drivers
in which road closure information has been provided through VMS. The analysis indi-
cated a dramatic reduction in speed, and almost all drivers avoided the closed road
section [16]. Likewise, the investigation specified that message content and location of the
accidents displayed on the VMS are important factors about drivers’ diversion behaviors
[17]. Peeta et al. investigated the relation of VMS message type and drivers’ behavior.
The results showed that the influence of the message content has a strong correlation
with system performance [12].
Conventionally, real-time communication between VMS and Traffic Management
Centers (TMC) is conducted by voice-grade telephone lines, either wire-based or cellular,
including fiber optic cable and copper twisted pair cable [18]. The recent emergence of
technology sought traffic engineers to deploy advanced communication equipment.
Kosch et al. proposed VMS as part of Roadside Equipment (RSE) and should be equipped
with communication hardware [19]. The VMS using the connection to the ITS integrated
network and the Internet, offers to relay information to passing vehicles.
The credibility of VMS is extremely important to achieving efficient operations.
Drivers eventually will not pay attention to messages they distrust [11]. VMS that have
been tampered with provide unsanctioned information, which distracts motorists and
affects the credibility of the system. Recent VMS targeted by hackers were mainly carried
out by physical attacks and were perpetrated for the attackers amusement.
Methodology of Risk Assessment

The risk assessment approach comprises basic steps common to National Institute
of Standard and Technology (NIST) Special Publication 800-30 publication [20].
Implementing NIST SP 800-30 provides a guide for conducting an organizational
risk assessment. The risk assessment process take-a-ways will be a key constituent of
a risk management process [21]. The risk-based approach employs NIST SP 800-30
risk model to conduct a risk assessment to perform qualitative vulnerability-oriented
threat analysis. To conduct the vulnerability-oriented risk assessment, this study
(1) identifies exploitable security gaps of the system, (2) identifies threat events that
could exercise those security gaps, and (3) seeks impacts or consequences of a VMS
hacking incident.
The methodology of risk assessment (Figure 1) comprises of (1) identification of
threat sources and event, (2) identification of the system vulnerabilities and predisposing
conditions, and (3) determination of the adverse impacts and magnitude of impact. To
this aim, physical and cyber-attacks will be represented in a tree structure [22] (i.e., attack
tree), with the goal of tampering VMS content and different ways of achieving that goal.
The attack tree, proposed by Schneier [22], specifies a range of actions that an adversary
undertakes to exploit the vulnerabilities of the VMS. Representing the attack in a tree
FIGURE 1 Methodology of Risk Assessment (adapted from NIST SP 800-30 [20]).

© SAE International

FIGURE 2 Attack tree for the physical attack. structure provides a better understanding of all different ways
that an adversary can compromise the system.
Following the threat source and attack tree, threat events
will be identified through a comprehensive survey of the
VMS hacking incidents occurred throughout the United
States. Threat events then will be accompanied by the iden-
tification of system vulnerabilities and predisposing condi-
tions, and determination of impacts of those vulnerabilities
being exploited.
Threat Source and

Attack Tree
An outsider adversary can compromise VMS functionality
through physical and cyber-attacks. In a physical attack, an
adversary with limited resources and expertise gains physical
access to the portable VMS cabinet by either breaking the
lock or using a key, which the attacker may purchase online.
Subsequently, by entering the default usernames and pass-

words, the attacker can override a VMS message [23, 24, 25]. In
the case of cyber-attack, an outsider adversary with moderate
resources and expertise alters VMS’ content with no need for
physical access. This attack can be initiated by gaining remote
FIGURE 3 Attack tree for the cyber-attack.
access through the Telnet port 23 or the Simple Network
Management Protocol (SNMP) [26, 27]. Then the attacker
needs to brute force the password (e.g., guessing a default
password or using a password breaker tool) and alter the VMS
message. To alter the VMS content remotely, the adversary
can employ LCD simulator software. A high-level graphical
representation (i.e., attack tree) and term-based syntax for
both the physical (Figure 2) and cyber-attacks (Figure 3) are
presented to understand the risk potential systematically. The
following section surveys VMS hacking events.
Threat Events and System

Vulnerabilities
Physical-Attacks
One of the early incidents of VMS hacking took place
in Boston, Massachusetts in April 2007. MIT students
tampered the message to read, “This sign has been hack[e]d” (Figure 4) [28]. Although
the intention was not malicious, and no one was hurt, authorities considered it as a
future potential threat [29].
In February 2009, hackers tampered with the messages on two portable VMS [29]. It
was more about entertainment than the actual message. It was not difficult to gain access
to a portable VMS. The control unit uses a simple password and user interface protection,
making it easy to hack. A hacker could easily gain the knowledge to gain access from the
internet [23, 24, 25]. Also, the portable VMS cabinet can be unlocked effortlessly [29]
because the keys are available for purchase. A similar event happened in January 2009, when
another VMS was broken into and manipulated to display the message “Zombies Ahead.”
Officials declared that it was the first time that one of these signs was hacked [30, 31].

In January 2009, hackers were able to tamper with the VMS FIGURE 4 Hacked road sign in Boston, MA (courtesy
in Collinsville, Illinois during the morning peak period [32, 33]. of C. Pentacoff) [28].
A legitimate warning message was altered to read, “Daily Lane
Closures Due to Zombies.” Such behavior alerted authorities to the
drawbacks of ITS’s infrastructure security. This deficit in security
should be noted and worked on by traffic safety engineers and trans-
portation officials [32]. In February 2009, drivers in Indianapolis,
Indiana dealt with a hacked road sign displaying similar language.
The sign stated, “Raptors Ahead Caution.” Although one of the
drivers claimed that he did not pay attention to the sign, another
was excited about the content but was skeptical regarding the truth
of the message [34].
In 2009, several hacks happened in New York, where three
VMS were hacked in one day [35]. The signs should have warned
drivers to decrease their speed due to construction activities
ahead of them. However, because of their messages (e.g., “PARTY
AT JULIE’S”), drivers did not pay attention to the signs, which
could have resulted in hazardous consequences for drivers and
pedestrians [35]. It is worth noting if the attacks were performed
on a weekday instead of the weekend the results might have
been different.
In December 2009, during the morning rush hour in Gainesville, Florida, two signs
were hacked using a similar message. Before this occurrence, the University of Florida
website displayed information regarding a Zombie awareness plan as a joke [36]. The sign
was tampered with to read “Zombie Attack! Evacuate”. Some drivers presumed the sign
might be authentic [37]. The Florida Department of Transportation (FDOT) believed that
the hack could have been perpetrated using a telephone or by actual physical involve-
ment. The following year, Miami-Dade County in the Florida suffered a hacking incident,
where the sign message should have warned drivers of a road closure on Northwest 25th
Street [38] but instead displayed an offensive message.
In October 2012, Portland, Maine experienced a tampered message board that
worried drivers with the message “Zombies Ahead” instead of “Nightwork 8 pm-6 am”
[39]. A similar event took place in November 2012 in Loomis, California where a message
board displayed “Caution Loose Gorilla” [41]. In February 2014, a message board in a
work zone area was hacked in California. The construction company was forced to shut
down the sign until they could reprogram it. Such behavior is not just offensive; it can
put construction crews’ and drivers’ lives in danger [40].
In September 2015, another occurrence happened in Mililani, Hawaii where a VMS
was hacked with an offensive message that created confusion for drivers. An investigation
revealed that hacking such infrastructure is not complicated since security countermea-
sures are not sophisticated enough to prevent such malicious behaviors [42]. In October
2015, a VMS in Sacramento, California was hacked by a physical-attack that involved
breaking into the VMS cabinet and tampering with the computer [43].
Principally, hacking VMS generates two main problems. First, drivers can get
distracted which can lead to a crash. Second, drivers do not see the message that they
are supposed to read. Missing traffic information may cause severe consequences not just
to drivers but also to construction crews at work. Table 1 provides a survey of hacking
cases (sorted by date) across the US. Beyond the cases provided in Table 1, additional
VMS hacking events can be found in [44].
Although many people consider hacking message boards a rare occurrence, the
difficulties and issues are serious, and the hacking itself is not particularly complicated.
Instructions are accessible to anyone who wants to behave maliciously [23, 24, 25]. Watch
Dogs (released on May 2014), a home console video game, also teaches gamers how to
hack critical infrastructure [27]. Watch Dogs pinpoints security vulnerabilities and
allow a player (hacker) to outwit smart cities’ operating systems (e.g., security cameras,
power grids and traffic lights) for the hacker’s gain [45]. All the scenarios above are
conducted by “Threat Agents: Group One”; individuals that breached the signs aim for

TABLE 1 Summary of known hacking events in the U.S.

Date Location Warning message Type of tampered message Short term consequence Reference
Cyber attack
Asheville, NC
Winston-Salem, NC
May-June Mount Airy, NC [26, 27, 45, 49,
Traffic Information Amusing /Fame Driver distracted and confused
2014 State of New Jersey 50]
State of Iowa
State of Wyoming
Physical attack
Oct. 2015 Sacramento, CA Work Zone Ahead Amusing/Offensive Driver distracted [43]
Sept. 2015 Mililani, HI Work Zone Ahead Offensive Driver confusion [42]
July 2015 Tucson, AZ Road Closure Amusing Driver double-take [57]
Jan. 2015 Los Angeles, CA Work Zone Ahead Amusing /Offensive Drivers distracted [58, 59, 60, 61]
Feb. 2014 Granite Bay, CA Work Zone Ahead Offensive Driver distracted [40]
Nov. 2012 Loomis, CA Road Closure Amusing Driver distracted [41]
Oct. 2012 Portland, ME Work Zone Ahead Amusing Driver worried and distracted [39]
Traffic Information - No
Aug. 2011 Flagstaff, AZ Amusing Driver confusion [62, 63]
left turn at intersection
Warning Message for
May 2011 Falls Church, VA Amusing Driver distracted [64]
bicyclist and hikers
May 2010 Miami, FL Work Zone Ahead Offensive Driver distracted [38, 65]
Dec. 2009 Gainesville, FL Work Zone Ahead Amusing Driver distracted [37]
Mar. 2009 New York, NY Work Zone Ahead Amusing Driver distracted [35]
Feb. 2009 Hamilton County, IN Work Zone Ahead Amusing Driver confusion [34]
Feb. 2009 Collinsville, IL Work Zone Ahead Amusing Driver distracted [33]
Feb. 2009 Austin, TX Work Zone Ahead Amusing Destabilized traffic norm [29]
Jan. 2009 Austin, TX Work Zone Ahead Amusing Driver distracted and confused [30]
Apr. 2007 Boston, MA Work Zone Ahead Amusing Driver distracted [28]
“fun or notoriety” [46]. These individuals use their technical knowledge to manipulate
and breach systems such as VMS [47].
Cyber-Attack: “Sun Hacker”

Vulnerabilities of cyber systems need to be identified, reduced, mitigated, and elimi-
nated throughout the entire supply chain, to ensure the physical security of assets [10].
Protection must be addressed throughout entire system lifespan. Most of the cases
reported in Table 1 were physical hacks. However, in Asheville, NC, someone using
the pseudonym “Sun Hacker” claimed responsibility for the tampering (Figure 5). Five
overhead VMS were hacked in North Carolina in the Ashville, Winston-Salem, and
Mount Airy areas on Friday, May 30, 2014 [26, 27, 45]. During the same period, from
May 27 to June 2, VMSs in three other states (New Jersey, Iowa, and Wyoming) were
also hacked by the same entity [27]. The hacked VMS were operated using a web-based
interface, which allowed the hacker was able to access them remotely. The hacker stated
in his twitter account that he used Telnet Port 23 and a password breaker to breach the
signs -there is evidence suggesting video game “Watch Dogs” was a direct inspiration
for “Sun Hacker” [27, 51].
The hacker [52] claimed that the tampering was accomplished by accessing the
Virtual Private Network (VPN) of the VMS [48]. The Department of Homeland Security
(DHS), based on communication with VMS manufacturer, stated that the password was
not hard-coded [26, 53]. Subsequently, the Federal Highway Administration (FHWA)
provided guidelines to prevent such an incident. Several recommendations concerned

with weak device’s password, the accessibility of the IP addresses, FIGURE 5 VMS hacked by “Sun Hacker” in Ashville, NC
and privacy of the network and web pages [54]. (courtesy of WNCN) [27].
Although some states used precautions such as Friend List
(New Hampshire), VPN Tunnel (New Hampshire, Nevada),
Firewalls (Nevada), Message Validation and Authentication
(Maryland), the consistencies of these strategies should be tested
and validated to ensure that they are robust enough to secure the
network from cyber-attacks.
Interestingly, similar tampering could have more benefits
than drawbacks. Instead of hardening systems against attacks
by hackers with mischievous intention - so-called black hats -,
authorities could encourage white hat hackers - people with non-
destructive motives - to try to break into these systems. Hackers
breach cyber systems daily, but concurrently they could bounce
hints to authorities to rethink ITS’s infrastructure security gaps to
prevent any harm to road users and financial losses. Subsequently
exercising white hat hackers’ knowledge to identify network
vulnerabilities would be essential.
Discussion
Although most of the reported hacking events range from amusing to mildly offensive,
these are cases that compromise road safety. VMS provide necessary information to
drivers regarding traffic conditions and the road ahead of them. Such information
might be related to road conditions and closures, construction zones, accidents, and
detours, among other things. Losing or neglecting such information could lead to a
disaster. The hacking which happened in Austin (February 2009) [29], was close to a high-
density trafficked boulevard. Although no incidents were reported for that particular
message, potential impacts that VMSs hacking might impose on driver’s behavior need
to be investigated. It is noteworthy that many studies presented results demonstrating
the significant correlation between the content of the VMS message and driver behavior
[12, 13, 14, 16, 17, 55].
Hacked VMS were investigated with regards to the impact on vehicles’ location
in a roadway and the function of their travels. Olofsson [77] defined VMS hacking as
“a means of combating the disciplinary regimes of roadways” that can destabilize the
correlation between vehicles’ location in the roadway, possible events ahead of their
trip, their destination, and ultimately jeopardize the institutionalized function of
the travel. Such an example can be seen in the Austin case (June 2010), when drivers
looked twice when confronting the sign reading, “NAZI ZOMBIES AHEAD!” on a
VMS. Some slowed down, and some even took pictures [56]. This reaction is concerning
because driver distraction can cause an increase in the risk of a crash, carbon emis-
sions, energy consumption. Security threats on VMS are inevitable, yet we cannot say
for sure what consequences such an attack can impose. Although researchers have
attended to the security aspect of the ITS, unfortunately very little attention has been
dedicated to assessing the impact of security breaches on the transportation network,
more specifically on the system’s operators and users. This limitation creates a need for
future research to assess the impact of cybersecurity vulnerabilities with data-driven
approaches [55, 79, 80, 82].
The threat events listed in Table 1 indicate that most of the VMS were seeking to
promote awareness regarding an upcoming construction zone. VMS not only provide
information regarding slower speed in work zones, but they are also able to enhance
safety, especially related to rear-end accidents, by displaying real-time information of
a queue-end location to alert drivers of upcoming slow-moving or stopped traffic [66].
Information provided by VMS in work zones should warn drivers to take necessary
precautionary actions. Such situations can be even more complicated when dealing
with different types of vehicles and emphasize the importance of functioning VMS [67].

Queue propagation speed, geometric conditions, and the familiarity of drivers are
important factors correlated with an accident’s intensity. Lack of real-time information
makes drivers more vulnerable to unforeseen traffic conditions [67]. One of the conse-
quences of an information deficit is rear-ended crashes. A naturalistic driving study
comprised of 100 cars found that almost 78% of total crashes and 65% of near-crashes
were related to driver inattention [68]. Out of all rear-end crashes, 87% involved driver
distraction [69]. In 2012, more than 1.7 million rear-end crashes that caused more than
1,700 fatalities were reported in the U.S. [70]. In addition to driver distraction, the unpre-
paredness of drivers to decrease speed in work zones were identified as another cause
of rear-end crashes [70]. This is concerning, since the rear-end crashes have significant
impact on crash severity [81].
Implementation of Automated Work Zone Information Systems (AWIS) is reported
to be a reliable and effective way to enhance work zone safety [71, 72]. However, the
comparison of AWIS information and the field observation revealed that, in situations
where information is not accurate, drivers become confused and some behave based on
their judgment and experience [73]. The discrepancy between provided information and
the actual condition results in drivers performing evasive behaviors in work zones that
increase the likelihood of a crash [74].
Adverse Impact and Countermeasures

A valid concern might not be over a single hacking scenario; rather about the penetration
of an integrated ITS network from the point that is most susceptible to a cyber-attack. As
the necessity for inclusive ITS implementation increases, the security issues regarding
cyber-attacks have gained more attention. In some cases, conventional VMSs were only
threatened by physical attacks, and protection was not difficult. However, an integrated
ITS network requires an advanced communication setup [19, 75], making security far
more difficult. Some challenges that security advancements face may include driver
distraction or confusion, incidents, congestion, fatalities, financial losses, and harm to
agencies and their employees. Dealing with cyber-attacks demands enhanced coordina-
tion and management, as these are serious matters that affect people’s lives.
Once a hacker breaches the system, which VMS becomes a malicious node in the
TMC network. From this first foothold, hackers could easily penetrate further and
expand their reach into the network. Spyware could be installed to steal usernames and
passwords, and later that information could feed Group 2 (e.g., ransomware viruses) and
Group 3 (terrorists engaged in cyber warfare) threat agents for exploitation and egress
[46]. It is best to create a robust security system in advance, rather than to consider after
VMS implementation, to save time and money. Moreover, with the rise of the connected
vehicle technology, adversaries could exploit vulnerabilities in existing, traditional
ITS installations, thus being able to compromise V2X installations despite their build
in security measures. Fabricated information could mislead vehicles to cause adverse
impacts (e.g., crashes) on road users and operators [76].
This survey is of special interest and usefulness to engineers concerned with
safety, operation and security issues that have been ensured by VMS hacking incidents.
Lessons learned from almost all the cases above depict that drivers were distracted by
the messages displayed on VMS; they either stopped or reduced vehicle speed and lost
concentration on the road for a short period. Driver distraction (e.g., to outside persons
and events) is one of the main causes of driving errors leading to crashes. On the other
side, tampered VMSs need to be fixed by authorities (e.g., Police, TMC) and will dictate
unexpected operational costs (e.g., labor costs) to the system. Besides, delay to adjust
the issues would affect system operational success and results in road users’ distrust of
a system. Consequently, VMS’ security shall design-fitted instead of retrofitted which
possibly costs considerably higher.
A 3 × 3 impact-likelihood matrix represented in Figure 6, attempts to map
adverse impacts of the VMS hacking events onto a meaningful, visual, matrix for risk

FIGURE 6 The impact-likelihood matrix maps the adverse impacts of VMS hacking
into high (colored in dark gray), medium (colored in light gray) and low-risk zones
(colored in white).
LIKELIHOOD
Low Medium High
Security Impact: VMS becomes Safety Impact: Crashes with
Behavioral Impact: Impulsive
High
a malicious node in the TMC severity level form Property

drivers' behavior
network damage to Injuries and fatalities
IMPACT
Operational Impact: Traffic

Medium
Operational Impact: System

congestion, increase in travel
operators' financial losses

time
Efficiency Impact: Increase in Reliability Impact: Road users’

Low
energy consumption distrust of a system
prioritization purposes. Three impact and likelihood classes (Low, Medium and High)
were identified to assist decision-makers to prioritize the risks that are associated with
hacked VMS. The impact-likelihood matrix maps the adverse impacts of VMS hacking
into high (colored in dark gray), medium (colored in light gray) and low-risk zones. Risk
assessment result indicates that (1) safety, operational (i.e., monetary loss) and behavioral
impacts are associated with the high-risk cluster, and (2) security, reliability, efficiency
and operational (i.e., congestion) impacts are associated with the medium-risk cluster.
Noteworthy that the impact-likelihood matrix is concluded based on the highest-level
attacks, which represented in Figures 2 and 3.
A few countermeasures should be considered by manufacturers and system opera-
tors, to mitigate against threats to VMSs [26]. These recommendations include:
1. to place the display on a private network or VPN,
2. to deactivate unnecessary telnet, webpage and LCD interfaces,
3. to avoid using hard-coded/default passwords, but instead secure the access with
strong and complex password,
4. to minimize network exposure,
5. to isolate control network from business network,
6. to secure the remote access for authorize users,
7. to implement an authentication mechanism against physical attacks,
8. to upgrade SNMP to the most current version,
9. to enable remote logging and Monitor the logs, and
10. to change all SNMP community string from the default.
This study aimed to raise awareness among traffic engineers to comprehend and
utilize best practices of cyber hygiene (steps to enhance cybersecurity), access control,
risk management, information security and monitoring [10]. Network redundancy is
required in advance to prevent the failure of critical infrastructures. Intrusion detection
systems (IDS) are effective in identifying odd activities in the network. Also, encryp-
tion methods can secure the network communications and stop hackers from easily
manipulating the infrastructures. More prominently a practical approach would be to
exercise white hat hackers’ skills to constantly monitor vulnerabilities of the ITS critical
infrastructure before adversaries could breach the system [78].
The result of this study supports risk response decision and might affect (1) ITS
security practices, policies, and guidance, (2) selection of common ITS security controls,
(3) ITS design, implementation and operational decisions, and (4) development of risk-
aware training. The main objective of this study is not to comprehensively study such
dynamic subject but is to raise awareness among policy-makers and engineers system-
atically. Awareness is a fundamental step in making transportation secure and resilient

against cyber-physical attacks. Secure and resilient transportation infrastructure enables

people and goods to be transported without significant disruption and frees resources
to make transportation safer and more efficient.
Conclusion
As more and more intelligence is applied in the field of transportation, cybersecurity
threats are more serious than ever before. These shifts in intelligence include introducing
connected vehicle and automated vehicle technologies. Broad integration of such intel-
ligent transportation systems, can make transportation faster, safer, more reliable, and
more convenient. However, unanswered controversies remain, including security issues
that demand attention before any large-scale implementation should take place. The
assumption of the trustworthiness of ITS network fails at the presence of any vulner-
ability, leading to brittle protection. For a future integrated ITS network, a hacked VMS
is a malicious node where group two and three threat agents would prolong their reach
into the network. Once access to a single node in a network is gained, adversaries can
launch various attacks that eventually can fail the system dramatically. In this study,
we outlined potential consequences of the VMS security breaches due to the cyber and
physical attacks. The concerns above need to be assessed more in-depth through crowd-
sourcing or simulation studies. There is a gap in literature to assess and evaluate the
impact of security vulnerabilities on the road users and the system operators. The result
of this study endorses the necessity of impact assessment on the VMS security breach.
Risk assessment is a key component of the risk management process, and since risk
management is an ongoing process, risk assessment should be conducted throughout
the system development lifecycle. In this study, we employed the NIST SP 800-30 risk
model to conduct a risk assessment to perform qualitative vulnerability-oriented threat
analysis. We summarize the key findings as follows:
•• The risk-based approach of this study delivers the impact-likelihood matrix, which
maps the adverse impacts of the threat events onto a meaningful, visual, matrix.
The result provides insights for system operators and decision-makers to prioritize
the risk of a VMS hacking event.
•• VMS hacking adverse impacts can be categorized mainly as high-risk and
medium-risk clusters. The safety, operational (i.e., monetary losses) and behavioral
impacts are associated with a high-risk cluster. While the security, reliability,
efficiency, and operational (i.e., congestion) impacts are associated with the
medium-risk cluster.
•• The VMS security vulnerabilities and predisposing conditions allow adversaries
to compromise the ITS functionality. System operators should consider adequate
physical and cyber security measures to improve transportation critical
infrastructure security and resiliency. Moreover, at the same time, since threats on
ITS are inevitable, impact assessment could guide decision-makers on the adverse
impact of a security breach.
It needs to be understood that with the progression of technology, the operator’s

understanding must also change. With the increasing complexity and the integration
of the system, operators have to look at this situation on a system level instead of as a
collection of isolated incidents. Future ITS implementations need to be designed with
adequate security in mind from inception. At this time, it can be concluded that the
physical and cyber hackings of VMS create the slowdown of traffic, they also have the
potential to threaten road users’ safety and to create financial losses in the affected
communities. Crashes, fatalities, congestion, and public chaos are among the possible
outcomes of tampering with transportation network critical infrastructures [78]. Sudden
changes in drivers’ behavior while passing a tampered message sign could lead to devas-
tating incidents. Also, from an operation and security standpoint, authorities need to
foresee the situation to plan efficient countermeasures to minimize the risk of partial
or complete losses of the system. Further research is needed to assess the risk conferred
by hacked VMS and the messages they convey. Also, countermeasures proposed here
(e.g., encryption and IDS) need to be studied and prioritized as potential long-term and
short-term strategies.
Contact information
Kaveh B. Kelarestaghi
The Charles Edward Via, Jr. Department of Civil and Environmental Engineering,
Virginia Polytechnic Institute and State University, Arlington, VA 22203
kavehbk@vt.edu
Acknowledgement
The authors would like to thank Edward Fok from Federal Highway Administration for
his input and help in preparing this article. In addition, we would like to thank three
anonymous reviewers for their valuable comments.
References
1. Graham, L., “Cyberattacks Are Surging and More Data Records Are Stolen,” 2017,
[Online], https://www.cnbc.com/2017/09/20/cyberattacks-are-surging-and-more-data-
records-are-stolen.html, accessed Feb. 14, 2018.
2. Huff Post, “With Cyber Attacks on the Rise, Here’s How Companies Can Mitigate Their
Risks,” 2017, [Online], https://www.huffingtonpost.com/entry/with-cyber-attacks-are-on-
the-rise-heres-how-companies_us_59c1787ae4b0c3e70e742883, accessed Feb. 14, 2018.
3. Morgan, S., “Top 5 Industries at Risk of Cyber-Attacks,” 2016, [Online], https://
www.forbes.com/sites/stevemorgan/2016/05/13/list-of-the-5-most-cyber-attacked-
industries/#8831db3715e9, accessed Feb. 14, 2018.
4. Polyakof, A., “What Cyberthreats Do Higher Education Institutions Face?,” 2017,
[Online], https://www.forbes.com/sites/forbestechcouncil/2017/08/21/what-cyberthreats-
do-higher-education-institutions-face/#24b35d7b640d, accessed Feb. 14, 2018.
5. Dahlberg, D., “Ransomware Cyber Attacks: Which Industries Are Being Hit the
Hardest?,” 2017, [Online], https://www.bitsighttech.com/blog/ransomware-cyber-attacks,
accessed Feb. 14, 2018.
6. Mimbela, L.E.Y. and Klein, L.A., “Summary of Vehicle Detection and Surveillance
Technologies Used in Intelligent Transportation Systems,” Handbook, 2000.
7. Dellios, K., Papanikas, D., and Polemi, D., “Information Security Compliance over
Intelligent Transport Systems: Is IT Possible?” IEEE Security and Privacy 3:9-15, 2015.
8. Fok, E., “An Introduction to Cybersecurity Issues in Modern Transportation Systems,”
Inst. Transportation Engineers J 83(7):18-21, 2013.
9. Cerrudo, C., “Hacking US (and UK, Australia, France, etc.) Traffic Control Systems,”
IOActive Labs Research, [Online], http://blog.ioactive.com/2014/04/hacking-us-and-uk-
australia-france-etc.html, accessed July 30, 2016.
10. CASE and WMC, “Protection of Transportation Infrastructure from Cyber Attacks:
A Primer, National Academies of Sciences, Engineering, and Medicine,” NCHRP
Document 221, 2015.
11. Dudek, C.L., “Changeable Message Sign Operation and Messaging Handbook,”
Publication FHWA-OP-03-070, FHWA, U.S. Department of Transportation, Aug. 2004.
12. Peeta, S., Ramos, J.L., and Pasupathy, R., “Content of Variable Message Signs and On-
Line Driver Behavior,” in Transportation Research Record: Journal of the Transportation
Research Board, No. 1725, TRB, National Research Council, Washington, DC, 2000,
102-108.
13. Garber, N.J. and Patel, S.T., “Control of Vehicle Speeds in Temporary Traffic Control Zones
(Work Zones) Using Changeable Message Signs with Radar,” in Transportation Research
Record No. 1509, TRB, National Research Council, Washington, DC, 1995, 73-81.
14. Wang C., Dixon, K., and Jared, D., “Evaluating Speed-Reduction Strategies for Highway
Work Zones,” in Transportation Research Record: Journal of the Transportation
Research Board, No. 1824, Washington, DC, 2003, 44-53.
15. Atkinson, J., Bauer, J., Hunt, K., Mullins, K. et al., “Designing for Transportation
Management and Operations: A Primer,” Report FHWA-HOP-13-013, SAIC and Gannett
Fleming, Inc. for Federal Highway Administration, McLean, VA, 2013.
16. Erke, A., Sagberg, F., and Hagman, R., “Effects of Route Guidance Variable Message Signs
(VMS) on Driver Behaviour,” Transportation Research F: Traffic Psychology and Behaviour
10:447-457, 2007.
17. Chatterjee, K., Hounsell, N.B., Firmin, P.E., and Bonsall, P.W., “Driver Response to
Variable Message Sign Information in London,” Transportation Research C: Emerging
Technologies 10(2):149-169, 2002.
18. WisDOT, “Intelligent Transportation Systems Design Manual,” Wisconsin Department
of Transportation, Madison, 2000.
19. Kosch, T., Kulp, I., Bechler, M., Strassberger, M. et al., “Communication Architecture
for Cooperative Systems in Europe,” IEEE Communications Magazine 47(5):116-
125, 2009.
20. Ross, R.S., “Guide for Conducting Risk Assessments,” No. Special Publication (NIST SP)-
800-30 Rev 1, 2012.
21. Initiative, NIST Joint Task Force Transformation, “Managing Information Security Risk:
Organization, Mission, and Information System View,” NIST Special Publication, 2011,
800-839.
22. Schneier, B., “Attack Trees,” Dr Dobb’s J. 24(12):21-29, 1999.
23. Stern, B., “HOW TO - Hack Construction Signs,” Make: DIY Projects, How-
Tos, Electronics, Crafts and Ideas for Makers, 2009, [Online], http://makezine.
com/2009/01/27/how-to-hack-construction-signs/, accessed Nov. 6, 2015.
24. Wojdyla, B., “How to Hack an Electronic Road Sign,” 2009, [Online], http://
jalopnik.com/5141430/how-to-hack-an-electronic-road-sign?trending_
test_d&utm_expid=66866090-62.H_y_0o51QhmMY_tue7bevQ.4&utm_
referrer=http%3A%2F%2Fjalopnik.com%2F5144628%2Ftraffic-sign-hacking-spreads-to-
indiana%3Ftrending_test_d, accessed Nov. 6, 2015.
25. Kraft, C., “Road Sign Hacking, Hackaday,” 2009, [Online], http://hackaday.
com/2009/01/24/road-sign-hacking/, accessed Nov. 10, 2015.
26. ICS-CERT, “Daktronics Vanguard Default Credentials (Update A),” 2014, [Online],
https://ics-cert.us-cert.gov/alerts/ICS-ALERT-14-155-01A, accessed Feb. 12, 2018.
27. Krebs, B., “They Hack Because They Can - Krebs on Security,” 2014, [Online], http://
krebsonsecurity.com/2014/06/they-hack-because-they-can/comment-page-1/, accessed
Nov. 11, 2015.
28. Interesting Hacks to Fascinate People: The MIT Gallery of Hacks, “This Sign Has Been
Hacked,” 2007, [Online], http://hacks.mit.edu/Hacks/by_year/2007/sign_hacked/index.
html, accessed Dec. 15, 2015.
29. Fermoso, J., “Austin Road Signs Hacked, Warn of Nazi Zombies and World’s End,”
Wired, 2009, [Online], http://www.wired.com/2009/02/austin-road-sig/, accessed Oct.
16, 2015.
30. Miller, J.R., “Hackers Crack into Texas Road Sign, Warn of Zombies Ahead,” Fox News,
2009, [Online], http://www.foxnews.com/story/2009/01/29/hackers-crack-into-texas-
road-sign-warn-zombies-ahead.html, accessed Dec. 15, 2015.
31. Impact Recovery Systems, “Road Sign Hacking: Harmless Prank or Safety Threat?,” 2009,
[Online], http://www.impactrecovery.com/resources/road_sign_hacking_harmless_
prank_or_safety_threat/, accessed Dec. 16, 2015.
32. Metro News, “Another Road Sign Warns of Zombies,” 2009, [Online], http://metro.
co.uk/2009/02/05/another-road-sign-warns-of-zombies-432840/, accessed Dec. 15, 2015.
33. NY Daily News, “Midwest Road Sign Prank Warns Drivers of Zombies,” 2009, [Online],
http://www.nydailynews.com/news/world/midwest-road-sign-prank-warns-drivers-
zombies-article-1.388519, accessed Nov. 6, 2015.
34. WRTV, “Raptors Ahead’ Sign Gets Stares, Chuckles,” 2009, [Online], http://www.
theindychannel.com/news/-raptors-ahead-sign-gets-stares-chuckles, accessed Nov. 6, 2015.
35. Soltis, A., “Making Light of the Law,” New York Post, 2009, [Online], http://nypost.
com/2009/03/17/making-light-of-the-law/, accessed Nov. 6, 2015.
36. Crabbe, N., “Thank goodness! UF Has a Plan for Zombie Invasions,” The Gainesville Sun,
2009, [Online], http://www.gainesville.com/article/20091002/articles/910021006, accessed
Dec. 15, 2015.
37. Byrnes, J. and Voyles, K., “Don’t Fear: Zombies Are Not Near,” The Gainesville Sun, 2009,
[Online], http://www.gainesville.com/article/20091222/articles/912221017, accessed Nov.
6, 2015.
38. Turnbell, M. and Sentinel, S., “Miami Construction Sign Hacked to Read ‘No Latinos,’ ‘No
Tacos’,” Sun Sentinel, 2010, [Online], http://articles.sun-sentinel.com/2010-05-26/news/fl-
roadsigns-hacked-20100525_1_sign-hacked-construction-site, accessed Nov. 6, 2015.
39. Leibowitz, B., “Zombies Ahead’: Construction Sign Tells Maine Drivers to Be Very
Afraid,” CBS News, 2012, [Online], http://www.cbsnews.com/news/zombies-ahead-
construction-sign-tells-maine-drivers-to-be-very-afraid/, accessed Nov. 6, 2015.
40. Sharp, R., “Construction Board Hacked in Granite Bay, Changed to Obscene Saying,”
News - KCRA Home, 2014, [Online], http://www.kcra.com/news/construction-board-
hacked-in-granite-bay-changed-to-obscene-saying/24277104, accessed Nov. 6, 2015.
41. Leibowitz, B., “Caution Loose Gorilla’: Electronic Road Sign Hacked in Northern
California,” CBS News, 2012, [Online], http://www.cbsnews.com/news/caution-loose-
gorilla-electronic-road-sign-hacked-in-northern-california/, accessed Nov. 6, 2015.
42. Cerball, A., “Mililani Digital Road Sign Hacked with Offensive Message,” KHON2, 2015,
[Online], http://khon2.com/2015/09/19/mililani-digital-road-sign-hacked-with-offensive-
message/, accessed Nov. 6, 2015.
43. Bizjak, T., “Dude! What Did That Traffic Sign Say?,” The Sacramento Bee, 2015, [Online],
http://www.sacbee.com/news/local/transportation/back-seat-driver/article39697362.
html, accessed Nov. 6, 2015.
4 4. Stuff You Should Know, “23 Great Hacked Road Signs,” 2014, [Online], http://www.
stuffyoushouldknow.com/blog/gallery/hacked-road-signs/, accessed Dec. 15, 2015.
45. Tassi, P., “Aiden Pearce, Snooping Superhero: The Strange Moral Compass of ‘Watch
Dogs’,” Forbes, 2014, [Online], https://www.forbes.com/sites/insertcoin/2014/05/28/aiden-
pearce-snooping-superhero-the-strange-moral-compass-of-watch-dogs/#33fb59d9588,
accessed Dec. 20, 2017.
46. Fok, E., “Cyber Security Challenges: Protecting Your Transportation Management
Center,” ITE Journal 85(2):32-36, 2015.
47. U.S. Department of Homeland Security, “Strategy for Securing Control Systems,” U.S.
Department of Homeland Security, Washington, DC, 2009.
48. WLOS News 13, “D.O.T. Signs Hacked in Asheville & Statewide,” YouTube, 2014,
[Online], https://www.youtube.com/watch?v=YJG9fRv_6Js, accessed Nov. 11, 2015.
49. WYFF4, “Computer Hackers Target Electronic Highway Signs,” 2014, [Online], http://
www.wyff4.com/news/computer-hackers-target-electronic-highway-signs/26263834,
accessed June 28, 2016.
50. Nextgov, “Hacker Rewrites 5 Highway Signs in North Carolina,” 2014, [Online], http://
www.nextgov.com/cybersecurity/threatwatch/2014/06/password-cracking-network-
intrusion-hacker/1105/, accessed June 28, 2016.
51. Hill, K., “Watch Dogs’ Video Game Will Inspire More ‘Godzilla Attack’ Road Sign
Hacks, Says Cybersecurity Warning,” Forbes, 2014, [Online], https://www.forbes.com/
sites/kashmirhill/2014/06/09/watch-dogs-road-sign-hacks/#3b63c3174e55, accessed Feb.
12, 2018.
52. SUN HACKER (@ISUN_HACKER) | Twitter,” [Online], https://twitter.com/isun_hacker,
53. Sternstein, A., “Flaw Lets Hackers Control Electronic Highway Billboards,” Nextgov.com,
2014, [Online], http://www.nextgov.com/cybersecurity/2014/06/flaw-lets-hackers-control-
electronic-highway-billboards/85849/, accessed June 28, 2016.
54. Traffic Technology International, “Godzilla Attack’ Prompts DMS Recommendations,”
2014, [Online], http://www.traffictechnologytoday.com/news.php?NewsID=59883,
55. Jeihani, M., NarooieNezhad, S., and Kelarestaghi, K.B., “Integration of a Driving
Simulator and a Traffic Simulator Case Study: Exploring Drivers’ Behavior in Response to
Variable Message Signs,” IATSS Res. 41(4):164-171, 2017.
56. KXAN, “Hacked Road Signs in Austin,” YouTube, 2010, [Online], https://www.youtube.
com/watch?v=1Lw0WMyChrM, accessed Oct. 16, 2015.
57. Marries, D., “Tucson Digital Traffic Sign Hacked, Warns of ‘Zombies Ahead’,” Tucson
News Now, 2015, [Online], http://www.tucsonnewsnow.com/story/29466781/tucson-
digital-traffic-sign-hacked-warns-of-zombies-ahead, accessed Dec. 15, 2015.
58. Romero, D., “L.A. Traffic Sign Is Hacked to Say ‘Read a F--ing Book’ (PHOTOS),” L.A.
Weekly, 2015, [Online], http://www.laweekly.com/news/la-traffic-sign-is-hacked-to-say-
read-a-f-ing-book-photos-5331670, accessed Nov. 10, 2015.
59. O’Brien, L., “Hackers Made an LA Traffic Sign Say ‘READ A FUCKING BOOK’,”
MobyLives, 2015, [Online], http://www.mhpbooks.com/hackers-made-an-la-traffic-sign-
say-read-a-fucking-book/, accessed: Dec. 15, 2015.
60. Watson, P., “Los Angeles Traffic Sign Hacked with Obscene Message (Photo),” 2015,
[Online], http://fansided.com/2015/01/11/los-angeles-traffic-sign-hacked-obscene-
message-photo/, accessed Dec. 15, 2015.
61. AOL News, “Downtown Los Angeles Traffic Sign Hacked,” 2015, [Online], http://www.
aol.com/article/2015/01/10/downtown-los-angeles-traffic-sign-hacked/21127825/,
62. The Associated Press, “Hacked Road Sign Warns Flagstaff Drivers of Panda Rampage,”
2011, [Online], http://tucson.com/news/state-and-regional/hacked-road-sign-warns-
flagstaff-drivers-of-panda-rampage/article_8c5e1e72-cda7-11e0-a057-001cc4c03286.html,
63. Daily Sun, “Police: No Rogue Pandas About,” 2011, [Online], http://azdailysun.com/news/
local/police-no-rogue-pandas-about/article_420be32f-7571-5507-9ce9-58b6f6ea8d4f.
html, accessed Dec. 15, 2015.
64. Fard, M.F., “Caution Zombies Ahead’: N. Va. Electronic Sign Hacked,” The Washington
Post, 2011, [Online], https://www.washingtonpost.com/blogs/the-buzz/post/caution-
zombies-ahead-n-va-road-sign-hacked/2011/05/11/AFPObEsG_blog.html, accessed Dec.
15, 2015.
65. Detwiller, B., 'Hacked Miami Road Sign Highlights Password and Physical Security
Lapses,” TechRepubli, 2010, [Online], http://www.techrepublic.com/blog/tr-dojo/hacked-
miami-road-sign-highlights-password-and-physical-security-lapses/, accessed Nov.
10, 2015.
66. Khan, A.M., “Intelligent Infrastructure-Based Queue-End Warning System for Avoiding
Rear Impacts,” IET International Transport Systems 1(2):138-143, 2007.
67. Wiles, P.B., Cooner, S.A., Walters, C.H., and Pultorak, E.J., “Advance Warning of Stopped
Traffic on Freeways: Current Practices and Field Studies of Queue Propagation Speeds,
No. FHWA/TX-03/4413-1, 2003.
68. Klauer, S.G., Dingus, T.A., Neale, V.L., Sudweeks, J.D. et al., “The Impact of Driver
Inattention on Near-Crash/Crash Risk: An Analysis Using the 100-Car Naturalistic
Driving Study Data,” 2006.
69. Lee, S.E., Llaneras, E., Klauer, S., and Sudweeks, J., “Analyses of Rear-End Crashes and
Near-Crashes in the 100-Car Naturalistic Driving Study to Support Rear-Signaling
Countermeasure Development,” DOT HS, Vol. 810, 2007, 846, National Highway Traffic
Safety Administration, Washington, DC.
70. National Transportation Safety Board, “The Use of Forward Collision Avoidance Systems
to Prevent and Mitigate Rear-End Crashes,” NTSB/SIR-15/01 PB2015-104098, May 2015.

71. Tudor, L., Meadors, A., and Plant, R., “Deployment of Smart Work Zone Technology in
Arkansas,” in Transportation Research Record: Journal of the Transportation Research
Board, No. 1824, TRB, National Research Council, Washington, DC, 2003, 3-14.
72. Chu, L., Kim, H.-K., Chung, Y., and Recker, W., “Evaluation of Effectiveness of
Automated Work Zone Information Systems,” in Transportation Research Record:
Journal of the Transportation Research Board, No. 1911, Transportation Research Board
of the National Academies, Washington, DC, 2005, 73-81.
73. Tooley, M., Gattis, J., Janarthanan, R., and Duncan, L., “Evaluation of Automated
Work Zone Information Systems,” in Transportation Research Record: Journal of the
Transportation Research Board, No. 1877, Transportation Research Board of the National
Academies, Washington, DC, 2004, 69-76.
74. Morris, T., Schwach, J.A., and Michalopoulos, P.G., “Low-Cost Portable Video-Based
Queue Detection for Work-Zone Safety,” Retrieved from the University of Minnesota
Digital Conservancy, http://hdl.handle.net/11299/101445, 2011.
75. GovMobile, “Wireless M2M Solutions for Changeable Message Signs for Traffic
Managers,” 2015, [Online], http://govmobile.com/author/govmobile/, accessed Dec.
16, 2015.
76. Muhammad, M. and Safdar, G.A., “Survey on Existing Authentication Issues for Cellular-
Assisted V2X Communication,” Veh. Commun. 12:50-65, 2018.
77. Olofsson, J., “‘Zombies Ahead! ‘A Study of How Hacked Digital Road Signs Destabilize
the Physical Space of Roadways,” Visual Communication 13(1):75-93, 2014.
78. Kelarestaghi, K.B., Heaslip, K., and Gerdes, R., “Vehicle Security: Risk Assessment in
Transportation,” ArXiv Preprint: ArXiv:1804.07381, 2018.
79. Dabiri, S. and Heaslip, K., “Transport-Domain Applications of Widely Used Data Sources
in the Smart Transportation: A Survey,” ArXiv Prepr. ArXiv180310902, 2018.
80. Dabiri, S. and Heaslip, K., “Inferring Transportation Modes from GPS Trajectories Using
a Convolutional Neural Network,” Transp. Res. Part C Emerg. Technol. 86:360-371, 2018.
81. Kelarestaghi, K.B., Zhang, W., Wang, Y., Xiao, L. et al., “Impacts to Crash Severity
Outcome due to Adverse Weather and Other Causation Dactors,” Advances in
Transportation Studies 43:31-42, 2017.
82. Shabanpour, R., Auld, J., Mohammadian, A.K., and Stephens, T.S., “Developing a
Platform to Analyze Behavioral Impacts of Connected Automated Vehicles at the
National Level,” No. 17-06283, 2017.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior written permission of SAE International.
Positions and opinions advanced in this article are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the
content of the article.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/329446999
An integrated approach based on system dynamics and ANP for evaluating

Article in International Journal of Systems Science Operations & Logistics · December 2018

DOI: 10.1080/23302674.2018.1554168
CITATIONS READS
38 290
2 authors:
Reza Sayadi Anjali Awasthi

York University Concordia University Montreal
4 PUBLICATIONS 93 CITATIONS 141 PUBLICATIONS 2,945 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Call for book chapters : Multicriteria Decision Models for Systems Engineering View project
Call for Chapters: Interdisciplinary Approaches to Decision Making for Sustainable Supply Chains View project
All content following this page was uploaded by Anjali Awasthi on 30 August 2020.
The user has requested enhancement of the downloaded file.

International Journal of Systems Science: Operations &
Logistics
ISSN: 2330-2674 (Print) 2330-2682 (Online) Journal homepage: https://www.tandfonline.com/loi/tsyb20
An integrated approach based on system

dynamics and ANP for evaluating sustainable
transportation policies
Reza Sayyadi & Anjali Awasthi
To cite this article: Reza Sayyadi & Anjali Awasthi (2018): An integrated approach based on
system dynamics and ANP for evaluating sustainable transportation policies, International Journal
of Systems Science: Operations & Logistics, DOI: 10.1080/23302674.2018.1554168
To link to this article: https://doi.org/10.1080/23302674.2018.1554168
View supplementary material
Published online: 05 Dec 2018.
Submit your article to this journal
Article views: 101
View related articles
View Crossmark data
Citing articles: 1 View citing articles
Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tsyb20
INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE: OPERATIONS & LOGISTICS
https://doi.org/10.1080/23302674.2018.1554168
An integrated approach based on system dynamics and ANP for evaluating

Reza Sayyadi and Anjali Awasthi
CIISE, Concordia University, Montreal, Canada
ABSTRACT ARTICLE HISTORY

Transportation systems are growing and complex systems. The presence of multiple, correlated, Received 8 December 2017
dynamically changing elements in this system with dependence and feedback add further com- Accepted 26 November 2018
plexity to the problem. In this paper, we present an integrated approach based on system dynamics KEYWORDS
(SD) simulation and analytic network process (ANP) for evaluating sustainable transport policies. Five Sustainable transportation;
policies namely trip sharing (TRS), trip rate reduction (TRR), reducing the length of the road network policy evaluation; multi
(LRN), car ownership (CAO), and average kilometres travelled (AKT) are evaluated against three cri- criteria decision analysis;
teria namely congestion level (CONG), fuel consumption (FULC), and emission (EMIS). The data for simulation; system dynamics;
the policies is generated via system dynamics simulation. ANP is used to rank the evaluation criteria analytic network process
and the alternatives (sustainable transportation policies). A numerical study is provided. The results
of our study reveal that trip sharing based policies perform better in comparison to the other policies
for achieving sustainability in a transportation system.
1. Introduction
consumption, congestion, air pollution etc., but the
Sustainable transportation systems play a vital role in negative side effect can be on the auto market eco-
developing a nation’s economy and human settlements. nomically. The auto industry may suffer through a
It involves policies, technologies and other initiatives that recession causing unemployment and joblessness of
ameliorate system effectiveness and minimise negative autoworkers. This is against economic growth which
impacts on the city environment and social life. Defin- is a part of sustainable development.
ing a sustainable policy for the transportation system is
complex and challenging due to the presence of: Because of these reasons, there is still not a unified
framework for defining and evaluating different sustain-
• Stakeholders with multiple, often conflicting interests. able transportation policies. Such multifaceted problems
• Socio-economic-environmental factors that change necessitate dynamic, flexible, careful, and continuous
their behaviour over time. For example, more trans- examinations.
portation causes less fossil fuel remaining for the In literature, there are few works that deal with the
future. subject of sustainable transport policy evaluation, and
• Growing, active and dynamic elements. The policies even less on assessing their combined performance. Most
need to be redesigned with system changes. of the studies use subjective data. In this paper, we are
• Multiple cultures. Defining policy on sustainability is addressing these challenges. A hybrid approach based
highly dependent on the culture and view of deci- on ANP and system dynamics simulation is proposed.
sion makers (DM). For instance, in Switzerland the ANP is chosen due to its ability to handle multiple,
transportation policies are more government oriented correlated and conflicting criteria while SD (Forrester,
but in Germany and the Netherlands, they display the 1969) models the dependencies between system com-
opposite pattern (Hirschi, Schenkel, & Widmer, 2002). ponents including feedback loops. Importance of the
• Elements with different characteristics. A policy may approach in predicting future behaviour of a system, rec-
positively influence one side of the system and be dis- ommending solutions and adapting policies has encour-
astrous for the other side. For e.g. decreasing rate of car aged lot of sustainability analysts and policy legislators
ownership may be effective in reducing rate of own- to select SD for analysing dynamic systems (Han &
ing and driving private cars and consequently energy Hayashi, 2008).
CONTACT Anjali Awasthi awasthi@ciise.concordia.ca

© 2018 Informa UK Limited, trading as Taylor & Francis Group
2 R. SAYYADI AND A. AWASTHI
The rest of the article is as follows. Section 2 presents • Optimisation: An optimisation model consists of an
the literature review. The methodology and numerical objective function (max, min), constraints and deci-
applications are presented in Sections 3 and 4. Section sion variables. In the context of sustainability, the goal
5 presents the managerial inferences. The last Section 6 is to optimise social, economic and environmental
concludes the paper and provides directions for future objectives. Their limitation is the inability to handle
works. qualitative and correlated data.
• Simulation: Simulation is used for testing new scenar-
ios or conducting what-if analysis under the absence
2. Literature review of real models. Their limitation is the use of assump-
tions in building the computer model.
A number of methods and techniques have been • Assessment indicator models: These models assess sys-
proposed in the literature for evaluating the impact tem performance against indicators. Identification of
of environment-friendly measures like mode sharing, the right number and type of indicators for measur-
multi-modal transport solutions, etc. Key techniques ing sustainability dimensions is critical to the effective
based on Wellar (2009) and Awasthi and Chauhan (2011) functioning of these models.
are listed as follows: • Data analysis: This category of models rely on numer-
ical data and use of statistical analysis techniques like
• Cost benefit analysis (CBA) and Cost effectiveness anal- surveys, hypothesis testing, structural equation mod-
ysis (CEA): CBA determines the benefits and costs of elling etc. to measure sustainability performance.
projects in dollar values. CEA is used where it may • Multicriteria decision-making (MCDM): MCDM tech-
be inappropriate to monetise the effects e.g. noise, niques involve group decision making to evaluate
emissions. Their limitation is that it is extremely dif- alternatives against a set of criteria. Their main advan-
ficult to estimate directly external and social costs tage is the ability to take into account qualitative,
(Kunreuther, Grossi, & Seeber, 2003). conflicting, multidimensional, incommensurable and
• Life cycle analysis/assessment: This approach models uncertain effects of decisions explicitly (Beinat, 2001).
GHG emissions inventory for a product’s complete The limitation is that the solutions generated are
life cycle, from raw materials to final disposal of the tradeoff among the multiple objectives and not opti-
product to assess the environmental impacts. How- mal ones due to the nature of the problem.
ever, its main limitation is that it does not take into • Fuzzy set theory: Fuzzy set theory is used to model
consideration social aspects. vagueness and uncertainty in systems when quanti-
• Environmental impact assessment (EIA): EIA evalu- tative data is limited or not available (Chen, Hsieh,
ates the likely environmental impacts of a proposed & Hung Do, 2015; Tyagi, 2016). Nazam, Jiuping,
project or development, taking into account inter- Zhimiao, Jamil, and Muhammad (2015) use it in com-
related socio-economic, cultural and human-health bination with other methods such as AHP for measur-
impacts, both beneficial and adverse. ing sustainability performance.
Table 1. Commonly used methods for sustainable mobility evaluation.
Technique Authors
Life cycle analysis Guine (2002)
Cost benefit analysis Tudela, Akiki, and Cisternas (2006), Eliasson (2009), Damart and Roy (2009), Browne and Ryan (2011)
Cost effectiveness analysis Tsamboulas and Mikroudis (2000), Kunreuther et al. (2003)
Environment impact analysis European Conference of Ministers of Transport (2004), Wood (2002)
Optimisation Tao and Hung (2003), Iniestra and Gutiérrez (2009), Sayyadi and Awasthi (2018)
Simulation Maoh and Kanaroglou (2009), Shafiei, Stefansson, and Asgeirsson (2013), Hivin (2014), Sayyadi and Awasthi (2017)
Assessment indicator models Black, Paez, and Suthanaya (2002), Gudmundsson (2003), Tao and Hung (2003), Jeon and Amekudzi (2005), Browne, O’Regan,
and Moles (2008), Litman (2009), Maoh and Kanaroglou (2009), Haghshenas and Vaziri(2012), Lima, da Silva Lima, and da
Silva (2014)
Data analysis Mohan and Tiwari (1999)
Multicriteria decision-making Brand, Mattarelli, Moon, and Wolfler-Calvo (2002), Yedla and Shrestha (2003), Sayers, Jessop, and Hills (2003), Shang, Tjader,
and Ding (2004), Tzeng, Lin, and Opricovic (2005), Brey et al. (2007), Awasthi, Breuil, Chauhan, Parent, and Reveillere
(2007), Awasthi, Chauhan, Hurteau, and Breuil (2008), Awasthi, Chauhan, and Breuil (2009), Awasthi, Chauhan, and Goyal
(2011), Awasthi, Chauhan, and Omrani (2011), Omrani, Awasthi, Ion, and Trigano (2009), Awasthi and Omrani (2009),
Campos Gouvêa, Rodrigues Ramos, de Miranda, and Correia (2009), Turcksin, Bernardini, and Macharis (2011), Awasthi and
Chauhan (2011), Awasthi and Chauhan (2012), Hickman, Saxena, Banister, and Ashiru (2012), Ivanović, Grujičić, Macura,
Jović, and Bojović (2013), Jones, Tefe, and Appiah-Opoku (2013), Pérez, Carrillo, and Montoya-Torres (2014), Vermote,
Macharis, Hollevoet, and Putman (2014), Bai, Fahimnia, and Sarkis (2015), Le Pira, Inturri, Ignaccolo, and Pluchino (2015),
Curiel-Esparza, Mazario-Diez, Canto-Perello, and Martin-Utrillas (2016)
Fuzzy set theory Teng and Tzeng (1996), Avineri, Praskker, and Ceder (2000), Awasthi, Chauhan, and Goyal (2011), Tyagi, Agrawal, Yang, and
Ying (2017)
INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE: OPERATIONS & LOGISTICS 3
Table 2. MCDM approaches for sustainability mobility evaluation.

Technique Problem Authors
AHP Carsharing stations selection Awasthi et al. (2007)
AHP Selecting car-sharing stations in medium size agglomeration Awasthi et al. (2008)
Survey study Success factors for carsharing Awasthi et al. (2009)
Fuzzy TOPSIS Location planning for urban distribution centres under uncertainty Awasthi, Chauhan, and Goyal (2011)
Fuzzy TOPSIS Evaluating sustainable transportation systems Awasthi, Chauhan, and Omrani (2011)
Affinity Diagram, AHP, fuzzy TOPSIS Sustainable city logistics planning Awasthi and Chauhan (2012)
AHP, Dempster Shafer Theory Evaluating sustainable transport solutions Awasthi and Chauhan (2011)
AHP, Belief theory DeSSIA: approche hybride et outil d’aide à l’évaluation pour le projet SUCCESS Omrani et al. (2009)
ANP Road transport project selection Ivanović et al. (2013)
Scenario analysis, MCA Examining transport futures with Hickman et al. (2012)
AHP Sustainable mobility evaluation in urban areas Campos Gouvêa et al. (2009)
DEA Evaluation of automobiles with alternative fuels Brey et al. (2007)
AHP, TOPSIS, VIKOR Analysis of alternative-fuel buses for public transportation Tzeng et al. (2005)
ANP Evaluation of transportation projects Shang et al. (2004)
AHP, Belief Theory Evaluating sustainable transportation solutions Awasthi and Omrani (2009)
AHP, PROMETHEE Policy scenario selection to stimulate a clean vehicle fleet Turcksin et al. (2011)
AHP Selection of environmentally sustainable transport system options in Delhi Yedla and Shrestha (2003)
NAIADE STEEDS: A strategic transport-energy environment decision support Brand et al. (2002)
PROMETHEE, System Dynamics Study the influence of an urban toll and flexible working hours on congestion Springael, Kunsch, and Brans (2000)
MCA, MAMCA Appraise Sustainability and Stakeholder Preferences for Mobility Projects Vermote et al. (2014)
AHP, Delphi technique, VIKOR Select optimal strategies in achieving sustainable mobility Curiel-Esparza et al. (2016)
Table 1 summarises the commonly used methods for simulated. The second step involves evaluating policy
sustainable mobility evaluation. It can be seen in Table 1 evaluation.
that multicriteria decision-making is the most popu-
lar technique followed by assessment indicator models,
3.1. Policies under study
simulation, optimisation, and cost benefit/effectiveness
analysis. Five policies are considered in this study namely trip
Table 2 lists MCDM techniques used in sustainable sharing (TS), trip rate reduction (TRR), reducing the
mobility evaluation. It can be seen that AHP and TOPSIS length of the road network (LRN), minimising car own-
are the most popularly used techniques. ership (CAO), and reducing the average kilometres trav-
The advantage of MCDM techniques is the ability to elled (AKT). Table 3 presents these five policies and their
handle multiple, conflicting and correlated criteria (fac- definitions.
tors) often present in sustainable transport decision mak-
ing. However, the drawback is that they rely on subjective
judgments for project evaluation. If quantitative informa- 3.2. System dynamics (SD) simulation model
tion can be made available, then a more reliable assess- Figure 1 presents the SD model used to generate rele-
ment can be made. This is the challenge we are addressing vant quantitative data for evaluating sustainable trans-
in this paper. port policies (Section 3.1). The model has been con-
ceptualised using the logical and mathematical relation
between system elements obtained from literature and
3. Methodology
discussion with 5 transportation experts from academia.
The proposed methodology involves two steps. In The model consists of five levels and forty auxiliary
the first step, sustainable transportation policies are variables. Level variables in rectangular boxes can be
Table 3. Alternatives and criteria for evaluation.
Cluster Element Definition
Goal Maintaining Sustainability in A MCDA problem
Transportation System (SU TA)
Criteria Congestion (CONG) Volume of private vehicular commuters (less is better)
Fuel Consumption (FULC) Fossil fuel resource depletion (less is better)
Emission (EMIS) Volume of pollutants in air (less is better)
Alternatives (Sustainable Trip sharing (TRS) A policy by which we merge private vehicular trips
transportation policies)
Trip rate reduction (TRR) A policy by which people make less trips
Length of road network (LRN) A policy by which we increase the road network
Car Ownership (CAO) A policy by which less people are able to have personal cars
Average kilometres travelled (AKT) A policy by which we centralise facility and residential locations and
people are not required to have long distance trips for daily affairs
or job
Figure 1. The proposed SD model.
accumulative (e.g. road network) or inborn (e.g. popu-

lation). Other elements are normal auxiliary variables,
which can be constant, rate, dependent or independent.
For more details on the SD model and involved equations,
please refer to Sayyadi and Awasthi (2017).
It is possible to separate the model into several sub-
systems. In travel supply and demand section (SD),
mostly private transit and public transit sub-systems,
Figure 2. Structural difference between AHP and ANP.
their volume and changes, congestion, and level of use-
fulness of the transit mode by people are considered.
The environmental (EN) sub-system mainly considers
components of the decision model and relies on the judg-
emission, fossil fuel resources depletion and waste pro-
ments of experts to derive priorities on absolute scales.
duction. Monetary related issues such as GDP and cost of
ANP models interdependencies using a feedback struc-
investments in the road network and public transit fleet,
ture among clusters or elements of a network (Saaty,
rate of owning car and its volume and domestic income
2001). The comparison is done between different clusters
are covered under Economic section (EC). The Social
of a network (outer dependency) or in the same cluster
section (SC) of the model deals with human-related
(inner dependency) to generate priority of elements. The
issues such as population, rate of death and birth.
general procedure of ANP involves the following steps:
• Step 1: Develop the decision model network by

3.3. Analytic network process
decomposing the ranking problem into a hierarchy
The SD model provides data for pairwise comparison containing main objective or goal cluster(s), criteria
matrices of ANP. The ANP against its ancestor AHP or attributes cluster(s), sub criteria and alternatives
(Figure 2) can address correlated model elements. cluster(s).
The ANP technique (Saaty, 1999) asks for perfor- • Step 2: Establish pairwise comparison matrices A =
mance measurement through pairwise comparisons of all [aij ]for all clusters and elements They should be square
matrices in which:

aij = aji = 1 ∀i = j
aji = 1/aij ∀i = j
The judiciousness-based scaled values of aij will be drawn

by the analyst(s) from the following table (Table 4) and
assigned to the evaluated performances. Step 3: Calcu-
late a list of the relative weights, importance, or value,
of the elements called wij to establish a Relative Value
Vector (RVV) or local priority weight vector w in the rel-
evant matrix. The principal eigenvalue λmax to examine
the consistency of the scaled decision matrix is obtained Figure 3. Sample of the super matrix.
through the following formulas:
(Am×m )(wm×1 ) = (λmax )(wm×1 )
Consistency Index: CI = (λmax − n)/(n − 1).

Consistency Ratio: CR = CI/RI.
where nis the number of elements in the decision matrix
and Random consistency Index (RI) is obtained from
Table 5 (Saaty, 1980)
If CR ≤ 0.1, then the comparison matrix is consid-
ered to be consistent enough for proceeding otherwise
the comparison matrix should be improved.
• Step 4: The primary super matrix (Figure 3) can

Figure 4. The hierarchy network of the decision model.
be established through synthesising the RVVs. For
componentCd , d = 1, 2, . . . , N with md elements of
matrix. It will show the global priority vectors.
ed1 , ed2 , . . . , edmd
• Step 5: Generally the primary super matrix is lim W k
unweighted. So, it should be transformed into the k→∞
weighted super matrix. It is necessary to provide
meaningful limiting priorities.
• Step 6: The limited super matrix is generated by raising 3.4. Establishing the network
the weighted super matrix to limit large power until it
converges to the same values for each column of the Figure 4 demonstrates the ANP hierarchical network to
represent dominance between clusters and within ele-
ments. The straight arrows present the links, considering
Table 4. Scales to convert comparative values into preference both top-down and bottom-up directions. The dashed
values. arrows illustrate feedbacks and interdependencies of ele-
Intensity of ments particularly with respect to the goal. Two com-
Importance (aij ) Explanation
ponents, each at a time, are compared with respect to
1 Equal importance another cluster or element for establishing dominance
3 Moderate importance of a component over another
5 Strong or essential importance matrices. Matrices are positive and reciprocal. Three cri-
7 Very strong or demonstrated importance teria namely Congestion (CONG), Fuel Consumption
9 Extreme importance
2,4,6,8 Intermediate values when needed (FULC), and Emission (EMIS) are used to evaluate the
1/aij Reciprocals for inverse comparisons policies.
Table 5. Values of RI. 4. Numerical illustration

n 3 4 5 6 7 8 9 10
Let’s consider a metropolitan with the following empiri-
RI 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.49
cal specifications (Table 6).
Table 6. Empirical input assumptions of the model. Table 8. Quantitative evaluation of alternatives.
Variable Input Value Unit Impacts on Criteria
Population 4,500,000 NO. (people) Alternative Policy CONG FULC (106 ) EMIS (106 )
Birth Rate 0.01 –
Death Rate 0.008 – TRS 18.12 4.15 9.79
GDP 25,000 × 4.5e6 $ TRR 27.19 6.22 14.61
GDP Rate 0.01 – LRN 30.58 8.29 19.64
Unit cost of bus 500,000 $ CAO 32.64 7.47 17.62
Trip Rate 2.8 – AKT 20.47 4.69 11.22
Car ownership rate 64% No. of car for 100 of
people
Vehicle travelled per lane 800 VKT
(Wang et al.,2008)
Table 9. Pairwise comparison matrix of criteria with respect to
Average Distance Travelled 8.5 km sustainability.
per Capita
SU TA CONG FULC EMIS RVV
Area of metropolitan 4500 km2
Network Length (in lane) 2700 km CONG 1 1/2 1/3 0.1634
Number of buses 1400 No. FULC 2 1 1/2 0.2969
Number of bus routes 190 No. EMIS 3 2 1 0.5396
Average travel time of bus 60 minute λmax = 3.0092→ CR = 0.0079 ≤0.1→ matrix consistency OK.
route
Ave. Vehicle emission 0.25 kg/km
factor (EPA standards)
Fuel Efficiency 0.106 litre/km Table 10. Pairwise comparison matrix of alternatives with
Working hours 16 hr respect to goal.
Goal TRS TRR LRN CAO AKT RVV
TRS 1 1 5 7 1 0.2974
TRR 1 1 5 7 1 0.2972
Table 7 shows the input data to model the five polices LRN 1/5 1/5 1 3 1/5 0.0712
in the SD model. TRR is a multiplier value by which CAO 1/7 1/7 1/3 1 1/7 0.0371
AKT 1 1 5 7 1 0.2972
number of population is converted to the number of λmax = 5.0711→ CR = 0.0159 ≤ 0.1 → matrix consistency OK.
trips made in a metropolitan. AKT as another multiplier
value considers an averaged kilometres length travelled
for each urban trip. The LRN represents a general con- respect to the Goal. The matrix entries are generated by
servative value or quantity of length of entire lanes in the group decision making committee.
the road network (it is a conservative value due to our Tables 11121314–15 present the pairwise comparison
assumption in which amalgamation of all roadways types matrices for the five policies (TRS, TRR, LRN, CAO and
used in a metropolitan such as freeways, highways, arteri- AKT) against the three criteria (CONG, FULC, EMIS)
als, collectors, etc is considered in one unit). Also, FE and and vice-versa. These matrices are generated based on SD
EF are the two technical factors affected by technology. model outputs (Table 8).
Each factor has two levels (Low and High).
Table 8 shows the outputs from the SD model namely
CONG, FULC and EMIS. Table 11. Pairwise comparison matrix of alternatives with
Table 9 presents the pairwise comparison matrix respect to congestion.
to determine the weights of the three criteria namely CONG TRS TRR LRN CAO AKT RVV
CONG, FULC and EMIS with respect to the sustainable TRS 1 5 7 9 3 0.5128
transportation objective. The matrix entries are gener- TRR 1/5 1 3 5 1/3 0.1289
LRN 1/7 1/3 1 3 1/5 0.0634
ated by consensus among the group decision makers. CAO 1/9 1/5 1/3 1 1/7 0.0334
Table 10 presents the pairwise comparison matrices of AKT 1/3 3 5 7 1 0.2615
λmax = 5.2375→ CR = 0.0530 ≤0.1→ matrix consistency OK.
the five alternatives (TRS, TRR, LRN, CAO, AKT) with
Table 7. Factors and their Values.

Impacts Levels Dependent Factors (Criteria)
Alias Factor Name Low (−) xLi < LL < xLj High (+)xHi < LH < xHj CONG FULC EMIS Unit
A TRR 2.1-2.6 3.2-3.8 N/A

B AKT 4.8-5.8 8.8-9.5 km
C LRN 2500–2750 2950–3200 km la
D FE 6.5-8.8 8.5-10 L
100gkm
E EF 150–170 180–210
km
Table 12. Pairwise comparison matrix of alternatives with 4.1. Establishing the super matrix
respect to fuel consumption.
The relative value vectors from the above matrices are
FULC TRS TRR LRN CAO AKT RVV
synthesised to establish the unweighted super matrix
TRS 1 5 7 6 3 0.4954
TRR 1/5 1 5 3 1/3 0.1347 (Table 16). The unweighted super matrix is normalised
LRN 1/7 1/5 1 1/3 1/6 0.0382 to obtain the weighted super matrix (Table 17). The rela-
CAO 1/6 1/3 3 1 1/5 0.0681 tive importance value is present in the column goal of the
AKT 1/3 3 6 5 1 0.2636
λmax = 5.3188→ CR = 0.0712≤0.1→ matrix consistency OK. final matrix.
Table 18 presents the limited super matrix. The results
Table 13. Pairwise comparison matrix of alternatives with (clusters) are presented in the GOAL column. Among
respect to emission. the alternatives, TRS scores the highest followed by AKT
EMIS TRS TRR LRN CAO AKT RVV and TRR. This highlights the importance of reducing
TRS 1 5 8 7 3 0.5079
vehicular air pollutants by trip sharing or decreasing trips
TRR 1/5 1 5 3 1/3 0.1301 distances of people. Among the three criteria, Emission
LRN 1/8 1/5 1 1/3 1/7 0.0348 level EMIS (0.1619) scores the highest weight followed
CAO 1/7 1/3 3 1 1/5 0.0641
AKT 1/3 3 7 5 1 0.2632 by FULC (0.1123), and CONG (0.059). The economic,
λmax = 5.2664→ CR = 0.0595≤0.1→ matrix consistency OK. environment and social dimensions contribute equally
towards the sustainability goal with a weight = 0.333.
LRN.
LRN CONG FULC EMIS RVV 4.2. Sensitivity analysis
CONG 1 3 2 0.5499
FULC 1/3 1 1 0.2099 Sensitivity analysis evaluates the effect of changes in the
EMIS 1/2 1 1 0.2402
λmax = 3.0183→ CR = 0.0316≤0.1→ matrix consistency OK.
input values on the sensitivity of the output parameters.
Figure 5 presents the results of sensitivity analysis. It can
be seen that the ranking results (for alternatives) are not
the other alternatives. very sensitive to the criteria weights.
TRS CONG FULC EMIS RVV
CONG 1 1/2 1/3 0.1692
FULC 2 1 1 0.3874 5. Managerial inferences
EMIS 3 1 1 0.4434
λmax = 3.0185→ CR = 0.0314 ≤0.1→ matrix consistency OK. The proposed research has several managerial inferences.
Using the proposed methodology, decision-makers can
Table 16. Primary (unweighted) super matrix.

TRS TRR LRN CAO AKT CONG FULC EMIS GOAL
TRS 1 0 0 0 0 0.5128 0.495 0.5079 0.2974
TRR 0 1 0 0 0 0.1289 0.134 0.1301 0.2972
LRN 0 0 1 0 0 0.0634 0.038 0.0348 0.0712
CAO 0 0 0 1 0 0.0334 0.068 0.064 0.0371
AKT 0 0 0 0 1 0.2615 0.263 0.2632 0.2972
CONG 0.1692 0.1692 0.5499 0.1692 0.1692 1 0 0 0.1634
FULC 0.3874 0.3874 0.2099 0.3874 0.3874 0 1 0 0.2969
EMIS 0.4434 0.4434 0.2402 0.4434 0.4434 0 0 1 0.5396
GOAL 1 1 1 1 1 1 1 1 1
Table 17. Weighted super matrix.

TRS 0.3333 0 0 0 0 0.1709 0.165 0.169 0.099
TRR 0 0.333 0 0 0 0.0430 0.0449 0.0434 0.099
LRN 0 0 0.333 0 0 0.0211 0.0127 0.0116 0.023
CAO 0 0 0 0.3333 0 0.0111 0.0227 0.0213 0.012
AKT 0 0 0 0 0.3333 0.0872 0.0879 0.0877 0.099
CONG 0.0564 0.056 0.1833 0.0564 0.0564 0.3333 0 0 0.054
FULC 0.1291 0.129 0.070 0.129 0.129 0.00 0.333 0.000 0.099
EMIS 0.1478 0.147 0.080 0.147 0.147 0.00 0.000 0.333 0.179
Goal 0.3333 0.333 0.333 0.333 0.333 0.3333 0.333 0.333 0.333
Table 18. Limited super matrix.

TRS 0.133 0.133 0.1337 0.1337 0.1337 0.1337 0.1337 0.1337 0.1337
TRR 0.0714 0.071 0.0714 0.0714 0.0714 0.0714 0.0714 0.0714 0.0714
LRN 0.0187 0.018 0.0187 0.0187 0.0187 0.0187 0.0187 0.0187 0.0187
CAO 0.0162 0.016 0.0162 0.0162 0.0162 0.0162 0.0162 0.0162 0.0162
AKT 0.0934 0.093 0.0934 0.0934 0.0934 0.0934 0.0934 0.0934 0.0934
CONG 0.0590 0.059 0.0590 0.0590 0.0590 0.0590 0.0590 0.0590 0.0590
FULC 0.1123 0.112 0.1123 0.1123 0.1123 0.1123 0.1123 0.1123 0.1123
EMIS 0.1619 0.161 0.1619 0.1619 0.1619 0.1619 0.1619 0.1619 0.1619
GOAL 0.3333 0.333 0.3333 0.3333 0.3333 0.3333 0.3333 0.3333 0.3333
emissions while the transport operators would like to

maximise trips to augment revenues. With the aid of the
proposed approach, decision makers can achieve goals
that respect everyone’s interest.
Thirdly, the proposed approach has strong practical
applicability. It can be easily coupled with city transport,
population, socio-economic, and demographic databases
for evaluating sustainable transportation policies.
Last but not least, the proposed work allows testing
several policies in tandem. Decision makers can assess
the gain or loss from implementation of policies in com-
bination. In addition, impact of time, space, context and
other external variables can also be evaluated.
6. Conclusions and future works

In this paper, we present an integrated approach based on
SD and ANP to evaluate the impact of sustainable trans-
portation policies. Five policies related to trip sharing,
trip rate reduction, length of road network, car owner-
Figure 5. Sensitivity analysis of alternatives under three criteria
with respect to goal. ship, and average kilometres travelled (AKT) are evalu-
ated against three criteria namely congestion level, fuel
consumption, and emission. System dynamics simulation
is used to generate data for sustainable transport policy
pro-actively plan and select the right sustainable trans- evaluation. ANP is used due to evaluate the policies using
port policies, thereby avoiding waste of time, money and expert opinions and simulated data. A numerical applica-
resources. The use of system dynamics simulation per- tion is provided. The results of the study yield trip sharing
mits generation of quantitative data for evaluation of policy (merging individual trips in private cars) at the
sustainable transportation policies. The decision-makers first place followed by policies which intend to reduce
can conduct what-if scenario analysis to determine which travelling distances and rate of urban trips. Future works
project(s) score better under which circumstances. Sen- involve:
sitivity analysis can be done to determine the influence
of variation in modelling parameters (criteria weights in • Extending the current study to integrate large size
our study) on the stability of model results. networks.
Secondly, the proposed methodology involves the use • Testing the proposed model with real data and com-
of group decision-making technique ANP thereby allow- paring the results with other comparable studies in the
ing multiple stakeholders opinion and correlated crite- literature
ria to be represented in the selection process. This will • Extending the study to integrated modelling of pas-
result in a win-win situation for all the involved decision- senger and freight transport policies.
makers (stakeholders). Often, the objectives of different
stakeholders are in conflict. For example, the municipal
administrators would like to impose regulations to limit Disclosure statement
vehicle movements and cut Green House Gas (GHG) No potential conflict of interest was reported by the authors.
Notes on contributors Black, J. A., Paez, A., & Suthanaya, P. A. (2002). Sustain-
able urban transportation: Performance indicators and some
Reza Sayyadi completed his master’s (MASc) in Quality Sys- analytical approaches. Journal of Urban Planning and Devel-
tems Engineering at Concordia Institute for Information Sys- opment, 128, 184–208.
tems Engineering (CIISE), in Concordia University, Montreal. Brand, C., Mattarelli, M., Moon, D., & Wolfler-Calvo, R. (2002).
His thesis addressed the subject of sustainable transportation STEEDS: A strategic transport-energy environment decision
policy planning. He is the author of one journal and one con- support. European Journal of Operational Research, 139(2),
ference paper in the area. 416–435.
Anjali Awasthi is an associate professor at Concordia Institute Brey, J. J., Contreras, I., Carazo, A. F., Brey, R., Hernández-Díaz,
for Information Systems Engineering (CIISE), in Concordia A. G., & Castro, A. (2007). Evaluation of automobiles with
University, Montreal. She received a Ph.D. in industrial engi- alternative fuels utilizing multicriteria techniques. Journal of
neering and automation from INRIA Rocquencourt & Uni- Power Sources, 169(1), 213–219.
versity of Metz, France. Awasthi has several years of industry Browne, D., O’Regan, B., & Moles, R. (2008). Use of ecologi-
and research experience in areas of automated transporta- cal footprinting to explore alternative policy scenarios in an
tion, sustainable mobility solutions, city logistics, and applied Irish city-region. Transportation Research D, 13(5), 315–322.
operations research. She is the author of several journal and Browne, D., & Ryan, L. (2011). Comparative analysis of evalua-
conference papers on these topics. tion techniques for transport policies. Environmental Impact
Assessment Review, 31, 226–233.
Campos Gouvêa, V. B., Rodrigues Ramos, R. A., de Miranda,
References D., & Correia, S. (2009). Multicriteria analysis procedure for
sustainable mobility evaluation in urban areas. Journal of
Avineri, E., Praskker, J., & Ceder, A. (2000). Transportation Advanced Transportation, 43(4), 371–390.
projects selection process using fuzzy sets theory. Fuzzy Sets Chen, J.-F., Hsieh, H.-N. H., & Hung Do, Q. (2015). Evaluating
and Systems, 116, 35–47. teaching performance based on fuzzy AHP and compre-
Awasthi, A., Breuil, D., Chauhan, S. S., Parent, M., & Reveillere, hensive evaluation approach. Applied Soft Computing, 28,
T. (2007). A multicriteria decision making approach for car- 100–108.
sharing stations selection. Journal of Decision Systems, 16(1), Curiel-Esparza, J., Mazario-Diez, J. L., Canto-Perello, J., &
57–78. Martin-Utrillas, M. (2016). Prioritization by consensus of
Awasthi, A., & Chauhan, S. S. (2011). Using AHP and Demp- enhancements for sustainable mobility in urban areas. Envi-
ster–Shafer theory for evaluating sustainable transport solu- ronmental Science & Policy, 55(1), 248–257.
tions. Environmental Modelling & Software, 26(6), 787–796. Damart, S., & Roy, B. (2009). The uses of cost- benefit anal-
Awasthi, A., & Chauhan, S. S. (2012). A hybrid approach inte- ysis in public transportation decision-making in France.
grating affinity diagram, AHP and fuzzy TOPSIS for sustain- Transport Policy, 16, 200–212. doi:10.1016/j.tranpol.2009.
able city logistics planning. Applied Mathematical Modelling, 06.002
36(2), 573–584. Eliasson, J. (2009). A cost-benefit analysis of the Stockholm
Awasthi, A., Chauhan, S. S., & Breuil, D. (2009). Sustain- congestion charging system. Transportation Research A,
able mobility solutions: A pre-implementation questionnaire 43(4), 468–480.
study for carsharing. International Journal of Services Sci- European Conference of Ministers of Transport. (2004). Assess-
ences, 2(3/4), 242–264. ment and decision making for sustainable transport. Paris:
Awasthi, A., Chauhan, S. S., & Goyal, S. K. (2011a). A multi- Organisation for Economic Co-Operation and Development
criteria decision making approach for location planning for (OECD).
urban distribution centers under uncertainty. Mathematical Forrester, J. W. (1969). Urban dynamics. Cambridge, MA: MIT
and Computer Modelling, 53(1-2), 98–109. Press.
Awasthi, A., Chauhan, S. S., Hurteau, X., & Breuil, D. (2008). Gudmundsson, H. (2003). Sustainable mobility and indicator
An Analytical hierarchical process-based decision-making systems in transport policy. Malden: UNESCO, Blackwell.
approach for selecting car-sharing stations in medium size Guine, J. B. (2002). Handbook on life cycle assessment. An oper-
agglomerations. International Journal of Information and ational guide to the ISO standard (p. 704). London: Kluwer
Decision Sciences, 1(1), 66–97. Academic.
Awasthi, A., Chauhan, S. S., & Omrani, H. (2011b). Appli- Haghshenas, H., & Vaziri, M. (2012). Urban sustainable trans-
cation of fuzzy TOPSIS in evaluating sustainable transportation indicators for global comparison. Ecological Indi-
portation systems. Expert Systems with Applications, 38(10), cators, 15, 115–121.
12270–12280. Han, J., & Hayashi, Y. (2008). A system dynamics model of
Awasthi, A., & Omrani, H. (2009). A hybrid approach based on CO2 mitigation in China’s inter-city passenger transport.
AHP and belief theory for evaluating sustainable transporta- Transportation Research Part D: Transport and Environment,
tion solutions. International Journal of Global Environmental 13(5), 298–305.
Issues, 9(3), 212–226. Hickman, R., Saxena, S., Banister, D., & Ashiru, O. (2012).
Bai, C., Fahimnia, B., & Sarkis, J. (2015). Sustainable transport Examining transport futures with scenario analysis and
fleet appraisal using a hybrid multi-objective decision mak- MCA. Transportation Research A, 46(3), 560–575.
ing approach. Annals of Operations Research, 250, 309–340. Hirschi, C., Schenkel, W., & Widmer, T. (2002). Designing sus-
Beinat, E. (2001). Multi-criteria analysis for environmental tainable transportation policy for acceptance: A comparison
management. Journal of Multi-Criteria Decision Analysis, of Germany, The Netherlands and Switzerland. Journal of
2001, 10–51. German Policy Studies, 2(4), 1–40.
Hivin, L. F. (2014). Sustainability of multimodal intercity trans- Sayyadi, R., & Awasthi, A. (2017). A system dynamics based
portation using a hybrid system dynamics and agent-based simulation model to evaluate regulatory policies for sus-
modeling approach (doctoral dissertation). Retrieved from tainable transportation planning. International Journal of
https://smartech.gatech.edu/handle/1853/53089 Modelling and Simulation, 37(1), 25–35.
Iniestra, J. G., & Gutiérrez, J. G. (2009). Multicriteria decisions Sayyadi, R., & Awasthi, A. (2018). A simulation-based optimi-
on interdependent infrastructure transportation projects sation approach for identifying key determinants for sus-
using an evolutionary-based framework. Applied Soft Com- tainable transportation planning. International Journal of
puting, 9(2), 512–526. Systems Science: Operations & Logistics, 5(2), 161–174.
Ivanović, I., Grujičić, D., Macura, D., Jović, J., & Bojović, N. Shafiei, E., Stefansson, H., & Asgeirsson, E. I. (2013). Integrated
(2013). One approach for road transport project selection. agent-based and system dynamics modelling for simulation
Transport Policy, 25, 22–29. of sustainable mobility. Transport Reviews, 33(1), 44–70.
Jeon, C. M., & Amekudzi, A. (2005). Addressing sustainabil- Shang, J. S., Tjader, Y., & Ding, Y. (2004). A unified frame-
ity in transportation systems: Definitions, indicators, and work for multicriteria evaluation of transportation projects.
metrics. Journal of Infrastructure Systems, 11, 31–50. IEEE Transactions on Engineering Management, 51(3),
Jones, S., Tefe, M., & Appiah-Opoku, S. (2013). Proposed 300–313.
framework for sustainability screening of urban transport Springael, J., Kunsch, P., & Brans, J. P. (2000). A group multi-
projects in developing countries: A case study of Accra. criteria decision aid and system dynamics approach to study
Ghana. Transportation Research A, 49, 21–34. the influence of an urban toll and flexible working hours
Kunreuther, H., Grossi, P., & Seeber, N. (2003). A framework for on the congestion problem. CSOO-Vrije Universiteit Brussel,
evaluating the cost-effectiveness of mitigation measures. New Belgium.
York, NY: Columbia University. Tao, C. C., & Hung, C. C. (2003). A comparative approach of
Le Pira, M., Inturri, G., Ignaccolo, M., & Pluchino, A. (2015). the quantitative models for sustainable transportation. Jour-
Analysis of AHP methods and the pairwise majority rule nal of the Eastern Asia Society for Transportation Studies, 5,
(PMR) for collective preference rankings of sustainable 3329–3344.
mobility solutions. Transportation Research Procedia, 10, Teng, J. Y., & Tzeng, G. H. (1996). Fuzzy multicriteria ranking
777–787. of urban transportation investment alternatives. Transporta-
Lima, J. P., da Silva Lima, R., & da Silva, A. N. R. (2014). Eval- tion Planning and Technology, 20(1), 15–31.
uation and selection of alternatives for the promotion of Tsamboulas, D., & Mikroudis, G. (2000). EFECT – evaluation
sustainable urban mobility. Procedia - Social and Behavioral framework of environmental impacts and costs of transport
Sciences, 162, 408–418. initiatives. Transportation Research D, 5(4), 283–303.
Litman, T. (2009). A good example of bad transportation per- Tudela, A., Akiki, N., & Cisternas, R. (2006). Comparing the
formance evaluation (Working paper). Victoria Transport output of cost benefit and multicriteria analysis: An applica-
Policy Institute. tion to urban transport investments. Transportation Research
Maoh, H., & Kanaroglou, P. (2009). A tool for evaluating urban A, 40(5), 414–423.
sustainability via integrated transportation and land use sim- Turcksin, L., Bernardini, A., & Macharis, C. (2011). A com-
ulation models. Urban Environment, 3, 28–46. bined AHP-PROMETHEE approach for selecting the most
Mohan, D., & Tiwari, G. (1999). Sustainable transport systems: appropriate policy scenario to stimulate a clean vehicle fleet.
Linkages between environmental issues, public transport, Procedia - Social and Behavioral Sciences, 20, 954–965.
non-motorised transport and safety. Economic and Political Tyagi, S. (2016). An improved fuzzy-AHP (IFAHP) approach
Weekly, 34(25), 1589–1596. to compare SECI modes. International Journal of Production
Nazam, M., Jiuping, X., Zhimiao, T., Jamil, A., & Muhammad, Research, 54(15), 4520–4536.
H. (2015). A fuzzy AHP-TOPSIS framework for the risk Tyagi, S., Agrawal, S., Yang, K., & Ying, H. (2017). An extended
assessment of green supply chain implementation in the tex- fuzzy-AHP approach to rank the influences of socialization-
tile industry. International Journal of Supply and Operations externalization-combination-internalization modes on the
Management, 2(1), 548–568. development phase. Applied Soft Computing, 52,
Omrani, H., Awasthi, A., Ion, L., & Trigano, P. (2009). DeSSIA: 505–518.
Approche hybride et outil d’aide à l’évaluation pour le projet Tzeng, G., Lin, C., & Opricovic, S. (2005). Multi-criteria analy-
SUCCESS. International Journal of Information - Interaction sis of alternative-fuel buses for public transportation. Energy
- Intelligence (I3), 8(1), 1–32. Policy, 33(11), 1373–1383.
Pérez, J. C., Carrillo, M. H., & Montoya-Torres, J. R. (2014). Vermote, L., Macharis, C., Hollevoet, J., & Putman, K. (2014).
Multi-criteria approaches for urban passenger transport sys- Participatory evaluation of regional light rail scenarios: A
tems: A literature review. Annals of Operations Research, Flemish case on sustainable mobility and land-use. Environ-
226(1), 69–87. mental Science & Policy, 37, 101–120.
Saaty, T. L. (1980). The analytic hierarchy process. New York, NY: Wellar, B. (2009). Results of an inquiry into the methodologies,
McGraw Hill. methods, and techniques used to make decisions about sus-
Saaty, T. L. (1999). Fundamentals of the analytic network process, tainable transport practices. Final Report. Retrieved from
ISAHP-1999. Kobe, Japan. http://www.wellarconsulting.com/
Saaty, T. L. (2001). Decision making with dependence and feed- Wood, C. (2002). Environmental impact assessment: A compar-
back, the analytic network process. Pittsburgh, PA: RWS Pub- ative review (p. 405). (2nd ed.). London: Prentice-Hall.
lications. Yedla, S., & Shrestha, R. (2003). Multi-criteria approach for the
Sayers, T. M., Jessop, A. T., & Hills, P. J. (2003). Multi-criteria selection of alternative options for environmentally sustain-
evaluation of transport options – flexible, transparent and able transport system in Delhi. Transportation Research A,
user-friendly? Transport Policy, 10, 95–105. 37(8), 717–729.
View publication stats

Integrating Shared Autonomous Vehicle in Public Transportation System:
A Supply-Side Simulation of the First-Mile Service in Singapore
Yu Shen, Hongmou Zhang and Jinhua Zhao
Yu Shen, PhD
Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji
University, Shanghai, China
Future Urban Mobility IRG, Singapore-MIT Alliance for Research and Technology Centre,
Singapore
Hongmou Zhang
Massachusetts Institute of Technology, Cambridge, MA
Jinhua Zhao, PhD (Corresponding Author)

Edward H. and Joyce Linde Associate Professor
Massachusetts Institute of Technology
77 Massachusetts Ave, Cambridge, MA 02139
Email: jinhua@mit.edu
Abstract
This paper proposes and simulates an integrated autonomous vehicle (AV) and public
transportation (PT) system. After discussing the attributes of and the interaction among the
prospective stakeholders in the system, we identify opportunities for synergy between AVs and
the PT system based on Singapore’s organizational structure and demand characteristics.
Envisioning an integrated system in the context of the first-mile problem during morning peak
hours, we propose to preserve high demand bus routes while repurposing low-demand bus
routes and using shared AVs as an alternative. An agent-based supply-side simulation is built
to assess the performance of the proposed service in fifty-two scenarios with different fleet
sizes and ridesharing preferences. Under a set of assumptions on AV operation costs and
dispatching algorithms, the results show that the integrated system has the potential of
enhancing service quality, occupying fewer road resources, being financially sustainable, and
utilizing bus services more efficiently.
Keywords:
Autonomous vehicle; Public transportation; Agent-based model; First-mile problem; Mobility-
on-demand
1
1. Introduction
Autonomous vehicles (AVs) are poised to represent a revolutionary future for urban mobility
(Silberg and Wallace, 2012). Recent literature on the potential operation of AVs primarily
regards them as an upgrade to conventional personal vehicles with the essential characteristics
of demand responsiveness, fleet repositioning, and shareability (Alonso-Mora et al., 2017;
Correia and van Arem, 2016; Fagnant et al., 2015; Fagnant and Kockelman, 2014). However,
if we merely deploy AVs as upgraded versions of human-driven vehicles, we may not derive
the optimal benefits of the new technology, especially in large metropolitan areas with high
population densities and limited road resources. In cities where public transportation (PT) plays
a critical role, the relationship between AVs and the PT system should not be ignored. Most
studies, however, do not take into account such relationships. Only a few offer limited insight
into the AV-PT interaction and the PT is mostly pitted as a competitor (Chen and Kockelman,
2016 and Mendes et al., 2017). Discussions regarding AV and PT as complementary and
integrated are scarce. Liang et al. (2016) compared the service offered by automated taxi
systems with those provided by human-driven taxis over the last mile to train service; however,
improvements in the PT performance was not the focus of the study. Vakayil et al. (2017)
explored a hybrid transit system with on-demand AVs as an additional service to improve metro
connectivity; nevertheless, the relationship between AV and bus networks was neglected. Lenz
and Fraedrich (2016) discussed, however conceptually, the possibilities of hybridizing AVs
with PT, including improvements in inter-modality and individualization of the transit service.
To fill the research gap, we examine the attributes of and the interaction among the prospective
players in an integrated system (AV operators, PT operators, riders, public authorities, and
automakers) and explore the opportunities that AVs can provide when integrated into PT
systems. We envision a scenario in which AVs provide a complementary on-demand service to
conventional fixed-schedule fixed-route buses for the first/last mile and assess whether the new
integrated service improves the performance of the overall system. This study is based on
empirical travel demand and transit operation details derived from the smart card data in
Singapore.
The remainder of this paper is organized as follows. Section 2 discusses the integrated AV and
PT framework and highlights a list of characteristics of AVs that distinguishes the new system
from the traditional hybrid public transit system. Section 3 presents the case of Singapore. After
analyzing the current PT travel demand focusing on the first-/last-mile trips, we propose a
design where we preserve the high demand bus routes, repurpose low-demand bus routes
introduce AVs service. Next, we describe the agent-based model and simulate 52 integrated AV
and PT scenarios for first-mile trips during morning peak hours. We also simulate the status-
quo bus operations as the benchmark, again which we evaluate whether the integrated AV-PT
system is well-suited to improve the quality of service, occupy fewer road resources, and be
financially sustainable, and utilize buses more efficiently. Section 4 concludes this paper.
2. Toward a framework for integrated AV and PT system

There have been decades of efforts to design and to operate an integrated PT system with on-
demand flexibly-routed service since Daganzo (1978) and Wilson and Hendrickson (1980).
Taking advantages of both operating styles—fixed-route fixed-schedule service for corridors
2
of high demand and density, and demand-responsive service for areas with low ridership, low
density and scattered demand (Adebisi and Hurdle, 1982; Chang and Schonfeld, 1991a),
several integrated PT systems have been proposed (Aldaihani et al., 2004; Chang and
Schonfeld, 1991b; Li and Quadrifoglio, 2009). Various conceptual models evaluating the fixed-
route and demand-responsive transit services have also been presented (Diana et al., 2009; Li
and Quadrifoglio, 2010; Qiu et al., 2015). Constrained by the 20th-century technology, the
integrated demand-responsive transit system faced critical challenges including high costs to
operate the service, difficulties to communicate with the riders and manage shared rides, and
problems to control drivers.
In recent years, the rapid development of Information and Communication Technology (ICT)
has led to the emergence of transportation network companies (TNCs), e.g., Uber and Lyft, and
revived interest in flexible on-demand systems. Online communication platforms manage
shared rides more efficiently by matching the real-time demand with dynamic fleet operation
strategies with lower price for the rides. Some operators have incorporated non-dedicated
vehicles into their service models to reduce capital costs, allowing the fleet size to vary
dynamically according to changes in demand (see the case of FlexDenmark).
When AVs become available, they may offer an opportunity to address many organizational
and technological challenges in the current system such as reducing labor cost, improving
compliance with planning and operation control, expanding service hours, avoiding erroneous
human-driving behavior, and optimizing the spatial and temporal allocation of the PT services.
Figure 1 summarizes the AV characteristics from the perspectives of operation (AV operator
and PT operator), governance (public authority), technology (AV producer), and consumption
(AV riders and PT riders), distinguishing the integrated AV-PT system from the traditional
hybrid transit system. In this study, we focus on the relationships between AV operators and PT
authority, and between AV operators and conventional transit operators. The rest, despite its
importance, is beyond the scope of this paper.
The AV operators offer a new travel option from the passengers’ point of view. Indicators of
conventional PT service quality also apply to the shared-AV service on demand (See the
comprehensive set of service indicators in Eboli and Mazzulla, 2011). AV service can be more
attractive if the system is able to provide a flexible door-to-door service, cheaper thanks to
lowering operating cost, and personalized to fit individual preferences and circumstances. The
service can also benefit the elderly or passengers with disabilities. There can also be drawbacks
associated with the AV service. The impact on the driver employment will have profound
implication on the labor relation, contracting, and social justice in general, which is part of the
broad discussion on the social impact of the automation. Without the appearance of human
drivers, the driverlessness may raise concerns from riders regarding safety (the maturing of
driverless technology), security (preventing crimes in the vehicle) and the quality of customer
service in general (e.g., human assistance). These concerns are partially demonstrated in Dong
et al. (2017).
3
Figure 1. Stakeholders around AV-PT operation and AV characteristics from the perspectives
of operation, governance, technology, and consumption
AV brings the attention of public authority to understanding the role and impacts of AVs on the
mobility system. A variety of organizational structures can be imagined with respect to the
ownership of the transit and AV operators, the interaction between the operators, and the degree
of regulation and intervention from the public authority. Possible organizational models of the
integrated AV and PT systems can be envisioned based on the existing transit governance
literature (Bruun, 2013; Costa, 1996; van de Velde, 1999; Wilson, 1991) and the emerging
experience in regulating the TNC firms in London, Singapore, and Boston. They include but
are not limited to 1) the laisse-faire structure; 2) the experience on regulating TNC firms (e.g.
in London); 3) the “on-demand paratransit pilot program” led by the Massachusetts Bay
Transportation Authority (MBTA); 4) the deregulation of PT in the UK (Wilson, 1991); 5) the
“Scandinavian” model (Costa, 1996; van de Velde, 1999); 6) the fully coordinated model.
Appendix A summarizes the key characteristics of the six models but does not get into the
details of the specific responsibilities and functions of the public authorities and other
stakeholders. For instance, within the Scandinavian model, the organizational structures in
Copenhagen (Denmark), Malmö (Sweden), and Adelaide (Australia) can be distinguished
further based on the level of central planning and the design of incentives to tendering parties
(van de Velde, 1999).
4
2.1. Assumptions of the AV-PT systems based on the organizational models in Singapore
Public transit in Singapore is highly regulated by the Land Transport Authority (LTA), which
is responsible for service integration including fares, information, and route design 1 . The
subway system, known as mass rapid transit (MRT), is planned and constructed by the LTA,
who then leases the operating licenses to transit operators. The LTA tenders out bus services
with a contract for five years plus a two-year extension based on service performance. The bus
operators compete for the market to win the contract to enter the market, while the bus routes
are still designed by the LTA.
Based on the characteristics and regulatory structure of the transport system in Singapore, we
identify the opportunity for synergy between AV and PT system and assume the following
characteristics of the integrated AV-PT system.
• Planning and regulation: the public authority LTA remains responsible for transit network
planning and the assignment of service areas and routes.
o On-demand AV operation is supported and regulated by the public authority.
o On-demand AV operation is provided in dedicated service areas and is limited to first-
and last-mile service in the current phase.
• Transit fare and subsidy: The transit fare is heavily regulated. The fare adjustment formula
is regularly reviewed to ensure that the fares are in line with the cost structure and
productivity achieved in the industry and that they are affordable to the public.
o The fare structure of the AV service is considered part of the transit fare review and
adjustment.
o The cost for an AV ride is calculated in Section 3 to test the financial viability of the
AV service.
• Coordination and competition: the current transit system in Singapore is a coordinated
system with the unified fare and information structure.
o AV operation is coordinated with the MRT system for feeder service. In the designated
area, the AV operation and the adjusted bus operation are also coordinated (see section
3 for the specific design).
• Fare, Ticketing and Information Integration: the AV service is integrated into the transit
system regarding:
o Ticketing: the transit smart card (or any other payment systems used by PT) can be
used for the AV service.
o Fare: the cost of a shared-AV ride is set to be close to the current transit fare.
o Information: the operating characteristics of both AV and conventional PT modes are
known to each other and available to the passengers.
3. An integrated AV-PT system for the first-mile service in Singapore

First, we examine the first-mile performance in the status quo PT system using the travel
records from the contactless e-purse application (CEPAS) card—the transit smart card in
Singapore. Second, based on the organizational structure assumed in Section 2, we propose one
feasible operational scenario of integrating AVs into the PT system, aiming at improving the
first-mile service during the a.m. peak. Finally, we assess the performance of the newly
integrated system via agent-based simulation and compare it with the current PT-only system.
3.1. Study area and data

We define the first-/last-mile trips in Singapore as the connecting trips to/from the MRT stations,
including walk and bus as the dominant modes. The CEPAS data used in this study covers all
1
TransitLink, the subsidiary of LTA, is in charge of the specific development of service integration and design.
5
PT trips in August 2013, with over 175 million travel records. Since the CEPAS requires the
tap-in and tap-out policy in both bus and MRT systems, each trip record archives the date and
time of every entry and exit activity and the boarding and alighting stops/stations. The fare
policy recognizes the temporally adjacent journey stages as one single journey, which may
contain a series of MRT and/or bus rides.2 The details on the bus trips made before and after
the MRT ride are clearly identified.
Appendix B discusses the first-/last-mile travel demand during the entire workdays, and we
focus here on the two-hour morning peak (7:00 to 9:00 a.m.). Figure 2 shows the volume of
passengers entering each MRT station during the morning peak and the mode shares by bus vs.
other modes. We highlight four stations with over 10,000 passengers and more than 50% of
bus mode share, among which, the Tampines area is chosen as the case study. Tampines has
one of the highest population densities in Singapore, with about 240,000 residents living in a
12 km2 area (HDB, 2015). The MRT travel demand in Tampines station is also one of the
highest: between 7 and 9 a.m. on a typical workday, more than 15,000 passengers use the MRT
service, including over 8,000 taking a bus to access the train station. Figure 3 ranks the 27 bus
routes (and directions) serving the MRT station by the average workday peak-hour bus
ridership. There is an obvious imbalance of patronage: the top five routes account for more
than 55% of the first-mile travel demand, and the bottom 11 routes account for less than 10%
of the demand. The footprints of the high-demand and low-demand bus routes are illustrated
in Figure B in Appendix C.
Figure 2. First-mile travel demand between 7 and 9 a.m.

(Calculated by the authors based on CEPAS smart card data)
2
According to the LTA, in a single journey, the transfer time should be less than 45 minutes while the total
travel time should be less than two hours.
6
Figure 3. Bus ridership to Tampines MRT by route
(Calculated by the authors based on CEPAS smart card data; “_1” and “_2” indicate the two
directions of the same bus route)
3.2. Design and assessment of the integrated AV-PT system

People in Singapore are highly dependent on the MRT system. The first-/last-mile connectivity
is one of the major issues in the current system of Singapore. Numerous solutions have been
designed and proposed (Lesh, 2013) to improve the connectivity: 1) by proactive planning,
such as enhancing walkability (Cervero, 2007), encouraging park-and-ride (Hamer, 2010), and
increasing intermodal coordination (Chien and Schonfeld, 1998; Sørensen and Longva, 2011);
and 2) by innovative travel modes, such as bike sharing (Cervero et al., 2013; DeMaio, 2009;
Shaheen et al., 2010), and electric scooters (Shaheen and Finson, 2003).
We propose an integrated AV-PT system by preserving the high demand bus routes and
repurposing low-demand bus routes while using the shared AV service on demand as an
alternative:
• The 16 busiest bus routes (shown in Figure 3) are kept to ensure that 90% of the travel
demand is served efficiently.
• The 11 low-demand routes are repurposed in four different ways depending on the specific
configuration of the routes: 1) rerouting to reduce the detour, 2) rerouting to bypass the
high traffic center, 3) maintaining the route but with larger stop spacing, and 4) shifting
the destination to the nearby Simei MRT station. The details of these repurposing options
are discussed in Appendix C.
• Only the remaining 10% of the demand previously serviced by the 11 low-demand buses
is to be served by the on-demand AV service.
7
We further assume the following additional AV operation features in the simulation:
On vehicle ownership and operation:
• Vehicle ownership: individually owned AVs are not taken into account in the model; and
the fleet-based AVs are rented from non-dedicated vehicle providers (e.g., car-rental
companies) whenever needed.
• Labor cost: there is no cost for human drivers. Service and maintenance cost of AVs are
accounted in the rental fees paid to the car rental companies.
• Operation Compliance: the AVs comply fully with the central controller and do not reject
service qualifying requests from customers.
• Fleet rebalancing: when the AV is idle, it is sent to a parking area3.
On communication and sharing:
• Mobility-on-demand: the AV service is requested on demand.
• Real-time shareability: The requests for ridesharing are processed and responded to in real-
time. Both the sharing and no-sharing scenarios are simulated and compared.
We built an agent-based model4, simulating the behaviors and interaction among three types of
agents in the system: the passenger, the AV, and the bus, shown in Figure 4.
Figure 4. Agent-based simulation of passengers, buses and AVs

(rounded rectangles refer to the agents, ellipses refer to locations, rectangles with solid line
refer to the movements of agents, rectangles with dashed line refer to the interaction between
agents, and diamonds refer to the decisions of agents.)
3
We do not introduce more sophisticated anticipatory rebalancing algorithms in this paper, but they can be
incorporated in the future research. See Wen et al. (2017). In the simulation vehicles simply return to the
parking lot of the Tampines MRT station, owned by public authority for both bus and shared AV parking. The
parking cost is thus not included in the model.
4
Both the analytical approaches and simulation approaches can be used to assess an integrated AV-PT system.
Analytical approaches are important, but the closed form solution may not be available in the context of the
dynamic and complex real world. They are often based on a hypothetical and small transit network (e.g.,
Aldaihani et al., 2004; Li and Quadrifoglio, 2009).
8
Passenger behaviors
This model focuses on the passengers who used to ride the bus to the MRT station. The
incidence of passenger trips based on the bus boarding in every 15 minutes during the peak
hours. A Voronoi diagram based on the location of bus stops is generated to assign travel
demand to each building5.
As AVs are not yet widely available to the public, current worldwide studies on AV preference
are hypothetical (e.g., Bansal et al., 2016; Krueger et al., 2016; Kyriakidis et al., 2015; Lu et
al., 2017; Payre et al., 2014; Shin et al., 2015; Yap et al., 2016) and the settings in the cases
from other countries may not transfer easily to the context of Singapore. Since the consumer
preference for AVs in Singapore is not the focus of this study, this paper does not model the
mode choice behavior. Instead the model assumes the fixed modal split between bus (90%) and
AV (10%) and focuses on the supply-side simulation. But it is a critical future research area to
have a dedicated AV preference study based on the local context in Singapore.
When the AV is chosen as the travel mode, the passengers start to call for the ride repeatedly
until the system has successfully booked a car for him/her. Once the vehicle is assigned, he/she
then moves to the pick-up location to wait for the AV. As the density of bus stops in Singapore
is very high, the pick-up location is set as the closest bus stop from the passenger. To prevent a
passenger from waiting forever when no AV is available, a maximum waiting time is set to 10
minutes in the study, beyond which he/she will forgo the AV request and travel by bus.
When the bus is chosen, the passenger will walk directly to the bus stop to wait for the next
bus. As the frequency of bus service is high (less than 10 min headway), the model assumes
that passenger arrival time is independent to the bus schedule (Abkowitz and Tozzi, 1987; Fan
and Machemehl, 2002; Frumin and Zhao, 2012). Thanks to the high service reliability of
Singapore’s bus system, all passengers get onto a bus within 10 minutes waiting time.
On-demand shared AV service

The concept of the mobility-on-demand system was proposed and demonstrated in the 1970s,
and was called Dial-a-Ride Transit (DART) (Wilson et al. 1976). Two decades later, Dial (1995)
proposed a fully automated DART system. More advanced transit systems integrating various
vehicle sharing services—including buses and taxis—were recently designed and tested (e.g.,
Atasoy et al., 2015; Djavadian and Chow, 2016; Jung and Jayakrishnan, 2016). In this study,
the on-demand AV service with ridesharing resembles a taxi-sharing system (Galland et al.,
2014; Martínez et al., 2015) but the service area is restricted to the Tampines town with only
the first/last mile service, consistent with the proposed usage of the AV prototype in Singapore
(Chong et al., 2011). The driving behaviors for all AVs are identical, and no booking requests
are rejected. The idle AVs are sent to the parking lot of the Tampines MRT station if there is no
demand for AV. Each AV allows a maximum of four passengers, the capacity of the AV
prototype, to share the ride. As discussed in Section 2.1, the AV operation is regulated by the
transit authority to be complementary to the bus. It does not compete against bus service for a
bigger market share.
The AV routing follows the shortest path. We assumed that a passenger prefers to ride alone if
possible even when he/she agrees to share the ride. Thus, when a passenger calls for a ride, the
system first scans all empty AVs. If there are empty AVs, the system assigns the closest
5
Due to the lack of demographic information at the building level, the population are assumed to be evenly
distributed to the buildings in the cell.
9
available AV to pick up the passenger. Once the AV and the passenger are matched, a notice is
sent to the passenger requesting a meet-up at the pick-up point. If there are no empty AVs, the
system searches for all occupied AVs, the passengers of which have agreed to share the ride,
based on pre-assigned passenger sharing preference. The criteria to match the shared rides are
as follows. For each shareable AV, after picking up the passenger at location i, the decision of
whether to pick up the next passenger at location j is determined by the following conditions:
• The vehicle must have enough available seats.
• Global detour constraint: For a passenger at location i in a shared trip, regardless of the
number of passengers in the AV, and regardless of the boarding order of passenger i, the
total travel time from pick-up location of passenger i to MRT station s, 𝑇"# , must be less
than the direct service time, 𝑡"# multiplied by an overall detour ratio, α:
Tis < a × tis , where a > 1 . (1)
• Incremental detour constraint: In each event of picking up an additional passenger at
location j, the deviated travel time for the previous passenger i due to picking up j (i.e.,
𝑡"% + 𝑡%# ) must be less than the direct service time, 𝑡"# , multiplied by the step-by-step
detour threshold, β.
tij + t js < b × tis , where b > 1. (2)
The pricing structure is purely distance-based, without a “base fare”. It also reflects the detour
ratio of each passenger. For ridesharing trips, a discount for passenger i is calculated based on
the actual detour ratio r. The discounted price, 𝑃" is computed as in Eq. (3).
𝑃" = 𝑝 ∙ 𝑑" ∙ (1 − 𝑟 0 ) (3)
where:
• p is the base AV fare per km;
• 𝑑" is the direct travel distance of passenger i in km;
• 𝐷" is the detoured actual distance;
3
• 𝑟 = 5 4 − 1, the detour ratio;
4
• 𝛾 is the degree of discount offered due to the detour.
In the simulation, we set 𝑝" to S$1 per kilometer which is at the same order of magnitude as
the base bus fare during peak hours6, α to 2, β to 1.2, and 𝛾 to 2. Please note that past literature
shows that given the bus pricing structure in Singapore, there is no price variation in the first
mile access by bus (Mo et al., 2018). We cannot find locally relevant willingness-to-pay
estimates from prior literature. Thus, we design the pricing discount and the distance detour to
roughly balance each other so that the fixed demand assumption remains reasonable.
Bus routing and dispatching

While the shared-AV service is on-demand, the bus service still operates on a fixed route and
fixed schedule basis. Two agents are introduced for bus routing and dispatching. The bus route
controller is a high-level agent that manages all bus routes, including the stop sequences to
follow and the headways of each bus route during different time periods. We assigned a bus
dispatcher for each bus route. Based on the stop lists and headways, the bus dispatchers
generate buses for the corresponding routes according to the actual bus schedule. The data for
bus routes and schedules are provided by the LTA. The routes used are the ones presented in
Figures 3, which intersect the Tampines area and stop at the MRT station.
6
The base bus fare is about 0.87 Singapore dollar.
10
Once a bus is generated by the dispatcher, the route is determined based on a sequential stop
list. The bus vehicle follows the route at a constant speed. Upon arrival at a bus stop, each bus
agent dwells for a certain amount of time to pick up all the passengers who are waiting at the
bus stop at the bus arrival time. The actual bus dwell time depends on the number of passengers
entering and exiting the bus, as well as the total number of passengers on the bus. However,
we simplify the dwell time to be fixed (30 seconds) in the simulation due to the lack of detailed
operational information and corresponding coefficients. The waiting time of each passenger is
archived, and the number of passengers onboard is logged. We also calculate the travel time
and travel kilometers for each passenger.
The model was coded in AnyLogic 7.3. We executed the simulation between 6:00 and 9:00 a.m.
(10,800 s) where one second in the simulation corresponds to one second in reality. The first
hour (6 to 7 a.m.) was regarded as the warm-up period; the outputs from the last two hours (7
to 9 a.m.) were recorded and analyzed.
3.3. Simulation results and analysis

Benchmark scenario: Bus-only service
We first simulated the status quo scenario with only the bus services, following the existing
bus routes, and used it as the benchmark to compare the new integrated AV-PT scenarios. The
simulation was run for 100 iterations. Figure 5 compares the observed numbers of boardings
in the CEPAS data with the simulated numbers in the benchmark model in every 15 minutes
interval. For most time intervals, the range of the simulated results captures the empirical
values.
Figure 5. Number of boardings comparison between simulation and CEPAS data
We also compared the share of ridership in each bus route between the simulated results and
the observation using the root-mean-square error (RMSE):
N R
åå ( rˆ - r nr )
2
nr
RMSE = n =1 r =1
, (4)
N ×R
where 𝜌89: is the simulated share of ridership of the bus route r at the n-th simulation; 𝜌9: is
the observed share of ridership of bus route r at the n-th simulation; N is the total number of
simulation runs; R is the total number of bus routes. The RMSE was 0.026, indicating that the
11
simulated demand in each bus route well fitted the actual patronage derived from the CEPAS
data7.
Integrated AV-PT scenarios

As proposed in Section 3.2, we simulated the scenarios where shared AVs would accommodate
the first-mile service as a replacement of the 11 low-demand bus routes. 90% of the passengers
would still choose the bus, as per usual, whereas the remaining 10% would hail an AV for the
ride to the MRT station. After repurposing the 11 bus routes, we evaluated the service
performance of different AV fleet sizes (from 10 to 35 vehicles) and with two extreme settings
of sharing preference of the AV riders: (1) no one is willing to share the ride, and (2) everyone
is willing to share. A total of 52 AV-PT scenarios were developed, and for each scenario, 100
times of simulations were run.
Among the many factors affecting system performance, we look into three perspectives to
evaluate the integrated AV-PT system:
• Passenger perspective (i.e., service quality): travel time, particularly the out-of-vehicle
travel times (OVT) including walking time and waiting time
• System perspective (i.e., road traffic): we use the passenger car unit kilometers (PCU-km)
as the approximation to the road resource usage for mixed-vehicle fleets (BITRE, 2015;
BTRE, 2007). The PCU-km is derived as the product of the vehicle-kilometers traveled
and the passenger car equivalent (PCE) factor. The empirical PCE factor for buses in
Singapore is approximately 2.75 (Fan, 1990; Yeung et al., 2015). As the AV is classified
as a personal car, its PCE factor is one.
• Business perspective (i.e., operating cost and revenue): whether the AV service can be
financially viable based on the proposed pricing model.
The simulation results of the 52 scenarios are plotted in Figure 6, where each dot represents
the average value of 100 simulation runs; the triangles represent the non-ridesharing setting,
whereas the circles represent the all-ridesharing setting. The dashed line shows the average
value of 100 simulation runs of the bus-only benchmark scenario.
Owing to the heat and humidity in Singapore, the outdoor walking time and the waiting time
are of high importance to service quality. The average OVT in the benchmark bus-only scenario
is approximately 420 seconds. We recorded in the simulation the total time spent by each
passenger from his/her incidence being generated to the time when he/she boards a bus or AV.
Figure 6(a) compares the average OVT of all passengers in the AV-PT scenarios with that in
the benchmark bus-only scenario. More AVs reduce the OVT until they saturate; and the OVT
is shorter when rides are shared for the AV fleet size less than 30, beyond which sharing no
longer affects the OVT. The first-mile travel demand can be served by only 17 AVs with
ridesharing for the average OVT to be on par with the benchmark scenario. However, without
sharing, we would need 22 AVs, an increase of fleet size by a third, to guarantee the same
service quality.
7
The visual comparison in Figure 5 and the RMSE are both aggregate measures, which do not capture the
variations in the number of passenger boardings at the bus stop level, resulting in an overestimation of the realism
of simulation.
12
Figure 6(b) indicates the total PCU-km in different AV-PT scenarios. By repurposing the 11
low-demand bus routes based on the strategies in Appendix C (mostly to reduce detours), we
can save a total of 860 PCU-km, which we consider as the acceptable “quota” for AVs in order
not to increase the total traffic in the study area. The simulation results indicate that about
18~19 AVs can be accommodated in the study area within the traffic “quota.” Since our
simulation model only uses the basic heuristic AV dispatching and assignment algorithms, the
AV-PT system performance can be further improved with more advanced algorithms.
Figure 6. Simulation results
Figure 6(c.1) and 6(c.2) plot the estimated revenue and cost per kilometer for various AV fleet
sizes. The operation cost of AVs is based on the car leasing plan between Uber and SMOVE8,
an emerging car-sharing company in Singapore. We assume that the AVs are owned by SMOVE,
or its equivalent, and rented to the AV operators using the same leasing plan: the costs being
calculated based on the usage of vehicles in terms of both time and distance. The simulation
results show that smaller fleet size has higher profits per kilometer. Sharing increases the profit
margin per kilometer. When the fleet size increases, the profit gap between sharing and non-
sharing scenarios shrinks. The integrated AV-PT service is financially viable in the tested range
of fleet size assuming the above cost structure and the pricing model.
8
See https://www.smove.sg/#pricing for more details.
13
4. Discussion
In addition to the 90%-10% scenario (90% by high demand bus routes and 10% by AV)
reported above, we also tested two alternative AV-PT scenarios: the first is an extreme case
where all first mile bus services are replaced by the AVs, and the second is to use AV to serve
45% of first-mile bus demand by preserving only the 5 busiest bus routes. Appendix D shows
that in both alternative scenarios, the system performance worsens, i.e., there will be far larger
PCU-km in the system to maintain the same level of service quality as in the benchmark case.
We should not take it for granted that an integrated AV-PT system will automatically improve
the status quo. Instead, it requires attentive research and innovative design to identify the
opportunity window where an optimal combination of AV and PT can enhance service quality,
utilize buses more efficiently, occupy fewer road resources, and be financially sustainable.
The Ministry of Transport (2017) of Singapore is planning for the pilot deployment of AVs in
several planning areas. The integration of AV into the PT system is one of the primary
objectives, including the exploration of operational characteristics and business models. Our
study proposes an integrated AV-PT system based on Singapore’s PT organizational structure
and demand characteristics. Specifically, we envision a synergic system in the context of the
first-mile service during morning peak hours, where high-demand bus routes are preserved,
low-demand routes are repurposed, and shared AVs are introduced as an alternative, and
evaluate the system from the passenger, business and system perspectives using agent-based
simulation.
The real world is a complex nonlinear system. There is a large room to improve the simulation
model in future research:
• The 90%-10% scenario is an intuitive design. A more systematic approach to optimize the
PT-AV split should be considered.
• The fixed AV demand is a strong assumption. A full multimodal demand model should be
developed based on consumer preference surveys conducted in Singapore for the
estimation of the mode choice between the bus, AVs and walking in the first-mile access
to MRT stations.
• A dynamic interaction between demand (modal choice) and supply (AV and bus fleets)
should be taken into consideration (See Wen et al. 2018).
• The dispatching of AV in the current simulation is based on simple heuristics. A more
efficient ridesharing model can be incorporated with more sophisticated algorithms to
optimize the ridesharing efficiency.
• Other sub-models such as traffic flow models and bus dwelling time models can also be
integrated to make the simulation closer to the real-world circumstances.
Finally, the case study presented in the paper is only one of many possible ways to integrate
the shared AVs into the PT system from a planning perspective. As in Figure 1, there are many
alternative schemata from the viewpoints of operation, governance, technology, and
consumption, e.g., pricing or ticketing integration, information sharing, institutional design, etc.
Private companies can also be potential AV fleet owners and operators. Various ownership
structures affect the interests and incentives of different players, leading to different behaviors
in the integrated mobility system. Each operator may have its own strategies toward its peers,
e.g., one may choose to compete against or to coordinate with the others.
14
Acknowledgement
The research is supported by the National Research Foundation (NRF), Prime Minister’s Office,
Singapore, under CREATE programme, Singapore-MIT Alliance for Research and Technology
(SMART) Centre, Future Urban Mobility (FM) IRG. The authors would like to thank the
anonymous reviewers and the members of MIT JTL Urban Mobility Lab and MIT Transit Lab
for their comments.
References
Abkowitz, M., Tozzi, J., 1987. Research contributions to managing transit service reliability. J. Adv.
Transp. 21, 47–65.
Adebisi, O., Hurdle, V.F., 1982. Comparing fixed-route and flexible-route strategies for intraurban
bus transit. Transp. Res. Rec. 854, 37–43.
Aldaihani, M.M., Quadrifoglio, L., Dessouky, M.M., Hall, R., 2004. Network design for a grid hybrid
transit service. Transp. Res. Part A Policy Pract. 38, 511–530.
Alonso-Mora, J., Samaranayake, S., Wallar, A., Frazzoli, E., Rus, D., 2017. On-demand high-capacity
ride-sharing via dynamic trip-vehicle assignment. Proc. Natl. Acad. Sci. U. S. A. 114, 462–467.
Atasoy, B., Ikeda, T., Song, X., Ben-Akiva, M., 2015. The concept and impact analysis of a flexible
mobility on demand system. Transp. Res. Part C Emerg. Technol. 56, 373–392.
Bansal, P., Kockelman, K.M., Singh, A., 2016. Assessing public opinions of and interest in new
vehicle technologies: An Austin perspective. Transp. Res. Part C Emerg. Technol. 67, 1–14.
Bruun, E.C., 2013. Organizational structures and contract models, in: Better Public Transit Systems :
Analyzing Investments and Performance. Routledge, Abingdon, UK, pp. 237–253.
Bureau of Infrastructure Transport and Regional Economics (BITRE), 2015. Traffic and congestion
cost trends for Australian capital cities. Australian Government, Canberra, Australia.
Bureau of Transport and Regional Economics (BTRE), 2007. Estimating urban traffic and congestion
cost trends for Australian cities, Working paper 71. Canberra, Australia.
Cervero, R., 2007. Transit-Oriented Development’s ridership bonus: A product of self-selection and
public policies. Environ. Plan. A 39, 2068–2085.
Cervero, R., Caldwell, B., Cuellar, J., 2013. Bike-and-ride: Build it and they will come. J. Public
Transp. 16, 83–105.
Chang, S.K., Schonfeld, P.M., 1991a. Optimization models for comparing conventional and
subscription bus feeder services. Transp. Sci. 25, 281–298.
Chang, S.K., Schonfeld, P.M., 1991b. Integration of Fixed- and Flexible-Route Bus Systems. Transp.
Res. Board J. Transp. Res. Board 1308, 51–57.
Chen, T.D., Kockelman, K.M., 2016. Management of a shared autonomous electric vehicle fleet:
Implications of pricing schemes. Transp. Res. Board J. Transp. Res. Board 2572, 37–46.
Chien, S., Schonfeld, P., 1998. Joint optimization of a rail transit line and its feeder bus system. J.
Adv. Transp. 32, 253–284.
Chong, Z.J., Qin, B., Bandyopadhyay, T., Wongpiromsarn, T., Rankin, E.S., Ang Jr., M.H., Frazzoli,
E., Rus, D., Hsu, D., Low, K.H., 2011. Autonomous personal vehicle for the first- and last-mile
transportation services. 2011 IEEE 5th Int. Conf. Cybern. Intell. Syst.
https://doi.org/10.1109/ICCIS.2011.6070337
Correia, G.H.A., van Arem, B., 2016. Solving the User Optimum Privately Owned Automated
Vehicles Assignment Problem (UO-POAVAP): A model to explore the impacts of self-driving
vehicles on urban mobility. Transp. Res. Part B Methodol. 87, 64–88.
Costa, Á., 1996. The organisation of urban public transport systems in Western European
metropolitan areas. Transp. Res. Part A Policy Pract. 30, 349–359. https://doi.org/10.1016/0965-
8564(96)00002-X
Daganzo, C.F., 1978. An approximate analytic model of many-to-many demand responsive
transportation systems. Transp. Res. 12, 325–333. https://doi.org/10.1016/0041-1647(78)90007-
2
DeMaio, P., 2009. Bike-sharing: history, impacts, models of provision, and future. J. Public Transp.
12, 41–56. https://doi.org/http://dx.doi.org/10.5038/2375-0901.12.4.3
Dial, R.B., 1995. Autonomous dial-a-ride transit introductory overview. Transp. Res. Part C Emerg.
15
Technol. 3, 261–275.
Diana, M., Quadrifoglio, L., Pronello, C., 2009. A methodology for comparing distances traveled by
performance-equivalent fixed-route and demand responsive transit services. Transp. Plan.
Technol. 32, 377–399. https://doi.org/10.1080/03081060903119618
Djavadian, S., Chow, J.Y.J., 2016. Agent-based day-to-day adjustment process to evaluate dynamic
flexible transport service policies. Transp. B Transp. Dyn. 1–26.
Dong, X., DiScenna, M., Guerra, E., 2017. Transit user perception of driverless buses, in: TRB 96th
Annual Meeting Compendium of Papers. Washington, DC, pp. 1–15.
Eboli, L., Mazzulla, G., 2011. A methodology for evaluating transit service quality based on
subjective and objective measures from the passenger’s point of view. Transp. Policy 18, 172–
181. https://doi.org/10.1016/j.tranpol.2010.07.007
Fagnant, D.J., Kockelman, K.M., 2014. The travel and environmental implications of shared
autonomous vehicles, using agent-based model scenarios. Transp. Res. Part C Emerg. Technol.
40, 1–13.
Fagnant, D.J., Kockelman, K.M., Bansal, P., 2015. Operations of shared autonomous vehicle fleet for
Austin, Texas, market. Transp. Res. Board J. Transp. Res. Board 2536, 98–106.
Fan, H.S.L., 1990. Passenger car equivalents for vehicles on Singapore expressways. Transp. Res.
Part A Gen. 24, 391–396.
Fan, W., Machemehl, R.B., 2002. Characterizing bus transit passenger waiting times. 2nd Mater.
Spec. Conf. Can. Soc. Civ. Eng. .
Frumin, M., Zhao, J., 2012. Analyzing passenger incidence behavior in heterogeneous transit services
using smartcard data and schedule-based assignment. Transp. Res. Rec. J. Transp. Res. Board
2274, 52–60.
Galland, S., Knapen, L., Yasar, A.-U.-H., Gaud, N., Janssens, D., Lamotte, O., Koukam, A., Wets, G.,
2014. Multi-agent simulation of individual mobility behavior in carpooling. Transp. Res. Part C
Emerg. Technol. 45, 83–98.
Hamer, P., 2010. Analysing the effectiveness of park and ride as a generator of public transport mode
shift. Road Transp. Res. A J. Aust. New Zeal. Res. Pract. 19, 51–61.
Housing & Development Board (HDB), 2015. Key statistics: HDB annual report 2014/2015 [WWW
Document]. URL http://www10.hdb.gov.sg/ebook/ar2015/html5/index.html?opf=tablet/Key
Statistics.xml
Jung, J., Jayakrishnan, R., 2016. Simulation framework for modeling large-scale flexible transit
systems. Transp. Res. Rec. 2466, 31–41.
Krueger, R., Rashidi, T.H., Rose, J.M., 2016. Preferences for shared autonomous vehicles. Transp.
Res. Part C Emerg. Technol. 69, 343–355.
Kyriakidis, M., Happee, R., de Winter, J.C.F., 2015. Public opinion on automated driving: Results of
an international questionnaire among 5000 respondents. Transp. Res. Part F Traffic Psychol.
Behav. 32, 127–140.
Lenz, B., Fraedrich, E., 2016. New mobility concepts and autonomous driving: The potential for
change, in: Markus Maurer, J. Christian Gerdes, Barbara Lenz, Hermann Winner (Eds.),
Autonomous Driving: Technical, Legal and Social Aspects. Springer Berlin Heidelberg, Berlin
Heidelberg, pp. 173–191. https://doi.org/10.1007/978-3-662-48847-8
Lesh, M.C., 2013. Innovative concepts in first-last mile connections to public transportation. Third
Int. Conf. Urban Public Transp. Syst. https://doi.org/10.1061/9780784413210.007
Li, X., Quadrifoglio, L., 2010. Feeder transit services: Choosing between fixed and demand
responsive policy. Transp. Res. Part C Emerg. Technol. 18, 770–780.
https://doi.org/10.1016/j.trc.2009.05.015
Li, X., Quadrifoglio, L., 2009. Optimal Zone Design for Feeder Transit Services. Transp. Res. Rec. J.
Transp. Res. Board 2111, 100–108. https://doi.org/10.3141/2111-13
Liang, X., Correia, G., van Arem, B., 2016. Optimizing the service area and trip selection of an
electric automated taxi system used for the last mile of train trips. Transp. Res. Part E Logist.
Transp. Rev. 93, 115–129.
Lu, Z., Du, R., Dunham-Jones, E., Park, H., Crittenden, J., 2017. Data-enabled public preferences
inform integration of autonomous vehicles with transit-oriented development in Atlanta. Cities
63, 118–127.
16
Martínez, L.M., Correia, G.H.A., Viegas, J.M., 2015. An agent-based simulation model to assess the
impacts of introducing a shared-taxi system: an application to Lisbon (Portugal). J. Adv. Transp.
49, 475–495.
Ministry of Transport, 2017. Request for information for pilot deployment of autonomous vehicles as
a form of public transportation in upcoming development areas in Singapore [WWW
Document]. URL https://www.gebiz.gov.sg/ptn/opportunity/opportunityDetails.xhtml (accessed
12.21.17).
Mo, B., Shen, Y., Zhao, J., 2018. Impact of built environment on first- and last-mile travel mode
choice. Transp. Res. Rec.
Payre, W., Cestac, J., Delhomme, P., 2014. Intention to use a fully automated car: Attitudes and a
priori acceptability. Transp. Res. Part F Traffic Psychol. Behav. 27, 252–263.
Qiu, F., Li, W., Haghani, A., 2015. A methodology for choosing between fixed-route and flex-route
policies for transit services. J. Adv. Transp. 49, 496–509. https://doi.org/10.1002/atr.1289
Shaheen, S.A., Finson, R., 2003. Bridging the last mile: A study of the behavioral, institutional, and
economic potential of the Segway human transporter. 82nd Annu. Meet. Transp. Res. Board.
Shaheen, S.A., Guzman, S., Zhang, H., 2010. Bikesharing in Europe, the Americas, and Asia: Past,
present, and future. Transp. Res. Rec. J. Transp. Res. Board 2143, 159–167.
Shin, J., Bhat, C.R., You, D., Garikapati, V.M., Pendyala, R.M., 2015. Consumer preferences and
willingness to pay for advanced vehicle technology options and fuel types. Transp. Res. Part C
Silberg, G., Wallace, R., 2012. Self-driving cars: The next revolution [WWW Document]. URL
https://www.kpmg.com/US/en/IssuesAndInsights/ArticlesPublications/Documents/self-driving-
cars-next-revolution.pdf
Sørensen, C.H., Longva, F., 2011. Increased coordination in public transport: Which mechanisms are
available? Transp. Policy 18, 117–125.
Vakayil, A., Gruel, W., Samaranayake, S., 2017. Integrating shared-vehicle mobility-on-demand
systems with public transit. 96th Annu. Meet. Transp. Res. Board.
van de Velde, D.., 1999. Organisational forms and entrepreneurship in public transport: classifying
organisational forms. Transp. Policy 6, 147–157. https://doi.org/10.1016/S0967-
070X(99)00016-5
Wen, J., Chen, Y.X., Nassir, N., Zhao, J., 2018. Transit-oriented autonomous vehicle operations with
integrated demand-supply interaction, Transportatio Research Part C, under review
Wen, J., Zhao, J., Jaillet, P., 2017. Rebalancing Shared Mobility-on-Demand Systems: a
Reinforcement Learning Approach, in: IEEE ITSC 2017 The Third International Workshop on
Intelligent Public Transports: Toward the Next Generation of Urban Mobility.
Wilson, N.H.M., 1991. Organizational options for public transportation in the US. Transp. Plan.
Wilson, N.H.M., Hendrickson, C., 1980. Performance models of flexibly routed transportation
services . Transp. Res. Part B Methodol. 14, 67–78. https://doi.org/10.1016/0191-
2615(80)90033-8
Wilson, N.H.M., Weissberg, R.W., Hauser, J.R., 1976. Advanced Dial-a-ride algorithms research
project: Final report. Massachusetts Institute of Technology, Cambridge, MA.
Yap, M.D., Correia, G., van Arem, B., 2016. Preferences of travellers for using automated vehicles as
last mile public transport of multimodal train trips. Transp. Res. Part A Policy Pract. 94, 1–16.
Yeung, J.S., Wong, Y.D., Secadiningrat, J.R., 2015. Lane-harmonised passenger car equivalents for
heterogeneous expressway traffic. Transp. Res. Part A Policy Pract. 78, 270–361.
17
Appendix: Supplementary Information
A. Six potential organizational models for the integrated AV-PT system
Model #1: Laissez-faire structure

In a laissez-faire model, the authority never intervenes in the operations. The private operators
aim to increase market share and maximize their profits. They only operate in profitable areas
or routes and compete against each other for ridership, service coverage, and profitability.
If the service could be potentially profitable, AV operators enter that market as well. The natural
advantage of AVs (e.g., lower operating cost with automation replacing human drivers) may
trigger the traditional PT operator to optimize its workforce planning and crew scheduling
strategies for cost savings to compete. It may also enhance the quality of customer service in
conventional transit since lack of personal maintenance is one of the potential weaknesses of
the driverless system. Another possibility is the conventional transit operator may also reform,
replacing its human driver with driverless technology, making the whole transit system
driverless. This would lead to competition between two driverless systems.
However, the absence of planning and regulation could result in a series of negative
consequences, including separate fare and ticketing system, lack of coordination in network
planning and scheduling. The lack of coordination creates a barrier for passengers to use AV in
conjunction with conventional transit systems owing to the separate fare systems. The
competition in the market can also result in more traffic congestion in profitable areas with lack
of service in unprofitable areas and routes. Without coordination, each AV operator may install
its own vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) system with high capital
cost, which may not even be compatible with that of the other operators. In this case, the
absence of an effective system design fragments the transit market, deteriorates the quality of
service, and reduces revenues. Even with the AV technology, this market may not be sustainable
owing to the reduction of fleet size, service frequency, and the layoff of employees.
Model #2: TNC-regulated structure

This structure is close to the current organizational model with regulations on TNC firms in
many cities like New York City, London, and Singapore, where a vocational driver’s license is
required by the local transport authorities before operating the TNC services. In this model, the
authorities in these cities regulate the private operators to impose service standards upon these
firms, e.g., safety. These TNC firms are not financially supported by the public.
TNC firms like Uber are attempting to develop their own AVs. It is possible that current TNC
firms may become AV operators in the future. In this case, the AV operation can be licensed or
regulated by public authorities, with the service still provided by the private operators and not
publicly subsidized. The AV service operates independently of transit operations. Without the
strings on labor cost for human drivers, AV operators may be more competitive. The
consequences may be similar to model #1, which may lead to a fully driverless system for all
operators to reduce operating cost or a fragmented system. Although there are certain service
standards promulgated by the public authority, the whole system is still unlikely to be efficient.
Model #3: MBTA paratransit structure

This model is another structure in which private operators can be supported by the public
authorities if they offer public services. The model can be understood as replicating the “on-
demand paratransit pilot program” led by the Massachusetts Bay Transportation Authority
18
(MBTA). In this program, the on-demand service provided by Uber and Lyft serve passengers
with disabilities. The fare for each ride is only two US dollars, and the rest of the operating cost
is subsidized by the MBTA. In the program, the MBTA could potentially save a large amount
of cost for operating its own paratransit services. The ticketing and information systems are
still separate. There is also no coordination between Uber/Lyft and MBTA travel modes.
In contrast to model #2, this model forges a new partnership between private operators and PT
agencies. If these TNC firms upgrade their fleet to AVs, the same organizational structure still
applies. The authority subsidizes private AV operations targeting a specific type of service, e.g.,
paratransit for the elderly, people with disabilities, and children. This structure helps to
incentivize the private AV operators to provide more social but (maybe) unprofitable service.
Compared with the current human-driven paratransit service, the on-demand AV service is able
to serve the vulnerable population with greater flexibility and reduced cost.
Model #4: UK deregulation structure

This structure presents an organization option akin to the UK deregulation model (in areas
outside London). In contrast to the TNC-oriented models above, there is a central planning
authority in this structure overseeing the operations at a certain level and the service is more
integrated, although the operators are private and compete against each other. In profitable
areas and routes, the scenario is close to the models above, where the operators set up their
own fare structure, routing and scheduling, and information systems, and compete against each
other. However, the public authority may intervene in the operations in unprofitable areas by
regulating fares and routes or may provide transit service directly. The pros and cons of this
model are briefly summarized by Wilson (1991).
In a model with AVs, the advantages of this model are similar to the virtuous circle discussed
in Model #1 with efficient service, quick response to market changes, and technology
innovation. With basic interventions and regulation, the negative consequences envisioned in
Model #1 may not occur. However, private AV operators are still for-profit, which may squeeze
out the conventional operators from the market. Even in a total driverless transit system, owing
to the unrestricted market entry and exit, the AV service may still be discontinuous in the
unprofitable areas, leading to less demand for PT service in these areas.
Model #5: Scandinavian structure

This model accords to the well-known “Scandinavian” or “London” model (Costa, 1996; van
de Velde, 1999). The public transport and service goals are set and planned by the authority,
which then contracts out the planned transit service to private operators. Competition is
introduced in the model where the operators attempt to reach the service standards set by the
authority to enter the market. For instance, in Stockholm (Sweden), bus service is contracted
to several public or private operators, including Arriva (from Deutsche Bahn), Keolis
(headquartered in Paris), and Nobina (from Sweden). The PT authority ensures the coordination
of the transit system.
In the AV scenario, to ensure coordination, central planning from a public authority is needed.
Both AV and conventional PT operators compete for contracts to enter the market. The
operating cost of AV is lower, but the conventional transit operators may have advantages in
other conditions, e.g., higher security standards and more help from service personnel. The
ticketing and the customer information systems are more likely to be coordinated and integrated,
which may attract more riders in general, owing to its convenience.
19
Model #6: Fully coordinated structure
This model envisions a fully coordinated scenario without competition in or for the market.
Similar to model #5, the fare and information systems are coordinated and integrated. The
authority is responsible for the design of the network route structure and the designation of the
service areas of each operator, aiming at maximizing social welfare. The operators are
responsible for the operation, management, and fleet maintenance. In contrast to model #5,
where the operators are free to compete for all service routes and areas, the operators in this
model run their services in different areas, making comparisons of service performance
difficult.
This model is highly regulated. Different service areas are designated to AV and PT operators
by the public authority, and the operators do not compete against each other for more service
areas. For instance, the planner may design a trunk-feeder or diameter-tangent network
structure. In the network, the trunk (or diameter) routes may be assigned to the conventional
fixed-route fixed-schedule transit operator, while the feeder (or tangent) service connecting to
the arterial routes is assigned to on-demand AVs.
B. First and last mile travel demand in Singapore during workdays

Singaporean residents largely rely on the PT system to travel. According to the statistics from
the LTA, with a total population of 5.47 million, as of 2014, less than 10% of people own
private cars. Moreover, the total number of private cars is continuously decreasing, owing to
the shrinking quota of Certificate of Entitlement. Thus, the availability of private vehicle-based
travel modes is limited. The household interview travel survey (HITS) conducted in 2012 also
suggests that during the a.m. peak hours, 70% of commuters travel to work using PT, including
buses and the MRT, whereas only 16% of them drive cars.
The average MRT travel demand per hour during workdays is shown in Figure A.1, along with
the volumes of the first- and last-mile bus trips9. The percentages of the first- and last-mile bus
trips to the connected MRT trips in each hour of the day are also shown. During morning peak
hours, first-mile bus trips are more than the last-mile bus trips, while the pattern gets reversed
in the evening peak, which is reasonable for any workday. The chart also shows a clear pattern
of increase in the MRT travel demand during morning and evening peak hours. For instance,
during morning peak hours—between 7 and 9 a.m.—there are in total 450,000 passengers
entering the MRT stations. Among them, nearly 134,000 passengers (over 30%) take a bus trip
to reach the MRT station. The findings from the CEPAS database are consistent with the
corresponding modal share from the HITS data. Observing Figure A.2, the sizes of first- and
last-mile trips are quite similar in all MRT stations, i.e., the number of passengers boarding and
alighting from the MRT are almost equal. Both two charts suggest that the travel pattern during
workdays is very regular. Figure A.2 also illustrates the share of first-/last-mile trips in the areas
close to the boundary of Singapore Island is greater than that in the CBD area, which is
probably due to the higher density of MRT stations in the CBD area.
9
For all bus trips connecting with the MRT, we filtered the trip distance of less than 5 km to be the first-/last-mile
(bus) trip.
20
(1) First-/last-mile travel demand during workdays by hour
(2) First-/last-mile travel demand during workdays by MRT station

Figure A. First-/last-mile travel demand on workdays
(Data source: calculations by the authors based on CEPAS smart card data)
C. Bus routes and the repurposing methods
Footprints of bus routes in Tampines

Figure B maps the footprints of bus routes. Figure B.1 shows the spatial coverage of high-
demand bus routes (i.e., from route 291_1 to 18_1 in Figure 3), which in total service over 90%
of the first-mile bus travel demand in the Tampines area. Figure B.2 shows the coverage of
low-demand routes (i.e., from 38_2 to 10_2). The low-demand buses do not significantly
21
enhance the service coverage in this area. The frequency of low-demand buses is quite high
with less than 10-min headway but they carry low first mile demand.
(1) Footprint of high-demand bus routes
(2) Footprint of low-demand bus routes

Figure B. First-mile bus routes and origin-destination patterns
(Data source: calculation by the authors)
Repurposing low-demand bus routes

The details on each of the low-demand bus routes and the corresponding repurposing scenarios
are summarized below:
22
i. Rerouting to reduce the detour. The bus routes presented in Figures C(1) and C(2) both
use the Tampines MRT station as the terminal. The routes deviate largely. After
repurposing, we let these buses go directly to the terminal.
ii. Rerouting to bypass the center. The bus routes presented in Figures C(3), C(4), and C(5)
all pass by the MRT station area. However, the low ridership indicates that, although the
buses are designed to connect to the MRT station, only a few passengers use the service.
Therefore, these routes were redesigned to bypass the station area to mitigate the pressure
on the road traffic around the MRT station.
iii. Keeping the route as is. The bus routes presented from Figure C(6) to C(10) all go to the
MRT station directly. We kept the route as is; however, the bus stops in the Tampines area
were skipped.
iv. Shifting the destination to the adjacent MRT station. The bus routes presented in Figure
C(11) connect Tampines and Simei, which is the MRT station next to Tampines station.
However, the travel demand between the two MRT stations by bus is low; the demand
between the two can be either serviced by the MRT or by the AVs. Hence, we shifted these
two bus routes to end at the Simei MRT station.
i. Rerouting to reduce detour:
(1) Bus route 3_2 (2) Bus route 39_2

ii. Rerouting to bypass the center:
(3) Bus route 21_1 (4) Bus route 21_2 (5) Bus route 168_2
iii. Keeping the route as is:
23
iv. Shifting the destination to the adjacent MRT station:
(11) Bus route 38_2

Figure C Repurposing low-demand bus routes
D. Simulation results for alternative scenarios

The section demonstrates two alternative scenarios with various repurposing proportions of
shared AV, based on Figure 3. For each scenario, 12 sub-scenarios were simulated with
different fleet sizes. With each AV fleet size, we run the simulation for 100 iterations. Every
passenger is assumed to be willing to share the ride. We focus on the impact on the road
resources, i.e. PCU-km. The results of the two scenarios are illustrated in Figure D.1 and D.2,
respectively.
Figure D. AV fleet size vs total PCU-kilometers traveled in alternative scenarios
Figure D.1 shows the PCU-km assuming the AVs serving 45% of the first-mile demand
originally served by bus. The five busiest bus routes are kept in the system while the other 22
routes are repurposed. The total saving of PCU-km by repurposing the bus routes are about
2,400 km. However, to serve the first-mile travel demand with the same level of service, at
least 96 shared AVs are required, generating nearly 4,300 PCU-km. The total PCU-km
increases. Similarly, Figure D.2 shows the results where all first-mile bus demands are replaced
by the shared AVs. By simply removing all bus routes in this area, about 7,600 PCU-km are
released. However, 208 shared AVs are needed, generating over 9,100 PCU-km. In both
24
scenarios, much more road resources are occupied, making the road traffic worse and harming
the performance of the overall mobility system. We are able to improve the system performance
based on an intuitive design of repurposing only the low-demand buses that serve 10% of the
demand. However, the comparison to the alternative scenarios shows that we cannot take it for
granted that an integrated AV-PT system will automatically improve the system performance.
25
J. Mod. Power Syst. Clean Energy (2019) 7(3):433–448
https://doi.org/10.1007/s40565-019-0516-7
Interdependence between transportation system and power

distribution system: a comprehensive review on models
and applications
Wei WEI1 , Danman WU1, Qiuwei WU2, Miadreza SHAFIE-KHAH3,4,
João P. S. CATALÃO4,5
Abstract The rapidly increasing penetration of electric appropriate incentives, the energy storage capability of
vehicles in modern metropolises has been witnessed during electric vehicle offers a unique opportunity to facilitate the
the past decade, inspired by financial subsidies as well as integration of distributed wind and solar power generation
public awareness of climate change and environment pro- into power distribution system. The above trends call for
tection. Integrating charging facilities, especially high- thorough investigation and research on the interdependence
power chargers in fast charging stations, into power dis- between transportation system and power distribution
tribution systems remarkably alters the traditional load system. This paper conducts a comprehensive survey on
flow pattern, and thus imposes great challenges on the this line of research. The basic models of transportation
operation of distribution network in which controllable system and power distribution system are introduced,
resources are rare. On the other hand, provided with especially the user equilibrium model, which describes the
vehicular flow on each road segment and is not familiar to
the readers in power system community. The modelling of
CrossCheck date: 15 January 2019 interdependence across the two systems is highlighted.
Taking into account such interdependence, applications
Received: 16 November 2018 / Accepted: 15 January 2019 / Published
ranging from long-term planning to short-term operation
online: 2 April 2019
The Author(s) 2019 are reviewed with emphasis on comparing the description
& Wei WEI
of traffic-power interdependence. Finally, an outlook of
wei-wei04@mails.tsinghua.edu.cn prospective directions and key technologies in future
Danman WU
research is summarized.
wudm18@mails.tsinghua.edu.cn
Qiuwei WU
Keywords Charging stations, Electric vehicles,
qw@elektro.dtu.dk Interdependence, Power flow, Power distribution network,
Miadreza SHAFIE-KHAH
Transportation network, User equilibrium
miadreza@gmail.com
João P. S. CATALÃO
catalao@ubi.pt 1 Introduction
1
State Key Laboratory of Power System, Department of Urban commute vehicles and coal-fired/gas-fired power
Electrical Engineering, Tsinghua University, Beijing, China
generators are considered as major fossil fuel consumers
2
Department of Electrical Engineering, Technical University and carbon dioxide emitters. To cut down fossil fuel
of Denmark, Copenhagen, Denmark
dependence, many countries have set clear goals for pro-
3
School of Technology and Innovations, University of Vaasa, moting renewable energy generation [1, 2] and electric
Vaasa, Finland
vehicles (EVs) [3], and promulgated market incentives to
4
Institute for Systems and Computer Engineering, Technology achieve their targets.
and Science (INESC TEC), Porto, Portugal
5
Faculty of Engineering, University of Porto, Porto, Portugal
123
434 Wei WEI et al.
Thanks to the development of battery technology, the distribution infrastructures. In the future, if EVs account
driving ranges of recent EV models have been significantly for a large portion of traffic flow, it is also possible to
improved. At current stage, standard EV models have a actively influence the traffic condition by elaborately
driving range of 100-250 km, while some elite ones can announcing difference electricity prices at individual
reach 300-500 km without charging [4]. However, the lack charging stations. Such interdependence phenomenon and
of public charging facility brings some inconvenience to synergetic potential have been reported in [10].
EV owners. To elevate the usage of EVs in modern smart To better understand traffic-power interdependence, this
cities, adequate charging facilities are indispensable. There paper launches an in-depth survey in this line of research.
are two different charging modes: slow charging and fast In Section 2, respective network flow models of the
charging. The former one usually takes place at home. transportation system and the power distribution system are
Because the driving patterns of most citizens are regular, presented; then the interactive model, namely the network
the residential charging power demand is predictable. The equilibrium, is introduced. In Section 3, applications found
latter one happens during travel, and the power demand in the current literature are reviewed, mainly from the
forecast for fast charging stations could be more difficult, aspects of system planning and operation. Section 4 envi-
because it is influenced by many factors such as the sions prospective topics in this young and active research
vehicular flow distribution, road congestion, electricity field. Section 5 concludes this paper.
price, etc. Today, the widely used charger CHAdeMO (for
fast charging) has a rated power of 50 kW, and that of
Tesla Supercharger is 120 kW [4]. If a fast charging station 2 Network flow models
serves dozens of EVs, the total demand easily reaches 1
MW or above. Meanwhile, dispatchable resources (espe- Network flow here refers to the steady-state distributions
cially large generators) in the companion power distribu- of vehicular flow on each road in the transportation net-
tion network (PDN) are rare. It has been widely work and bus voltage/line power flow in the distribution
acknowledged that distribution system operator should take network, which are respectively determined by a traffic
effective measures to resolve the impact of massive assignment problem (TAP) and an optimal power flow
charging of EV fleets [5–7]. (OPF) problem introduced in Section 2.1 and Section 2.2.
Existing research on domestic EV charging scheduling Their connection is elaborated in Section 2.3.
is relatively mild, because the daily driving pattern is
regular. This paper will mainly focus on the on-road fast 2.1 Transportation network model
charging station, which creates a bilateral tie between the
transportation system and the power distribution system. Similar to the power flow problem, it is important for the
From the transportation system side, EVs plan their routes transportation authority to know road traffic flow distri-
according to road congestion pattern and locations of fast butions in steady state. Such an analog in transportation
charging stations. The number of EVs receiving service engineering is called TAP. However, unlike a power sys-
determines the electric power demand at a fast charging tem where an operator can directly control generators, the
station. In traditional power system research, to acquire the traffic flow distribution in a transportation system sponta-
power demand, vehicle arrival rates are exogenously given neously reaches a steady state owing to the rationality of
as constants or probability distributions, or developed from individual travelers. The classic model in [11] will be
queueing theory [8]. This assumption is acceptable when introduced in this section.
the penetration of EVs is low and charging stations are A transportation network can be described as a con-
distant from each other; otherwise, spatial correlation could nected graph GT ¼ ðTN ; TA Þ, where TN denotes the set of
appear: if long queue occurs at one charging station, dri- nodes, including origins and destinations of vehicles, as
vers would probably seek service elsewhere. From the well as intersections in the physical system; TA is the set of
power system side, a charging station has certain capacity arcs, representing roads in the physical system. The con-
and should also obey system security constraints, such as nection topology is described by node-link incidence
bus voltage limits. Unexpected outage may influence traffic matrix K, whose dimension is jTN j jTA j. An element of K
condition. According to [9], on 19 May 2018, in the city of is described as (1). Each column of K is associated with a
Shenzhen in south China, more than 500 charging poles in link with 1 (or - 1) at the entry corresponding to the
5 charging stations were out of service due to power system entrance (or exit) node.
maintenance. As a result, nearly 2700 electric taxis failed
to get recharged in time, and long queues emerged in
adjacent charging stations. This event evidently showed the
interdependence between transportation and power
123
Interdependence between transportation system and power distribution system: a… 435
8
SO is an ideal concept in theoretical research; UE is
< þ1
> node i is the head of link j
Kij ¼ 1 node i is the tail of link j likely to happen because it captures the selfish behavior of
>
: ð1Þ motorists. Hereinafter, any reasonable outcome of traffic
0 node i and link j are disconnected
flow distribution will be referred to as a traffic assignment,
including SO and UE. In this sense, traffic assignment has a
broader content than UE. Nonetheless, because UE better
Let (r, s) be a pair of origin and destination (O-D).
fits the reality, we will mainly discuss UE in the rest of this
Traffic demand qrs is defined as the quantity of vehicles
paper.
travelling from origin node r to destination node s, and the
jTN j jTN j matrix Q is the system O-D demand matrix,
2.1.1 Beckmann model
which can be non-symmetric. A path consists of connected
arcs between ðr; sÞ. All available paths connecting O-D
To quantify road travel time, a natural assumption is that
pair ðr; sÞ are denoted by K rs , and each path is indexed by
the travel time ta on arc a solely depends on xa and is
k 2 K rs . fkrs is the traffic flow carried by path k
independent of the condition in remaining parts of the
between ðr; sÞ. The sum of path flows must meet the
transportation system. A commonly used candidate is the
traffic demand, i.e.
X bureau of public road (BPR) function in [12]:
fkrs ¼ qrs 8ðr; sÞ ð2Þ " 4 #
k2K rs 0 xa
ta ðxa Þ ¼ ta 1 þ 0:15 8a ð5Þ
ca
Topological relations between paths and links are
depicted via a link-path incidence matrix D = (Drs), where ta0 is the free travel time; ca is sometimes called the
8ðr; sÞ, where the sub-matrix Drs with a dimension of capacity of arc a. However, it is not a mandatory upper
jTA j jK rs j corresponds to certain O-D pair ðr; sÞ, and its bound of xa; instead, the travel time ta quickly grows
element is defined as: when xa [ ca , preventing further congestion. To model a

1 arc a belongs to path k strict capacity limit ca, the Davidson function has been
drs
ak ¼ proposed in [13]:
0 otherwise ð3Þ

0 Jxa
t a ðx a Þ ¼ t a 1 þ 8a ð6Þ
ca xa
With the help of link-path incidence matrix D, the traffic
flow xa on arc a is a linear function of path flow fkrs : where J is a parameter which should be calibrated from
XX real data.
xa ¼ fkrs drs
ak 8a ð4Þ
k
Provided with arc travel time, the total travel
ðr;sÞ
time crs
k perceived by a single traveler on path k
It should be mentioned that the vehicular flow in traffic between ðr; sÞ is given by
assignment is non-atomic: the flow is a real number; the X
crs
k ¼ ta ðxa Þdrs
ak 8k; 8ðr; sÞ ð7Þ
system impact of a single vehicle is negligible. a2TA
For most drivers, travel time is the primary concern. Let
ta be the travel time on link a; it only depends on xa in the Apparently, when someone can shorten travel time by
simplest case (classic model). Traffic assignment aims to using an alternative path, the traffic flow pattern will not be
identify a reasonable outcome of traffic flow distribution stable. The UE pattern occurs only if travel times on all
from feasible set (2)-(4). Two criteria have been used paths are equal. This is similar to the equal
introduced: incremental cost criterion in power system economic
dispatch, and is known as the Wardrop principle in
1) Social optimum (SO). Traffic flow pattern is said to be
P transportation engineering, which is formally stated as
social optimal if the total travel time xa ta reaches follows [11, 14].
a
minimum. In this setting, a central operator determines Wardrop principle: the traffic flow reaches a UE if travel
travel plans on behalf of all travelers, and travelers are times on all used paths between any O-D pair are equal,
willing to cooperate in order to use the entire and no greater than those which would be perceived on any
transportation system in the most effective way. unused path.
2) User equilibrium (UE). Travelers determine their The above condition has a logic form: Aurs,
routes individually in order to minimize their own
travel times. In this setting, a stable state emerges if no
one is willing to alter his current route.
123
436 Wei WEI et al.

crs
k ¼u
rs
k 2 K rs and fkrs [ 0; 8ðr; sÞ vehicle must keep a safety distance with adjacent ones, and
ð8Þ
ck [ urs
rs
k 2 K rs and fkrs ¼ 0; 8ðr; sÞ the safety distance increases with the growth of speed,
resulting in a drop in vehicle density. Consider an extreme
where urs is the minimal travel time between (r, s). These case: the safety distance is a constant and the vehicle
logical conditions can be easily interpreted by a nonlinear density is maximum regardless of vehicle speed. Under this
complementarity problem (NCP): situation, in order to increase xa , every vehicle must move

0 fkrs ?ðcrs rs
k 2 K rs ; 8ðr; sÞ faster, which in turn leads to a drop (not rise) of the travel
k u Þ0 ð9Þ
s:t: ð2Þ; ð4Þ; ð5Þ; ð7Þ time. To reconcile this conflict, a new traffic assignment
model has been developed in [16, 17] based on weaker
where notation 0 a?b 0 stands for a 0; b 0, and assumptions without relying on a specific latency function
ab ¼ 0, or in other words, at most one of a and b can take ta .
a strictly positive value. However, solving NCP (9) for a
large-scale system is still challenging due to its non-con- Assumption 1 Arc flow cannot exceed its capacity:
vexity introduced by the first complementarity and slack- xa ca 8a ð12Þ
ness condition.
Here the capacity ca may have a different value
Fortunately, it turns out that (9) exactly constitutes the
compared to that in Beckmann model.
Karush-Kuhn-Tucher (KKT) optimality conditions of the
following convex optimization problem [11, 15]:
8 P R xa Assumption 2 Arc travel time is equal to ta0 if arc flow is
< min 0 ta ðhÞdh less than its capacity; delay occurs only if arc flow reaches
a2TA
: ð10Þ its capacity, i.e.
s:t: ð2Þ; ð4Þ; fkrs 0 8k; 8ðr; sÞ (
ta ¼ ta0 xa ca
0
ta ta xa ¼ ca ð13Þ
where the arc travel time ta can be BPR function (5),
Davidson function (6), or other empirical function ta ðxa Þ

which is increasing. To see the convexity of the objective Define vectors x ¼ ðxa Þ; c ¼ ðca Þ; t0 ¼ ta0 ; 8a. The
function, its second-order derivative is calculated: Nesterov SO model is cast as a linear program (LP) as
Z xa follows:
d2 dta ðxa Þ
t a ð h Þdh ¼ ð11Þ
dx2a 0 dxa min xT t0 ð14Þ
x; f
Because arc travel time ta must be increasing in xa , the
s:t: x Kf ¼ 0 ð15Þ
second-order derivative is always positive, and thus the
objective function is strictly convex. Therefore, problem Ef ¼ q f 0 ð16Þ
(10) can be easily solved by a general-purpose nonlinear xc : k ð17Þ
program (NLP) solver. The optimal solution xa uniquely
determines traffic flow distribution at UE pattern. Please where (15) and (16) are the compact forms of (4) and (2),
note that Davidson function actually imposes bounds on respectively; K is the node-link incidence matrix; f is the
xa , so problem (10) may become infeasible if the traffic vector of path flow variables; E is a zero-one matrix cor-
demand is too high. responding to the coefficients of (2). As long as the
capacity is adequate, congestion will not happen in the SO
2.1.2 Nesterov model model; otherwise, problem (14)-(17) will be infeasible.
According to the physical interpretation of Lagrangian dual
The intuition of latency function ta ðxa Þ in Beckmann multipliers, k ¼ ðka Þ, 8a 2 TA corresponding to the flow
model is apparent: the more vehicles travelling on a road, capacity constraint (17) represents the delay that a single
the more it will be congested. This is reasonable if the user would perceive when travelling through a congested
traffic flow xa varies in a certain range. If we dig into its arc. Let x and k be the optimal primal and dual variables,
physical reality, inconsistency would emerge when xa an important conclusion drawn in [16, 17] is: traffic
keeps growing, which has been reported in [16, 17]. Traffic assignments ðx ; t0 Þ and ðx ; t0 þ k Þ reach SO and UE,
flow is the number of vehicles passing through a cross- respectively.
section per unit time, i.e.: flow equals speed multiplied by In Beckmann model, road travel time ta ðxa Þ only
density. When xa grows large, neither the vehicle speed depends on xa . However, in Nesterov model, the delay on
nor density can take a very small value. However, every each road is determined by the utilization of the whole
123
network. The two models are comprehensively compared destination, route, and parking location choices of GVs and
on some benchmark instances and real-world transportation EVs subject to driving distances is proposed in [26].
networks in Switzerland [18], demonstrating that both To mimic drivers’ charging decision meticulously, en-
models provide similar congestion patterns. However, the route energy consumption should be simulated in the UE
travel times offered by two models are generally incom- model. Based on different assumptions on the relations
parable, because they rest on different assumptions of arc among battery state-of-charge (SoC), recharging time and
latency. Nesterov model is an LP, and can be embedded road traffic flows, three sophisticated UE models are
into another high-level optimization problem more con- expounded in [27]. Two of them are cast as convex traffic
veniently in form of KKT optimality condition or primal- assignment problems, and the third one yields an NCP.
dual optimality condition. Similar work has been found in [28, 29] considering battery
An implicit assumption in setting up problems (10) and swapping stations. Trip chains of EVs are incorporated in
(14) is that the path sets K rs ; 8ðr; sÞ are available. This can the UE model studied in [30]. Limited driving ranges and
be done by enumerating all usable paths K rs for every O-D recharging needs of EVs are main concerns. A UE model
pair in advance. Nevertheless, such an enumeration could with charging-in-motion lanes are portrayed in [31]. A
be time-consuming and also unnecessary for large-scale mixed UE model with GVs and EVs as well as autono-
systems, as most paths existing in theory will never be used mous-vehicle-only lanes is formulated in [32]. A special
in practice. To vanquish this difficulty, a reasonable mean UE model is put forward in [33] to investigate the impact
is to start with a restricted traffic assignment problem based of travel time display on driver route choices.
on subsets of K rs ; 8ðr; sÞin which the paths are most likely Above work originating from transportation research
to be visited, and then test whether a new path will be community puts more emphasis on the traffic flow distribu-
activated according to the congestion pattern at the tion. Operation of charging stations and its connection with
restricted UE. This procedure is mathematically stream- power systems are neglected, but deserve more attentions.
lined in [19]. An adaptive path generation algorithm is
developed, in which path generation is formulated as an 2.2 PDN model
MILP based on K.
Economical power system operation entails solving an
2.1.3 Other UE models OPF problem [34]. The most well-known OPF model renders
a non-convex quadratically constrained program with the
The basic UE models presented above capture conges- traditional bus-injection model (BIM). The active power
tion effect, the foremost concern in traffic engineering. For flow can be approximated via the direct-current OPF
more extensive discussions about UE models considering (DCOPF) [34], which is an LP. However, the charging
elastic traffic demands, travel pattern and destination facilities in urban areas are connected to a PDN; distribution
choices, as well as link interactions, please refer to [11]. In lines possess higher resistance to reactance ratios, bringing
addition to the Davidson function method, tailored UE tight correlation among power flow variables. The traditional
models with arc capacity constraints can be found in lossless DCOPF which assumes constant bus voltage and
[19–22]. The core idea is to introduce a generalized path neglects reactive power is no longer satisfactory for mod-
travel cost, which introduces a penalty whenever xa [ ca . elling a PDN. In contrast to a transmission network which is
The cost can be extracted from KKT condition of (10) with usually meshed in topology, a PDN is often intentionally
an additional constraint xa ca . operated with radial topology. In such circumstance, the
Some earlier EV models have limited driving ranges. power flow equation can be recursively constructed via the
Researchers have developed distance-aware UE models branch flow model (BFM) developed in [34–37]:
which highlight the special route selection behaviors of X
Plij rijl Iijl þ PNj ¼ Pljk 8j ð18Þ
EVs. For example, length limit is incorporated in the model k2cð jÞ
proposed in [23], which is further generalized in [24] to X
include trip chains. Qlij xlij Iijl þ QNj ¼ Qljk 8j ð19Þ
In the near future, gasoline vehicles (GVs) will still k2cð jÞ
account for the majority part of traffic demand; GVs and 2
EVs will co-exist in the next a few decades. It is essential Uj ¼ Ui 2 Plij rijl þ Qlij xlij þ zlij Iijl 8l ð20Þ
to investigate UE models with both kinds of vehicles.
Along this line, a mixed UE model is developed in [25]; Ui Iijl ¼ ðPlij Þ2 þ ðQlij Þ2 8l ð21Þ
two algorithms are devised to solve the model efficiently. where cðjÞ is the set of child buses of bus j; Plij ; Qlij ; and Iijl
A more comprehensive UE model that jointly considers are active power, reactive power and squared current in
123
438 Wei WEI et al.
line l between its head bus i and tail bus j, respectively; 2.3 Interdependent network model
rijl ; xlij and zlij ¼ rijl þ jxlij are resistance, reactance, and
impedance of line l, respectively; PNj and QNj are nodal 2.3.1 Coupling points
power injections; Pljk and Qljk are power flows in down-
Charging facilities tie the two traditionally isolated
stream lines connected to bus j; Uj ¼ Vj2 is the squared bus systems. The typical connection is shown in Fig. 1. In the
voltage magnitude. Equations (18) and (19) depict nodal OPF problem, equality constraints correspond to the power
power balance; (20) is voltage drop equation; (21) defines flow model, while inequality constraints include variable
apparent power injected into the head bus of each line. lower and upper bounds as well as (22). The UE problem is
BFM can be derived from BIM, so they actually offer associated with (10). The interaction between two systems
the same power flow solutions in terms of bus voltage and is described as follows.
line flow. However, BFM (18)-(21) is more tractable than The traffic flow forms a UE in the transportation system.
BIM, because (18)-(20) are linear, and only (21) is non- Suppose a fast charging station rests on arc a where the
linear. Another advantage of BFM is that it allows an traffic flow is xa . If NE vehicles receive battery charging
efficient convex relaxation by replacing (21) with a rotated service during a period of Dt, the flow at the charging
second-order cone: station link is xea ¼ NE =Dt. Suppose the average energy
Ui Iijl ðPlij Þ2 þ ðQlij Þ2 ð22Þ demand of each EV is EB , then the total electric power
demand Pdc j at charging station a connected with bus j is:
So the relaxed model yields a second-order cone
NE EB
program (SOCP). For radial networks, this relaxation is Pdc
j ¼ ¼ / xea ¼ xea EB ð27Þ
shown to be exact under mild conditions [37–40], and is Dt
more tractable than the semidefinite program relaxation for In the existing literature, there are two representative
BIM-based OPF problems. For meshed networks, because methods to model electric vehicular flow xea .
bus angle variables are neglected, BFM and BIM may no
1) Proportional assumption (Ass-P for short). The sim-
longer be consistent.
plest way is to assume xea ¼ k0 xa , where k0 is a
If we neglect the lossy items in BFM, (18)-(20) again
constant, as in [43–46]. This assumption is more
define a proper power flow solution; (21) and squared
suitable for wireless charging road: charging facilities
current Iijl variable can be neglected, leading to the lin-
are paved under the surface of roadway segments, and
earized BFM: EVs are charged in motion [45, 47]. However, it may
X
Plij þ PNj ¼ Pljk 8j ð23Þ be difficult to determine the value of k in practice,
k2cð jÞ because it could change over time, and may be
X different for each road. It is also worth mentioning that
Qlij þ QNj ¼ Qljk 8j ð24Þ not all EVs have to recharge on travel. Only those in
k2cð jÞ
need of the service are included in xea .

Uj ¼ Ui 2 Plij rijl þ Qlij xlij 8l ð25Þ 2) Independent modelling of vehicle types (Ass-I for
short). Another way is to treat GV flow xga and EV
Since reactive power and bus voltage are taken into flow xea separately, and xa ¼ xga þ xea . This needs a
account, model (23)-(25) is still more competent than the particular route choice model for EVs, which can take
traditional DC power flow model. more factors such as electricity prices and queuing into
If we assume that all bus voltages are close to that at the
reference bus, ð Vi V0 Þ 2 0 holds, implying
Uj ¼ 2V0 Vj U0 . Substituting it into (25), we obtain: Distribution network
OPF
PjN, QjN Pikl 1, Qikl 1
Plij rijl þ Qlij xlij i Pijl , Qijl j min cost
Vj ¼ Vi 8l ð26Þ Pikl 2, Qikl 2 s.t. power flow
V0 Ui Uj variable bounds
Pikl 3, Qikl 3
Locational
Equations (25) and (26) are interchangeably used in the Charging
marginal
demand
existing literature. Linearized BFM is shown to perform xae price
well in distribution system related studies [40–43]. UE
xa xa xae Beckmann model
or
Transportation network Nesterov model
Fig. 1 Coupling between the two systems
123
account. This paradigm is used in [19]. If EV charging three phases are unbalanced, the equivalent network
demands are not the same, they can be categorized into topology is no longer radial, the three-phase distribution
several clusters, and each cluster shares the same OPF model in [50] and semidefinite programming relax-
parameter [48]. ation method can provide an OPF solution.
The charging load in (27) is added with other loads in
2.3.2 Coupled networks
the PDN. In non-deregulated case, the system power flow is
determined by BFM. The electricity price at charging sta-
As shown in Fig. 1, the traffic assignment problem and
tions may influence the route choices of EVs (this requires
OPF problem are independently solved; an equilibrium in
a special treatment in the UE model, as explained previ-
the coupled networks will emerge when no one has the
ously), if power system is allowed to announce different
incentive to change its strategy unilaterally.
prices for individual charging stations, then charging loads
Mathematically, such an equilibrium is usually descri-
will spontaneously respond to the price signal, offering
bed by a fixed-point problem: suppose the OPF solution is
additional flexibility to system operation. This dimension
y ðxÞ, where x corresponds to the UE pattern in the trans-
of flexibility could be significant because dispatchable
portation system; LMP can be determined from a mapping
resource in distribution system is often rare. In this context,
l ¼ LMPðy Þ; x ðlÞ is the UE pattern with a given LMP
it is more interesting and promising to envision deregulated
l . A fixed point is reached if (28) is satisfied, inspiring an
power markets in distribution system. In particular, two
iterative method to identify the network equilibrium [19],
pricing schemes were found in existing work.
described as follows.
1) Retail price (RTP). The distribution system can l ¼ LMPðy ðx ðl ÞÞÞ ð28Þ
directly set electricity prices at individual fast charging
stations, as in [46]. To ensure fairness, the price maker Step 1: Initiate an arbitrary LMP l.
must obey certain pricing policy, such as limits on Step 2: Solve the traffic assignment problem (referring
maximum price in each period and daily average price. to the UE models in this paper).
2) Locational marginal price (LMP), which is similar to Step 3: Update charging demands; solve the market
the wholesale power market. Given demand bidding at clearing problem and calculated LMP.
all buses, the distribution market is cleared according Step 4: If the change of LMP in two consecutive itera-
to an OPF problem, as illustrated in Fig. 1, where the tions is small enough, terminate; otherwise, return to Step
equality collects active power balance condition in 2.
(18) or (23), and inequality encapsulates remaining We choose to initiate LMP because it does not impact
constraints in BFM as well as variable lower and upper feasibility. If the equilibrium exists and is stable, this
bounds. LMP reflects the marginal cost of supplying iterative algorithm is likely to converge. It relies on convex
per unit incremental load at particular charging station, optimization and is thus preferred in many applications.
and has been adopted in [19, 44]. Another viable option is to simultaneously solve the
optimality conditions of traffic assignment problem and
Mature LMP markets rely on DCOPF model, which is OPF problem. For the former one, the KKT condition is
appropriate for transmission power grid. As distribution like (9); the convex objective can be approximated by a
systems possess high resistance to reactance ratios, the piece-wise linear function, and (9) can be linearized by the
market clearing model should take into account network integer programming method [51]. For the latter one, we
losses. Fortunately, BFM-based OPF formulation admits recommend linearizing the market clearing problem by
exact SOCP relaxation. LMPs can be extracted from dual performing polyhedral approximation [49] on constraint
variable l, which is supported in MOSEK solver. Alter- (22), because KKT condition of an LP is a linear com-
natively, LMP can be obtained by solving the dual problem plementarity problem, and can be linearized by the same
(an SOCP) or the KKT condition with the given primal technique in [51]. The advantage of KKT system method
optimal solution (a set of linear equations). stems from its compactness: the network equilibrium is
An LP market clearing problem can be obtained by expressed by closed-form constraints.
applying polyhedral approximation technique [49] on sec- The above modelling framework has been reported in
ond-order cone constraint (22). It provides a more conve- [19]. It also discusses the existence and stability of the
nient computational tool for distribution market studies, network equilibrium and provided some intuitive expla-
because the KKT or primal-dual optimality condition can nations. To our knowledge, reference [44] may be the first
be easily linearized, and the market clearing problem can work that sheds light on this equilibrium problem in the
be embedded in another optimization problem that mimic coupled transportation and power systems, in which
the strategic decision of market participants. If the loads in DCOPF is used to clear the power market.
123
440 Wei WEI et al.
3 Applications UE model captures traffic flow distributions. Similar

framework is also adopted in [54]; however, queuing time
Since EVs have shown their advantages in low envi- and tie line allocation are taken into account. In afore-
ronmental impact and great market potential, the research mentioned work, the UE appears in the form of KKT
on integration of EVs in transportation system and power condition, and the nonlinear BPR function (3) makes the
system has been a hot topic for more than a decade. Among complementarity condition hard to tackle. Reference [55]
transportation research community, traffic assignment borrows the Nesterov UE model, and suggests a mixed-
models with EVs received most attentions. However, the integer linear program (MILP) model for coordinate
charging station and its impact on the power system is expansion planning, including roadway segments in the
usually simplified or neglected. Among power system transportation system, generators and distribution lines in
research community, there have been extensive publica- the PDN, as well as charging facilities which couple the
tions on various EV charging schemes and their system two systems.
impact. Nevertheless, the driving patterns of EVs are Researchers also propose alternative transportation sys-
assumed to be given in either a deterministic or stochastic tem models for charging station planning problem. A
manner. This section aims to bridge the gap: we give a capacitated flow-refueling location model is reported in
comprehensive review on interdisciplinary research which [56] which mainly concerns limited driving ranges of EVs,
considers the reaction from the traffic side (although some instead of congestion, and the fast charging station plan-
neglects the power grid), covering the main topics of sys- ning problem yields an MILP. Notably, the PDN model is
tem planning and operation. The models of respective the linearized power flow model in [57]. In [58], the
system and the mathematical formulation of the whole capacitated flow-refueling location model is improved by
problem will be compared. considering heterogeneous EV driving ranges and
stochastic arriving rates, and the power system model is
3.1 System expansion planning also replaced with the BFM which is more suitable for
distribution systems. As a result, the planning problem
The task of system expansion planning aims to deter- comes down to a mixed-integer SOCP (MISOCP). Photo-
mine the construction strategies on one or more system voltaic (PV) generation is growing fast in distribution
components including charging stations/lanes, roadway systems. However, the output of PV power plant is volatile.
segments, distribution lines, generators, etc. Charging A two-stage stochastic MISOCP model is offered in [59] to
facility planning has received most attentions up to now, determine the optimal size of charging stations and PV
amid the research on new UE models. generators. In practice, an entirely accurate probability
Reference [30] proposes a tour-based UE model with distribution of uncertain factor is difficult to acquire. To
several classes of EVs; the charging station planning vanquish this difficulty, PV generation is represented via
problem is then formulated as a bi-level program (BLP), ambiguous distribution in [60]. Traffic demands on a
and solved by a genetic-algorithm-based procedure. Ref- highway network are obtained via Monte Carlo simulation.
erence [31] studies optimal deployment of charging lanes A data-driven robust stochastic program is proposed in [60]
under an energy-aware UE model. The planning problem is for siting and sizing stand-alone renewable powered
cast as a mathematical program with complementarity charging stations.
constraints (MPCC), which is solved by an active-set al- All literatures mentioned above are summarized and
gorithm. Above work does not consider the PDN. compared in Table 1, where the symbol ‘‘H’’ (or ‘‘9’’)
Reference [44] presents the first study on interdependent signifies that the listed item is (or is not) considered;
transportation and power infrastructure from a network ACOPF is the abbreviation of alternating current optimal
equilibrium perspective. The UE model captures destina- power flow.
tion choice via a multinomial logit function. The network
equilibrium consists of KKT conditions of UE and DCOPF, 3.2 Coordinated operation
and the charging station allocation problem is posed as an
MPCC. An active-set algorithm combining with NLP sol- With a high penetration of EVs, the interactions between
vers is used to solve the problem. A multi-objective transportation system and power distribution system
charging station planning model is developed in [52], impose challenges on real-time operation, as the decision
which maximizes the vehicle flows captured by charging space grows higher. On the other hand, the synergetic
stations and compromises network losses and bus voltage potential also creates opportunity to ameliorate operating
deviations. Collaborative planning of distribution system conditions of both systems through proper price signals,
and charging stations is studied in [53], where Beckmann calling for careful coordination between two systems. A
basic model is the network equilibrium elaborated in
123
Table 1 Comparison of models in system expansion planning

Reference Network model Vehicular flow Components for expansion Final model
Transportation network PDN GV EV
[30] Tour-based UE 9 H H Charging stations BLP

[31] UE (SoC, recharge rate) 9 9 H Charging lanes MPCC
[44] UE (destination choice) DCOPF H H Charging stations MPCC
[52] Shortest path ACOPF (BIM) 9 H Charging stations Multi-objective NLP
[53] UE (Beckmann model) ACOPF (BIM) Ass-P ? Ass-P ? Charging facilities, feeders, Multi-objective NLP
(27) (27) substations
[54] UE (Beckmann model) ACOPF (BIM) Ass-P ? Ass-P ? Charging stations, tie lines NLP
(27) (27)
[55] UE (Nesterov model) Linearized BFM Ass-P ? Ass-P ? Charging facilities, local MILP
(23)-(25) (27) (27) generators, distribution lines,
roadway segments
[56] Capacitated flow- Linearized power 9 H Charging stations MILP
refueling flow model in
location model [57]
[58] Modified capacitated BFM (18)-(21) 9 H Charging stations MISOCP
flow-refueling
location model
[59] Modified capacitated BFM (18)-(21) 9 H Charging stations, PV power plants Two-stage
flow-refueling stochastic
location model MISOCP
[60] Monte Carlo simulation 9 9 H Sites of charging station, sizes of Data-driven
PV panels, size of energy storage robust
unit stochastic
program
Section 2.3.2. This equilibrium state is often described by identified by iteratively solving an SOCP master problem
complementarity constraints, as in many publications and a max-min SOCP subproblem. By dualizing the inner
reviewed in Section 3.1. In operation related studies, SOCP, a bi-convex program arises, which is further
locations and capacities of charging facilities are fixed; transformed into an MISOCP. Reference [63] applies the
electricity price is an effective mean to influence traffic interdependence model in the unit commitment problem,
flows. As charging stations mainly demand active power, and proposes to use EV fleets as a battery storage unit.
generators in the distribution systems also appear to be Above work assumes that there is a central agency that
important dispatchable resources. can anticipate the network flow in both systems. However,
Along this line, coordinated system operation with if the two systems are independently operated, or the pri-
road/electricity pricing is studied in [45, 46, 61]. Beckmann vacy of system data should be preserved, new decision-
UE model or its modification is adopted for traffic flow, making methods are needed. Reference [47] considers the
and congestion toll is imposed on each roadway segment, coordinated system operation as a decentralized optimiza-
while DCOPF/BFM is employed for power flow. The tion (DO) problem, and develops an algorithm based on
system operation models in [46, 61] render MPCCs which alternative direction method of multiplier. A bilevel game
can be solved by a manifold optimization algorithm. model is proposed in [64] to coordinate power and traffic
Because the solution is often suboptimal, derivative-free flows, shave peak loads and smooth load ramps in both
search algorithms are suggested in [45, 46] for cases with systems. The upper level designs the network equilibrium
only a few pricing variables. MISOCP can be solved by (a welfare equilibrium in the sense of [65]) between the
off-the-shelf solvers such as CPLEX, GUROBI and two systems; in the lower level, a quasi-dynamic UE model
MOSEK. To consider traffic demand uncertainty, a two- is incorporated to simulate time-varying demands and
stage robust optimization (TSRO) model is proposed in spontaneous delay of departure time. GVs and EVs are
[62]. By solving traffic assignment problem (10) in extreme separately treated in [19, 48]; charging stations are repre-
scenarios, traffic demand variation is mapped into power sented by virtual nodes and arcs in an extended trans-
demand uncertainty. The optimal solution of TSRO is portation network. Network equilibrium is introduced in
123
442 Wei WEI et al.
Table 2 Comparison of models in system operation

Reference Network model Vehicular flow Pricing scheme Final model
TN PDN GV EV TN PDN
[61] UE (Beckmann model) DCOPF Ass-P ? Ass-P ? Road toll LMP MPCC
(27) (27)
[45] UE (Beckmann model) BFM (18)-(21) Ass-P ? Ass-P ? Road toll 9 MISOCP
(27) (27)
[46] UE (destination choice) BFM (18)-(21) Ass-P ? Ass-P ? Road toll Retail price MPCC
(27) (27)
[62] UE (Beckmann model) BFM (18)-(21) Ass-P ? Ass-P ? 9 9 TSRO
(27) (27)
[63] UE (Beckmann model) DCOPF Ass-P ? Ass-P ? 9 9 MILP
(27) (27)
[47] Modified UE DCOPF Ass-P ? Ass-P ? 9 LMP DO
(27) (27)
[64] UE (Beckmann model) Unbalanced BFM Ass-P ? Ass-P ? 9 Welfare equilibrium [65] Bilevel game
(27) (27) equivalent to LMP
[48] SO traffic assignment DCOPF H H Road toll LMP LP
[19] UE (GVs ? EVs) BFM (18)-(21) H H 9 LMP Fixed-point problem
[66] Joint transportation and 9 9 H 9 Price function Normalized Nash
charging scheduling equilibrium
[67] Dynamic routing for DCOPF 9 H 9 LMP Fixed-point problem
individual vehicles
[19]. Collaborative pricing schemes are suggested in [48] power flow is solved by fast-decoupled Newton-Raphson
to help each network achieve SO while preserving the method.
privacy of operational data. The impact of wireless charging roads on electricity
Some literature studies inter-system collaboration market is investigated in [70]. EV arrival rate at each street
through more tailored models. For example, a joint trans- is assumed to follow Poisson process, and thus trans-
portation and charging scheduling model is portrayed in portation network model is not considered. Power market is
[66], and is formulated as a normalized Nash equilibrium cleared according to DCOPF. A retail pricing mechanism is
problem. The distribution system is omitted; the LMP is recommended for load-serving entities to reduce LMP
approximated by a power function in nodal demand. In variation and improve the social welfare.
[67], a routing optimization algorithm is presented to Vulnerability of interdependent transportation and
navigate individual vehicles; vehicle routing and distribu- power distribution systems is studied in [71]. Traffic flow is
tion system will reach a fixed point linked by LMP. All determined from Beckmann UE model, where road
literatures mentioned above are summarized and compared capacity ca also acts as a variable. An MILP model is
in Table 2. proposed to identify the most vulnerable arcs, whose
degradation have the worst impact on total vehicle travel
3.3 Other applications time. The distribution system is modeled via linearized
BFM (23), (24), (26). It is found that road capacity
Besides the aforementioned interdependence model, degradation may cause over load in some charging stations,
researchers have proposed alternative ways to forecast which may trigger security issues in the power grid. In
loads of fast charging stations. In [68], EV arrival rate is [72], traffic demand is described by hourly O-D matrices,
derived from a fluid dynamic model, and then the charging and the impact of road block on the coupled networks is
load is predicted by queueing theory. The spatial and investigated via simulation. Resilience of the coupled
temporal distribution of EV charging demands are esti- systems in extreme events is studied in [73]. A dynamic UE
mated by Monte Carlo simulation method in [69]. Geo- model is used to capture time-varying traffic flows after a
graphical movements of EVs are indicated by twenty-four disruption occurs, taking into account traffic signal control,
O-D matrices with a time granularity of one hour. BIM of which is shown to be effective in traffic flow management
[74, 75]. Distribution system is modeled by linearized BFM
123
(23), (24), (26). A min-max-min problem is set forth to Continuous-time DTA models have been proposed in
enhance system resilience by strategically hardening lines form of optimal control [89], differential game [90, 91],
and placing DGs. differential variational inequality [92], differential com-
plementarity problem [93] and partial-differential com-
plementarity problem [94]. Meanwhile, continuous-time
4 Future research directions power system dispatch methods are also developed in [95]
for unit commitment and in [96] for power market. Com-
With the increasing penetration of EVs, the system-wide bining dynamic models of transportation system and power
interdependence across transportation and power distribu- system, it is more convenient to investigate the transient
tion infrastructures will become more prominent. The topic behavior and stability of network flow evolution. However,
reviewed in this paper is a very young and active research the computational tractability may be a main issue for
field. Authors believe the following directions deserve practical usage.
more attentions in the future.
4.1.2 Ride-sharing UE
4.1 Traffic assignment models
In light of the fast development of information and
4.1.1 Dynamic traffic assignment communication technology, as well as the proliferation of
smart phones, ride-sharing has become a prevalent trend in
The UE model introduced in Section 2.1 assumes all recently years, and related services have witnessed rapid
trips are completed instantly. In fact, not all vehicles enter growth.
the transportation system and begin to travel at the same Beckmann UE model is revamped in [97] by combining
time. To simulate spatial and temporal distributions of a ride-sharing demand function with elastic traffic
traffic flows more precisely, dynamic traffic assignment demands. The model can be utilized to analyze ride-sharing
models should be used. Indeed, as the power system load price, driver’s willingness to share, as well as their inter-
and electricity price are time-varying, it is always more action with road congestion level. In [98], travelers are
appropriate to use dynamic traffic assignment in the study categorized into solo/ride-sharing drivers and passengers; a
of system interdependence. However, dynamic traffic mixed complementarity program is devised for the ride-
assignment theory is much more complicated than the sharing UE over an extended network. Path-flow based
static one. There is no universal model; existing ones ride-sharing UE is formulated as NCP in [99], and is
usually require more sophisticated mathematical tools. improved to a link-node based one in [100]. The latter one
There have been extensive publications in this area during drastically reduces problem size and computational burden,
the past decades. A review on some earlier dynamic traffic and also allows problem decomposition for large-scale
assignment (DTA) models can be found in [76], and an up- systems. An alternative method is suggested in [101] to
to-date survey can be found in [77]. This section only sheds consider ride-sharing behaviors in Beckmann UE model
light on some most representative ones that might be with travel mode choice. Waiting time that depends on the
combined with power system research. number of available drivers is taken into account.
Continuous-time traffic flow can be described by kine-
matic wave equation, or namely, the Lighthill-Whitham- 4.2 Future distribution systems
Richards (LWR) model [78, 79]. Cell transmission model
(CTM) proposed in [80, 81] is a discrete-time approxima- Due to the constantly increasing penetration of renew-
tion of the LWR model. It simulates traffic flow and den- able energy and distributed generation [102, 103], the
sity on a corridor (represented by a finite number of distribution system itself is undergoing rapid change.
segments) over a certain period. Discrete-time and con- Power flow could be reversed depending on renewable
tinuous-time dynamic user equilibrium (DUE) models power output; a fraction of demand such as EV charging
based on CTM is developed in [82, 83], respectively. becomes deferrable and more flexible; energy storage
Because system resilience entails time elapse, the discrete- devices are deployed to facilitate system operation; above
time DUE model is employed in [73] as complementarity changes pave the way to the so-called active distribution
constrains [84]. Several other discrete-time DTA models network [104, 105], and call for innovative ideas, tech-
can be found in [84–88]. Discrete-time DTA models are nologies and initiatives to support such a transition. To our
more likely to be used in the interdependence research, knowledge, energy storage sharing and its business model
because mainstream power system dispatch and power have already attracted much attention [105–108]. In a
market models give rise to discrete-time optimization broader sense, prosumers [109] and distribution power
problems. market [50, 110] are playing increasingly important roles in
123
444 Wei WEI et al.
the operation of distribution systems. In the energy internet 4.3.3 System resilience
era [111], we believe new business models will emerge,
such as sharing economy [112], multi-energy market Transportation system and power distribution system are
[113, 114], peer-to-peer exchange [114–117], and so on. In critical infrastructures in modern society. It is important to
such circumstances, EVs and charging stations have better study the resilience of such infrastructures under extreme
opportunities and stronger incentives to take part in future weather conditions and natural disasters, activating fast
distribution power markets. plans for system restoration and repairing, and emergency
evacuation. In the case of a major blackout, EVs can pro-
4.3 Prospective applications vide valued backup of electric energy. In addition, con-
tingency in one system can propagate to the other, such as
Provided with more advanced UE models, some inter- the event in the city of Shenzhen mentioned in Section 1.
esting directions are open for future research. Unlike power system blackout which rarely happens, traffic
accidents and traffic flow control take place more fre-
4.3.1 Demand response quently. It is also essential to study how traffic flow and
power flow are influenced after a contingency, so as to
Because controllability (or flexibility) of a power dis- identify vulnerable components that need upgrade or
tribution system is not as high as a transmission system, it expansion.
is very promising to ameliorate system operating condition
by leveraging demand response potentials of private elec- 4.3.4 Robust system operation
tric cars and electric buses. Please bear in mind that private
cars, buses and metro trains have different driving patterns: Finally, uncertainty in such coupled infrastructures is
buses cannot alter their routes, while metro train will not ubiquitous: distributed renewable generation, power
suffer congestions. Electric buses and metro trains are good demand, traffic demand, various contingencies, and so on.
providers of demand response, because they are regularly Robust system operation still deserves research attention,
scheduled by a transportation authority. In the future, pri- especially in a decentralized manner.
vate EVs and fast charging stations are likely to join in
demand response programs in the presence of appropriate
economic incentives. This entails accurate time-dependent 5 Conclusion
DTA models with multiple travel mode choices to capture
the reaction from the transportation system in response to The penetration of EVs in modern smart cities is
the price signal from the power grid, as well as thorough growing fast, and the interaction between transportation
investigation of market stability using dynamic interde- and power distribution systems are becoming more evident.
pendence models. This paper provides an up-to-date survey on modelling and
applications of such interdependent infrastructures, and
4.3.2 Shared EVs envisions prospective research directions in the near future.
The power flow model of distribution system is relatively
Shared EVs which resemble shared bikes are developing mature, while the user equilibrium model of the trans-
rapidly in recent years. Unlike fast charging stations which portation system varies significantly under different
draw electricity from the power system, unused EVs in a assumptions, especially in the presence of EVs. System-
parking lot can act as an energy storage unit that can either level coordination also yields very different problems with/
consume or provide electricity. However, the usage of without a central authority. We believe the distributed
shared EVs depends on the service price and congestion mode and network equilibrium are more practical in the
condition in the transportation system. An interesting topic future. In conclusion, this direction is still in its infancy
for the EV-sharing platform is to determine real-time ser- stage. With further proliferation of EVs and the advent of
vice prices which encourage commuters to avoid traffic- sharing economy, the benefits gained from traffic-power
peak hours, as well as charging-discharging schedules coordination for enhancing system reliability, efficiency
which bring the platform additional benefits via providing and resilience will be prominent.
auxiliary service to the power grid or participating demand
response programs. The strategic behaviors of EV-sharing Acknowledgements Wei WEI acknowledges the support by the
Young Elite Scientists Program of CSEE (No. JLB-2018-95) and the
platforms need careful investigation using the time-varying National Natural Science Foundation of China (No. 51621065, No.
traffic flow patterns in the transportation network, calling U1766203). João P. S. CATALÃO acknowledges the support by
for thorough study on modelling dynamic interdependence. FEDER funds through COMPETE 2020 and by Portuguese funds
through FCT, under SAICT-PAC/0004/2015 (POCI-01-0145-
123
FEDER-016434), 02/SAICT/2017 (POCI-01-0145-FEDER-029803) Nesterov & de Palma models. In: Proceedings of Swiss trans-
and UID/EEA/50014/2019 (POCI-01-0145-FEDER-006961). port research conference, Ascona, Switzerland, 12–14
September 2007, pp 1–23
Open Access This article is distributed under the terms of the [19] Wei W, Wu L, Wang J et al (2018) Network equilibrium of
Creative Commons Attribution 4.0 International License (http:// coupled transportation and power distribution systems. IEEE
creativecommons.org/licenses/by/4.0/), which permits unrestricted Trans Smart Grid 9(6):6764–6779
use, distribution, and reproduction in any medium, provided you give [20] Larsson T, Patriksson M (1995) An augmented Lagrangean
appropriate credit to the original author(s) and the source, provide a dual algorithm for link capacity side constrained traffic
link to the Creative Commons license, and indicate if changes were assignment problems. Transp Res B Methodol 29(6):433–455
made. [21] Larsson T, Patriksson M (1999) Side constrained traffic equi-
librium models—analysis, computation and applications.
Transp Res B Methodol 33(4):233–264
References [22] Nie Y, Zhang HM, Lee DH (2004) Models and algorithms for
the traffic assignment problem with link capacity constraints.
[1] Abdmouleh Z, Alammari RAM, Gastli A (2015) Review of Transp Res B Methodol 38(4):285–312
policies encouraging renewable energy integration & best [23] Jiang N, Xie C, Waller T (2012) Path-constrained traffic
practices. Renew Sustain Energy Rev 45:249–262 assignment: model and algorithm. Transp Res Rec J Transp Res
[2] Tan Z, Chen K, Ju L et al (2016) Issues and solutions of Board 2283:25–33
China’s generation resource utilization based on sustainable [24] Wang T, Xie C, Xie J et al (2016) Pathconstrained traffic
development. J Mod Power Syst Clean Energy 4(2):147–160 assignment: a trip chain analysis under range anxiety. Transp
[3] Zhou Y, Wang M, Hao H et al (2015) Plug-in electric vehicle Res C Emerg 68:447–461
market penetration and incentives: a global review. Mitig [25] Jiang N, Xie C (2014) Computing and analyzing mixed equi-
Adapt Strateg Glob Change 20:777–795 librium network flows with gasoline and electric vehicles.
[4] Grunditz E, Thiringer T (2016) Performance analysis of current Comput Aided Civ Infrastruct Eng 29(8):626–641
BEVs based on a comprehensive review of specifications. IEEE [26] Jiang N, Xie C, Duthie JC et al (2014) A network equilibrium
Trans Transp Electr 2(3):270–289 analysis on destination, route and parking choices with mixed
[5] Yong J, Ramachandaramurthy V, Tan K et al (2015) A review gasoline and electric vehicular flows. Eur J Transp Logist
on the state-of-theart technologies of electric vehicle, its 3(1):55–92
impacts and prospects. Renew Sustain Energy Rev 49:365–385 [27] He F, Yin Y, Lawphongpanich S (2014) Network equilibrium
[6] Xiang Y, Liu Z, Liu J et al (2018) Integrated traffic-power models with battery electric vehicles. Transp Res B Methodol
simulation framework for electric vehicle charging stations 67:306–319
based on cellular automaton. J Mod Power Syst Clean Energy [28] Xu M, Meng Q, Liu K (2017) Network user equilibrium
6(4):816–820 problems for the mixed battery electric vehicles and gasoline
[7] Liu P, Yu J (2018) Identification of charging behavior char- vehicles subject to battery swapping stations and road grade
acteristic for large-scale heterogeneous electric vehicle fleet. constraints. Transp Res B Methodol 99:138–166
J Mod Power Syst Clean Energy 6(3):567–581 [29] Liu Z, Song Z (2018) Network user equilibrium of battery
[8] Kwasinski A, Bae S (2012) Spatial and temporal model of electric vehicles considering flowdependent electricity con-
electric vehicle charging demand. IEEE Trans Smart Grid sumption. Transp Res C Emerg 95:516–544
3(1):394–403 [30] He F, Yin YF, Zhou J (2015) Deploying public charging sta-
[9] Shenzhen News Website (2018) Charging stations out of ser- tions for electric vehicles on urban road networks. Transp Res
vice caused large-scale outage of taxis in Shenzhen. http:// C Emerg 60:227–240
news.sznews.com/content/2018-05/22/content_19164366.htm. [31] Chen Z, He F, Yin Y (2016) Optimal deployment of charging
Accessed 15 November 2018 lanes in transportation networks. Transp Res B Methodol
[10] Galus M, Waraich R, Noembrini F et al (2012) Integrating 91:344–365
power systems, transport systems and vehicle technology for [32] Chen Z, He F, Zhang L et al (2016) Optimal deployment of
electric mobility impact assessment and efficient control. IEEE autonomous vehicle lanes with endogenous market penetration.
Trans Smart Grid 3(2):934–949 Transp Res C Emerg 72:143–156
[11] Sheffi Y (1985) Urban transportation network: equilibrium [33] Li M, Lin X, He F et al (2016) Optimal locations and travel
analysis with mathematical programming methods. Pretice- time display for variable message signs. Transp Res C Emerg
Hall, Upper Saddle River 69:418–435
[12] Bureau of Public Roads (1964) Traffic assignment manual. U.S. [34] Zhu J (2015) Optimization of power system operation. Wiley,
Department of Commerce, Washington Hoboken
[13] Davidson KB (1966) A flow-travel time relationship for use in [35] Baran ME, Wu FF (1989) Network reconfiguration in distri-
transportation planning. https://trid.trb.org/view/1209266. bution systems for loss reduction and load balancing. IEEE
Accessed 1 November 2018 Trans Power Deliv 4(2):1401–1407
[14] Wardrop JG (1952) Some theoretical aspects of road traffic [36] Baran ME, Wu FF (1989) Optimal capacitor placement on
research. Proc Inst Civ Eng 1(3):325–362 radial distribution systems. IEEE Trans Power Deliv
[15] Beckmann M, McGuire C, Winsten C (1956) Studies in the 4(1):725–734
economics of transportation. Yale University Press, New Haven [37] Baran ME, Wu FF (1989) Optimal sizing of capacitors placed
[16] Nesterov Y (2000) Stable traffic equilibria: properties and on a radial distribution system. IEEE Trans Power Deliv
applications. Optim Eng 1(1):29–50 4(1):735–743
[17] Nesterov Y, de Palma A (2003) Stationary dynamic solutions in [38] Farivar M, Low SH (2013) Branch ow model: relaxations and
congested transportation networks: summary and perspectives. convexification. IEEE Trans Power Syst 28(3):2554–2572
Netw Spat Econ 3(3):371–395 [39] Gan L, Li N, Topcu U et al (2015) Exact convex relaxation of
[18] Chudak FA, Eleuterio VDS, Nesterov Y (2007) Static traffic optimal power flow in radial networks. IEEE Trans Autom
assignment problem: a comparison between Beckmann and Control 60(1):72–87
123
446 Wei WEI et al.
[40] Huang S, Wu Q, Wang J et al (2017) A sufficient condition on driven robust optimization approach. IEEE Trans Transp Electr
convex relaxation of AC optimal power flow in distribution 4(3):817–830
networks. IEEE Trans Power Syst 32(2):1359–1368 [61] He F, Yin Y, Zhou J (2013) Integrated pricing of roads and
[41] Yeh HG, Gayme DF, Low SH (2012) Adaptive VAR control electricity enabled by wireless power transfer. Transp Res C
for distribution circuits with photovoltaic generators. IEEE Emerg 34:1–15
Trans Power Syst 27(3):1656–1663 [62] Wei W, Mei S, Wu L et al (2017) Robust operation of distri-
[42] Wang Z, Chen B, Wang J et al (2015) Coordinated energy bution networks coupled with urban transportation infrastruc-
management of networked microgrids in distribution systems. tures. IEEE Trans Power Syst 32(3):2118–2130
IEEE Trans Smart Grid 6(1):45–53 [63] Sun Y, Chen Z, Li Z et al (2018) EV charging schedule in
[43] Wang Z, Chen B, Wang J et al (2015) Stochastic DG placement coupled constrained networks of transportation and power
for conservation voltage reduction based on multiple replica- system. IEEE Trans Smart Grid. https://doi.org/10.1109/tsg.
tions procedure. IEEE Trans Power Deliv 30(3):1039–1047 2018.2864258
[44] He F, Wu D, Yin Y et al (2013) Optimal deployment of public [64] Jiang H, Zhang Y, Chen Y et al (2018) Power-traffic coordi-
charging stations for plug-in hybrid electric vehicles. Transp nated operation for bi-peak shaving and bi-ramp smoothing-a
Res B Methodol 47:87–101 hierarchical data-driven approach. Appl Energy 229:756–766
[45] Wei W, Mei S, Wu L et al (2017) Optimal traffic-power flow in [65] Li N, Chen L, Low S (2011) Optimal demand response based
urban electrified transportation networks. IEEE Trans Smart on utility maximization in power networks. In: Proceedings of
Grid 8(1):84–95 IEEE PES general meeting, Detroit, USA, 24–29 July 2011,
[46] He F, Yin Y, Wang J et al (2016) Sustainability SI: optimal pp 1–8
prices of electricity at public charging stations for plug-in [66] Zhu M, Liu X, Wang X (2018) Joint transportation and
electric vehicles. Netw Spat Econ 16(1):131–154 charging scheduling in public vehicle systems—a game theo-
[47] Manshadi SD, Khodayar ME, Abdelghany A et al (2018) retic approach. IEEE Trans Intell Transp Syst 19(8):2407–2419
Wireless charging of electric vehicles in electricity and trans- [67] Amini M, Karabasoglu O (2018) Optimal operation of inter-
portation networks. IEEE Trans Smart Grid 9(5):4503–4512 dependent power systems and electrified transportation net-
[48] Alizadeh M, Wai H, Chowdhury M et al (2017) Optimal works. Energies 11(1):196
pricing to manage electric vehicles in coupled power and [68] Bae S, Kwasinski A (2012) Spatial and temporal model of
transportation networks. IEEE Trans Control Netw Syst electric vehicle charging demand. IEEE Trans Smart Grid
4(4):863–875 3(1):394–403
[49] Ben-Tal A, Nemirovski A (2001) On polyhedral approxima- [69] Mu Y, Wu J, Jenkins N et al (2014) A spatial-temporal model
tions of the second-order cone. Math Oper Res 26(2):193–205 for grid impact analysis of plug-in electric vehicles. Appl
[50] Liu Y, Li J, Wu L (2018) Distribution system restructuring: Energy 114:456–465
distribution LMP via unbalanced ACOPF. IEEE Trans Smart [70] Ou C, Liang H, Zhang W (2015) Investigating wireless
Grid 9(5):4038–4048 charging and mobility of electric vehicles on electricity market.
[51] Fortuny-Amat J, Mccarl B (1981) A representation and eco- IEEE Trans Ind Electron 62(5):3123–3133
nomic interpretation of a two-level programming problem. [71] Wei W, Wang J, Wu L (2016) Quantifying the impact of road
J Oper Res Soc 32(9):783–792 capacity loss on urban electrified transportation networks: an
[52] Wang G, Xu Z, Wen F et al (2013) Traffic-constrained multi- optimization based approach. Int J Transp Sci Technol
objective planning of electric vehicle charging stations. IEEE 5(4):268–288
Trans Power Deliv 28(4):2363–2372 [72] Yang X, Li Y, Cai Y et al (2018) Impact of roadblock on peak-
[53] Yao W, Zhao J, Wen F et al (2014) A multiobjective collab- load of coupled traffic and energy transportation networks.
orative planning strategy for integrated power distribution and Energies 11(7):1–12
electric vehicle charging systems. IEEE Trans Power Syst [73] Wang X, Shahidehpour M, Jiang C et al (2018) Resilience
29(4):1811–1821 enhancement strategies for power distribution network coupled
[54] Xiang Y, Liu J, Li R et al (2016) Economic planning of electric with urban transportation system. IEEE Trans Smart Grid.
vehicle charging stations considering traffic constraints and https://doi.org/10.1109/tsg.2018.2848970
load profile templates. Appl Energy 178:647–659 [74] Li Z, Shahidehpour M, Bahramirad S et al (2017) Optimizing
[55] Wei W, Wu L, Wang J et al (2017) Expansion planning of traffic signal settings in smart cities. IEEE Trans Smart Grid
urban electrified transportation networks: a mixed-integer 8(5):2382–2393
convex programming approach. IEEE Trans Transp Electr [75] Li Z, Al Hassan R, Shahidehpour M et al (2019) A hierarchical
3(1):210–224 framework for intelligent traffic management in smart cities.
[56] Zhang H, Moura S, Hu Z et al (2018) PEV fast-charging station IEEE Trans Smart Grid 10(1):691–701
siting and sizing on coupled transportation and power netowrks. [76] Peeta S, Ziliaskopoulos AK (2001) Foundations of dynamic
IEEE Trans Smart Grid 9(4):2595–2605 traffic assignment: the past, the present and the future. Netw
[57] Haffner S, Pereira LFA, Pereira LA et al (2008) Multistage Spat Econ 1(3–4):233–265
model for distribution expansion planning with distributed [77] Wang Y, Szeto WY, Han K et al (2018) Dynamic traffic
generation part I: problem formulation. IEEE Trans Power assignment: a review of the methodological advances for
Deliv 23(2):915–923 environmentally sustainable road transportation applications.
[58] Zhang H, Moura S, Hu Z et al (2018) A second order cone Transp Res B Methodol 111:370–394
programming model for planning PEV fast-charging stations. [78] Lighthill M, Whitham G (1955) On kinematic waves, II: a
IEEE Trans Power Syst 33(3):2763–2777 theory of traffic flow on long crowded roads. Proc R Soc Lond
[59] Zhang H, Moura S, Hu Z (2018) Joint planning of PEV fast- Ser A Math Phys Sci 229(1178):317–345
charging network and distributed PV generation using the [79] Richards PI (1956) Shockwaves on the highway. Oper Res
accelerated generalized benders decomposition. IEEE Trans 4(1):42–51
Transp Electr 4(3):789–803 [80] Daganzo CF (1994) The cell transmission model: a dynamic
[60] Xie R, Wei W, Khodayar ME et al (2018) Planning fully representation of highway traffic consistent with the hydrody-
renewable powered charging stations on highways: a data- namic theory. Transp Res B Methodol 28(4):269–287
123
[81] Daganzo CF (1995) The cell transmission model, part II: net- comparison of numerical and mathematical modelling methods.
work traffic. Transp Res B Methodol 29(2):79–93 Renew Sustain Energy Rev 67:531–573
[82] Ukkusuri SV, Han L, Doan K (2012) Dynamic user equilibrium [103] Georgilakis PS, Hatziargyriou ND (2013) Optimal distributed
with a path based cell transmission model for general traffic generation placement in power distribution networks: models,
networks. Transp Res B Methodol 46(10):1657–1684 methods, and future research. IEEE Trans Power Syst
[83] Friesz TL, Han K, Neto PA et al (2013) Dynamic user equi- 28(3):3420–3428
librium based on a hydrodynamic model. Transp Res B [104] Hu J, Cong H, Jiang C (2018) Coordinated scheduling model of
Methodol 47(1):102–126 power system with active distribution networks based on multi-
[84] Han L, Ukkusuri S, Doan K (2011) Complementarity formu- agent system. J Mod Power Syst Clean Energy 6(3):521–531
lations for the cell transmission model based dynamic user [105] Liu J, Gao H, Ma Z et al (2015) Review and prospect of active
equilibrium with departure time choice, elastic demand and distribution system planning. J Mod Power Syst Clean Energy
user heterogeneity. Transp Res B Methodol 45(10):1749–1767 3(4):457–467
[85] Nakayama S, Connors R (2014) A quasi-dynamic assignment [106] Lombardi P, Schwabe F (2017) Sharing economy as a new
model that guarantees unique network equilibrium. Transp A business model for energy storage systems. Appl Energy
Transp Sci 10(7):669–692 188:485–496
[86] Zheng QP, Arulselvan A (2011) Discrete time dynamic traffic [107] Kalathil D, Wu C, Poolla K et al (2019) The sharing economy
assignment models and solution algorithm for managed lanes. for the electricity storage. IEEE Trans Smart Grid
J Glob Optim 51(1):47–68 10(1):556–567
[87] Nahapetyan A, Lawphongpanich S (2007) Discretetime [108] Liu J, Zhang N, Kang C et al (2017) Cloud energy storage for
dynamic traffic assignment models with periodic planning residential and small commercial consumers: a business case
horizon: system optimum. J Glob Optim 38(1):41–60 study. Appl Energy 188:226–236
[88] Waller ST, Fajardo D, Duell M et al (2013) Linear program- [109] Parag Y, Sovacool BK (2016) Electricity market design for the
ming formulation for strategic dynamic traffic assignment. prosumer era. Nat Energy 1(4):16032
Netw Spat Econ 13(4):427–443 [110] Bai L, Wang J, Wang C et al (2018) Distribution locational
[89] Friesz TL, Luque J, Tobin RL et al (1989) Dynamic network marginal pricing (DLMP) for congestion management and
traffic assignment considered as a continuous time optimal voltage support. IEEE Trans Power Syst 33(4):4061–4073
control problem. Oper Res 37(6):893–901 [111] Xue Y (2015) Energy internet or comprehensive energy net-
[90] Wie BW (1995) A differential game approach to the dynamic work? J Mod Power Syst Clean Energy 3(3):297–301
mixed behavior traffic network equilibrium problem. Eur J [112] Hamari J, Sjöklint M, Ukkonen A (2016) The sharing econ-
Oper Res 83(1):117–136 omy: why people participate in collaborative consumption.
[91] Friesz TL (2010) Dynamic optimization and differential games. J Assoc Inf Sci Technol 67(9):2047–2059
Springer, Boston [113] Chen Y, Wei W, Liu F et al (2018) Energy trading and market
[92] Friesz TL, Han K (2018) The mathematical foundations of equilibrium in integrated heat-power distribution systems.
dynamic user equilibrium. Transp Res B Methodol. https://doi. IEEE Trans Smart Grid. https://doi.org/10.1109/tsg.2018.
org/10.1016/j.trb.2018.08.015 2849227
[93] Liu HX, Ban XJ, Pang JS et al (2012) Modeling and solving [114] Wang C, Wei W, Wang J et al (2018) Equilibrium of inter-
continuous-time instantaneous dynamic user equilibria: a dif- dependent gas and electricity markets with marginal price
ferential complementarity systems approach. Transp Res B based bilateral energy trading. IEEE Trans Power Syst
Methodol 46(3):389–408 33(5):4854–4867
[94] Wang DZW, Du B (2016) Continuum modelling of spatial and [115] Liu N, Yu X, Wang C et al (2017) Energy-sharing model with
dynamic equilibrium in a travel corridor with heterogeneous price-based demand response for microgrids of peer-to-peer
commuters—a partial differential complementarity system prosumers. IEEE Trans Power Syst 32(5):3569–3583
approach. Transp Res B Methodol 85:1–18 [116] Morstyn T, Teytelboym A, McCulloch MD (2018) Bilateral
[95] Parvania M, Scaglione A (2016) Unit commitment with con- contract networks for peer-to-peer energy trading. IEEE Trans
tinuous-time generation and ramping trajectory models. IEEE Smart Grid. https://doi.org/10.1109/tsg.2017.2786668
Trans Power Syst 31(4):3169–3178 [117] Morstyn T, Farrell N, Darby SJ et al (2018) Using peer-to-peer
[96] Parvania M, Khatami R (2017) Continuous-time marginal energy-trading platforms to incentivize prosumers to form
pricing of electricity. IEEE Trans Power Syst 32(3):1960–1969 federated power plants. Nat Energy 3(2):94
[97] Xu H, Ordónez F, Dessouky M (2015) A traffic assignment
model for a ridesharing transportation market. J Adv Transp
49(7):793–816 Wei WEI received the Bachelor and Ph.D. degrees both in electrical
[98] Xu H, Pang JS, Ordónez F et al (2015) Complementarity engineering from Tsinghua University, Beijing, China, in 2008 and
models for traffic equilibrium with ridesharing. Transp Res B 2013, respectively. He was a Postdoctoral Research Associate with
Methodol 81:161–182 Tsinghua University from 2013 to 2015. He was a Visiting Scholar
[99] Di X, Liu HX, Ban X et al (2017) Ridesharing user equilibrium with Cornell University, Ithaca, USA, in 2014, and with Harvard
and its implications for highoccupancy toll lane pricing. Transp University, Cambridge, USA, in 2015. He is currently an Associate
Res Rec J Transp Res Board 2667:39–50 Professor with Tsinghua University. His research interests include
[100] Di X, Ma R, Liu HX et al (2018) A link-node reformulation of applied optimization and energy system economics.
ridesharing user equilibrium with network design. Transp Res
B Methodol 112:230–255 Danman WU received the Bachelor degree in electrical engineering
[101] Bahat O, Bekhor S (2016) Incorporating ridesharing in the from China Agricultural University, Beijing, China, in 2018. She is
static traffic assignment model. Netw Spat Econ currently pursuing Ph.D. degree in Tsinghua University, Beijing,
16(4):1125–1149 China. Her research interests include operation of interdependent
[102] Theo WL, Lim JS, Ho WS et al (2017) Review of distributed energy systems.
generation (DG) system planning and optimization techniques:
123
448 Wei WEI et al.
papers, 5 books, 35 book chapters, and 14 technical reports, with an

Qiuwei WU obtained the Ph.D. degree in power system engineering
h-index of 45, an i10-index of 190, and over 8300 citations (according
from Nanyang Technological University, Singapore, in 2009. He is an
to Google Scholar), having supervised more than 70 post-docs, Ph.D.
Associate Professor at Center for Electric Power and Energy,
and M.Sc. students. He is the Editor of the books entitled ‘‘Electric
Department of Electrical Engineering, Technical University of
Power Systems: Advanced Forecasting Techniques and Optimal
Denmark (DTU). He was a visiting scholar at Department of
Generation Scheduling’’ and ‘‘Smart and Sustainable Power Systems:
Industrial Engineering & Operations Research (IEOR), University
Operations, Planning and Economics of Insular Electricity Grids’’
of California, Berkeley, from February 2012 to May 2012 funded by
(Boca Raton, FL, USA: CRC Press, 2012 and 2015, respectively). His
Danish Agency for Science, Technology and Innovation (DASTI),
research interests include power system operations and planning,
Denmark. He was a visiting scholar at the Harvard China Project,
hydro and thermal scheduling, wind and price forecasting, distributed
School of Engineering and Applied Sciences, Harvard University
renewable generation, demand response and smart grids. Prof.
from November 2017 to October 2018 funded by the Otto Mønsted
CATALÃO is an Editor of the IEEE Transactions on Smart Grid,
Fond. His research interests include operation and control of power
an Editor of the IEEE Transactions on Power Systems, an Associate
systems with high penetration of renewables.
Editor of the IEEE Transactions on Industrial Informatics, and a
Subject Editor of the IET Renewable Power Generation. From 2011
Miadreza SHAFIE-KHAH received the M.Sc. and Ph.D. degrees in till 2018 (seven years) he was an Editor of the IEEE Transactions on
electrical engineering from Tarbiat Modares University, Tehran, Iran, Sustainable Energy and an Associate Editor of the IET Renewable
in 2008 and 2012, respectively. He received his first postdoctor from Power Generation. He was the Guest Editor-in-Chief for the Special
the University of Beira Interior (UBI), Covilha, Portugal in 2015. He Section on ‘‘Real-Time Demand Response’’ of the IEEE Transactions
received his second postdoctor from the University of Salerno, on Smart Grid, published in December 2012, the Guest Editor-in-
Salerno, Italy in 2016. He is currently an Assistant Professor at the Chief for the Special Section on ‘‘Reserve and Flexibility for
University of Vaasa, Vaasa, Finland. His research interests include Handling Variability and Uncertainty of Renewable Generation’’ of
power market simulation, market power monitoring, power system the IEEE Transactions on Sustainable Energy, published in April
optimization, demand response, electric vehicles, price forecasting 2016, and the Corresponding Guest Editor for the Special Section on
and smart grids. ‘‘Industrial and Commercial Demand Response’’ of the IEEE
Transactions on Industrial Informatics, published in November
João P. S. CATALÃO received the M.Sc. degree from the Instituto 2018. Since March 2018, he is the Lead Guest Editor for the Special
Superior Técnico (IST), Lisbon, Portugal, in 2003, and the Ph.D. Issue on ‘‘Demand Side Management and Market Design for
degree and Habilitation for Full Professor (‘‘Agregação’’) from the Renewable Energy Support and Integration’’ of the IET Renewable
University of Beira Interior (UBI), Covilha, Portugal, in 2007 and Power Generation. He was the recipient of the 2011 Scientific Merit
2013, respectively. Currently, he is a Professor at the Faculty of Award UBI-FE/Santander Universities, the 2012 Scientific Award
Engineering of the University of Porto (FEUP), Porto, Portugal, and UTL/Santander Totta, the 2016 and 2017 FEUP Diplomas of
Research Coordinator at INESC TEC. He was also appointed as Scientific Recognition, the 2017 Best INESC-ID Researcher Award,
Visiting Professor by North China Electric Power University, Beijing, and the 2018 Scientific Award ULisboa/Santander Universities, in
China. He was the Primary Coordinator of the EU-funded FP7 project addition to an Honorable Mention in the 2017 Scientific Award
SiNGULAR (‘‘Smart and Sustainable Insular Electricity Grids Under ULisboa/Santander Universities. Moreover, he has won 4 Best Paper
Large-Scale Renewable Integration’’), a 5.2-million-euro project Awards at IEEE Conferences.
involving 11 industry partners. He has authored or coauthored more
than 680 publications, including 260 journal papers (more than 75
IEEE Transactions/Journal papers), 370 conference proceedings
123
Transport and Telecommunication Vol. 19, no. 1, 2018
Transport and Telecommunication, 2018, volume 19, no. 1, 45–58

Transport and Telecommunication Institute, Lomonosova 1, Riga, LV-1019, Latvia
DOI 10.2478/ttj-2018-0005
METHODOLOGY OF SELECTION OF HEAVY AND OVERSIZED

FREIGHT TRANSPORTATION SYSTEM
Artūras Petraška1, Kristina Čižiūnienė2, Olegas Prentkovskis3, Aldona Jarašūnienė4
1, 2, 4
Department of Logistics and Transport Management,
Vilnius Gediminas Technical University
Plytinės g. 27, LT-10105, Vilnius, Lithuania
1
arturas.petraska@vgtu.lt; 2 kristina.ciziuniene@vgtu.lt; 4 aldona.jarasuniene@vgtu.lt
3
Department of Mobile Machinery and Railway Transport,
Vilnius Gediminas Technical University,
Plytinės g. 27, LT-10105, Vilnius, Lithuania
3
olegas.prentkovskis@vgtu.lt
Dynamically changing economic conditions influence the growing demand of various freights transportation. Inevitably,
the demand for heavy and oversized freight transportation is increasing, which is quite problematic. In many cases, it is very
difficult to standardize the technology of heavy and oversized freight transportation. Decisions are necessary, which would allow to
deliver heavy or oversized freight to the destination place, allocating fewer funds as possible for infrastructure improvemen ts,
choosing the most appropriate mode of transport for such freight transportation or using advantages of multimodal (combined)
transport. In this article there will be selected methodology of heavy and oversized freight transportation system, carrying h eavy and
oversized freight by road transport. There will also be made an approval of this methodology, based on a hypothetical rout e within
Lithuanian territory.
Keywords: heavy and oversized freight, transportation system, road, selection methodology
1. Introduction
The growing demand of various freights transportation is influenced by increasing countries

economic development, dynamically changing economic conditions. Building large economic facilities
such as a large thermal, nuclear or wind power plants, oil refineries and chemical plants will inevitably
provoke a demand of heavy and oversized freight (hereafter – HOF) transportation, which is quite
problematic. Given the fact that the HOF can be transported by more than one mode of transport, it can be
said that the proper freight transportation route selection is a complex, multifunctional task (Xu and Hoel,
2001; Wang et al., 2011; Petraška and Palšaitis, 2011, 2012; Bazaras et al., 2013; Petraška et al., 2017;
Newnam et al., 2017; Palšaitis and Petraška, 2012; Pronello et al., 2017; Sánchez-Díaz, 2017; Andrés and
Padilla, 2015; Li et al., 2014). Road transport is one of the most complicated modes of transport, and
demanding most technological solutions (Sivilevičius, 2011; Podvezko and Sivilevičius, 2013; El-
Rashidy and Grant-Muller, 2016; Carrara and Longden, 2017).
Purpose of this article is to create the methodology to select a system of HOF transportation by
road transport.
There are following tasks to achieve the purpose:
 to perform the analysis of theoretical aspects concerning transport routing, when carrying
HOF;
 to create HOF transportation methodology, after evaluating the routing principles;
 to perform hypothetical inspection of created methodology applicability using an example of
Lithuania.
The paper comprises five sections, the first one being the introduction, followed by Section 2 that
is dedicated to relevant literature review about theoretical aspects of transportation routing, in case of
transportation of heavy or oversized freight by road transport. Section 3 presents data collection about
evaluation of HOF transport routing principles and new methodology of universal system for HOF
routing creation. Results and Discussion about application example of universal system for HOF routing
by road transport are given in Section 4. Finally, the 5 Section presents the Conclusion.
45
2. Theoretical Formulation of the Problem
Durham and Faghri (2002) in their analyses investigate the capabilities using Geographic
Information System (GIS) to estimate potential transportation routes of HOF. However, these systems are
not always allow to assess what is the turning radius of the oversized vehicle, which can be critical and
may limit the movement of the vehicle on a regular road network (Sarker et al., 2012) and other specific
elements. On the other hand, it is very important that oversized vehicle remains undeviating from the
planned transportation route, chosen by evaluating the vehicle’s dimensions together with the load,
weight and other characteristics, in order to ensure traffic safety and to prevent damage to the road
infrastructure (ECMT, 2006; Vaitkus et al., 2016; Kombe et al., 2017; Dell’Acqua et al., 2016; Ušpalytė-
Vitkūnienė and Laureshyn, 2017; Pauer, 2017).
Using GIS software aims to find the shortest route in the existing road network taking into account
the vehicle’s height and weight for each road segment/section (Datla et al., 2004). Thus, during the
analysis of the entire road network, based on oversized vehicle parameters, those roads, which do not
meet the vehicle parameters, are eliminated. During the analysis of the road network, it is appropriate to
assess the improvement possibilities of “bottlenecks” or restrictive elements of HOF transportation. In
this case GIS system would become inaccurate (Durham and Faghri 2002).
Many researchers (Hall et al., 2007; Sousa et al., 2016; Abulizi et al., 2016; Gardziejczyk and
Zabicki, 2017; Doumiati et al., 2014; Sivilevičius et al., 2017; Bartuška et al., 2016; Sivilevičius and
Šukevičius 2006) examine in their works a road plan, curves marking methods, longitudinal road profile,
bank, they provide a methodology for selecting the pavement structure, there are presented calculations of
the road plan and longitudinal profile elements, of the designed Annual Average Daily Traffic Intensity
(AADTId), of bank, of earthwork volumes. Given the fact that roads in most countries are divided into
categories, the designed road category is determined by the designed Annual Average Daily Traffic
Intensity. For example, in Lithuania, the determination of road category comprises the 20-year long-term
period.
HOF transportation conditions, compared with the standard freight transportation are different, but
the principles of the classification of road sections according to the traffic intensity are similar, so the
traffic intensity assessment is an important factor in the planning of the sections of HOF transportation
routes in different areas.
During the planning of HOF transportation route, it is necessary to evaluate the existing obstacles
on the ground surface, which may need to overtake during the transportation process (Petraška and
Palšaitis, 2012; Bazaras et al., 2013; Petraška et al. 2017; Palšaitis and Petraška, 2012; et al. 2017;
Benekos and Diamantidis 2017; Conca et al. 2016).
The road is designed taking into account technical and economic calculations, which are based on
road economic necessity and expedience expediency (Armstrong and Taylor, 2000; Park, 2000; Iwański
et al., 2016; Van Damme et al., 2016; Bickel et al., 2006). One of the most important documents,
according to which issues concerning road design and construction expediency, their sequence and
volume of works are dealt, it is the road development scheme. Preparing this scheme, there are following
actions: formation of region characteristic, network paths analysis, calculation of transported freight and
passengers dimension and traffic intensity on network roads, presentation of proposals to develop the road
network, description of natural conditions in the road construction area, determination of capital
investment efficiency, presentation of road technical and economic characteristics, presentation of
conclusions describing the road network compliance with current and prospective traffic, presentation of
recommendations for road network configuration development and for road category increase. Designing
the new road construction, economic research is carried out. One of the most important signs of economic
research in the area are focus points of freight and passenger, which determine the volume of freight
transported [42]. Therefore, an important factor is the assessment of necessary elements, which allows
properly selecting the transportation route.
3. Data Collection and Creation of Methodology
3.1. Evaluation of HOF Transport Routing Principles

Having the necessity to carry HOF, there are different conditions to define that must be met, that
the transport process would be successful. The conditions determining the appearance of that demand
may be twofold: it is a specific demand for transporting specific freight from point A to point B, or
forecast that it will appear a multiple demand to carry such freight within a certain territory.
46
The examination of the HOF transportation process starting from the selection of route, mode of
transport and transportation vehicle and ending with transportation action planning and implementation, it
can be seen that the individual transportation process parts and operations in different modes of transport
has obvious qualitative similarities, but can vary depending on the implementation cost or time (for
example, road sections, not meeting the requirements of HOF transportation, the demand to improve the
road surface or to deepen inland waterway watercourse, to increase road, railway or waterway turning
radius etc.). The financial costs and time spent will be needed to carry out all the necessary works. The
most rational transportation route is that, which will incur the minimum financial and time costs. It is
appropriate to classify systematically whole reasons determining the above factors, by spotlighting
repetitive route sections attributing the adequate criterion weights. After formalizing the entire HOF route
evaluation process, it appears a possibility to objectively compare different routes and different modes of
transport by the same scoring system and to found the most acceptable transportation option.
3.2. Creation of New Methodology

In order to ensure proper transportation of the HOF, it is proposed to evaluate the following
criteria.
1. Restrictions on the physical characteristics of the road – this group includes two criteria relate
to the following:
1.1 Road section pavement. Influence of road section pavement to freight transportation
speed – this criterion is designed to compare the route advantages for the different modes of
transport. In case of road route assessment, there are distinguished two alternatives of road
section criterion: asphalt or gravel. These options allow the assessment of freight movement
speed on the route. Option of the lack of road section allows evaluating the possibility of a
new road construction or searching for other alternatives.
Index “A” in formulas means sort of road transport. Influence of road section pavement to
transport speed:
S AD  xA1  k AD  xA2  k AZ , (1)
whereas: SAD – total duration of HOF transportation by road on route (in months); xA1 –
length of road with asphalt pavement (in km); xA2 – length of road with gravel pavement (in
km); kAD – coefficient – the inverse rate to the average freight transportation speed on
asphalt pavement; kAZ – coefficient – the inverse rate to the average freight transportation
speed on gravel pavement;
x 
kij   A1 
 x A2  , x A1  0, ..., n , x A 2  0, ..., n  ;
(2)
1
kij    
b  vaver , (3)
whereas: kij – a road pavement evaluation criterion numeric value in the analysed transport
category; b – number of hours per month (b = 720 hours, it is assumed that month consist
of 30 days); vaver – average speed of HOF transportation in the analysed transport category;
α – route longevity (in years); γ – HOF transportation frequency per year; n – the number of
road sections with different road pavements.
Sizes α and γ gain influence on the result with the values, if the HOF transportation route is
long-term. In case of a single transportation, these values practically have no influence to
result.
1.2 The physical quality of the road pavement at the assessment moment. Physical road
pavement quality at the assessment moment – this criterion allows evaluating the road
pavement for freight transportation and the demand for one or the other action to ensure the
required quality of the road. For the road, the total value of the financial costs, ensuring the
required quality of the road pavement for HOF transportation is assessed according to
formula (4). The value of financial costs is assessed by Relative Financial Units (hereafter –
RFU), which concrete value depends on the economic-financial environment of examined
territory:
47
FAQ  xAF1  k AF1  xAF 2  k AF 2  xAF 3  k AF 3 , (4)
whereas: FAQ – total value of the financial costs required for the route improvement for
HOF transportation by road transport (in RFU); xAF1 – number of road sections of 100 m,
requiring minor improvements (in pcs.); xAF2 – number of road sections of 100 m, requiring
major works (in pcs.); xAF3 – number of road sections of 100 m, requiring new road
construction (in pcs.); kAF1 – coefficient – dimension, meaning the average financial costs
of small improvements to make on the route (in RFU); kAF2 – coefficient – dimension,
meaning the average financial costs of road improvement major works (in RFU); kAF3 –
coefficient – dimension, meaning the average financial costs of arranging a new road
section (in RFU).
kFij – dimension, which means the financial costs incurred in preparing route for HOF
transportation, when it is necessary to remove small obstacles. The coefficient value
depends on rates of route improvement operations and work standards existing on a specific
area.
 c , c  0, ..., n ,

k Fij   z , z  0, ..., n ,
 g , g  0, ..., n , (5)

whereas: c – size assessing financial costs of road structures destruction and rebuild (in
RFU); z – size assessing financial losses resulting from time spent destructing or rebuilding
the road structures, and the costs for the time spent in the necessary legal procedures (in
RFU); g – financial cost value of necessary to carry out additional works (in RFU); n –
number of critical points, where improvements must be carried out (in pcs.).
During calculation of real value of quality criterion of physical road pavement, it is used
indicator d of possible financial loss. This is value estimating possible financial loss per
unit of time, related to the loss of time in accomplishing the necessary procedures (for
example, designing, planning, receiving of permits etc.). This indicator is measured in
money terms. The rate of this indicator depends on the policy of banks, interest rates and
labour force rates in the analysed area.
Ec  Pb
d  Uw , (6)
12
whereas: d – value of possible financial costs (losses) (in RFU); Ec – annual amount of
credit for implementation of HOF transportation project (in RFU); Pb – bank annual interest
rate (in %); Ud – monthly wage of staff involved in the project implementation process (in
RFU).
Obstacles due to road section geometry – to describe this criterion, the best example is the
freight corridor term that defines a freight transportation height, width and curvature
parameters. Further there are examined aspects of this criterion for each of mentioned
characteristics.
2. Road turning radius of curvature. In both road and inland waterway routes assessment cases,
the criterion has three alternatives: the curvature radius corresponds to the conditions of freight
transportation; there are necessary minor improvements of route – which means minor works
of straightening road radius, minor widening of road embankment etc.; major works would
mean a substantial reorganization of the road section.
 Small radius road turnings. The total value financial costs (FAS) in route of HOF
transportation by road transport, needed to increase the radius of curvature, is defined as
follows:
FAS  x AS1  k AS1  x AS 2  k AS 2 , (7)
whereas: xAS1 – number of small improvements on the road route, in order to ensure the
required turning radius (in pcs.); xAS2 – number of sections of the road route, in which major
works must be carried out to ensure the required turning radius (in pcs.); kAS1 – value
corresponding to the average financial costs necessary for small improvements of road
48
turning radius on a route (in RFU); kAS2 – value corresponding to the average financial costs
necessary for major improvements of road turning radius on a route (in RFU).
The weight of criterion evaluating the radius improvement of the road turning curvature
is equivalent to construction of 100 m of new road. The demand to change road curvature
radius occurs depending on the technical capabilities of the vehicle to turn within a
certain radius. These works are attributed to the major works category.
Evaluation of minor works ensuring proper road curvature radius, road corridor width or
quality:
 Destruction of road structures (minor works) includes destruction and rebuild of
kerbs, road signs and hedges.
 Road processing in order to reduce the potential negative impact on the road
pavement. An example could be the processing of road pavement using dolomite
chippings.
 Grading – temporary pavement improvement, for temporary route.
K grading  kgr  tgr  d  , (8)
whereas: Kgrading – numeric value of criterion “road grading”; kgr – financial expenses
of 100 m road grading (in RFU); tgr – road grading time (in months); d – value of
possible financial costs (losses) (in RFU).
 Road graveling – temporary road pavement improvement.
K graveling  k grv  t grv  d  , (9)
whereas: Kgraveling – numeric value of road graveling coefficient; kgrv – financial
expenses of 100 m road graveling (in RFU); tgrv – time required to perform graveling
works on route (in months); d – value of possible financial costs (losses) (in RFU).
3. Transportation corridor on road section is too narrow. Freight transportation corridor on
road section is too narrow – assessing road route, criterion consists of four alternatives: the
corridor width corresponds to the freight transportation conditions; there are necessary minor
routing improvements – which means minor works of road widening, a minor road
embankment widening etc.; major works would mean essential reorganization of the road
section; a version that the problem cannot be solved rationally, means that it is necessary to
select another transportation alternative.
FAKS  x AKS1  k AKS1  x AKS 2  k AKS 2  xAKS 3  k AKS 3 , (10)
whereas: FAKS – the total value of the financial costs required for HOF carried by road, corridor
width increase (in RFU); xAKS1 – number of minor improvements on the road route, ensuring
the required transportation corridor width (in pcs.); xAKS2 – number of road sections on the
route, where it is necessary to carry out major works, ensuring the required transportation
corridor width (in pcs.); xAKS3 – number of impassable sections of road route due to too low
width of transportation corridor (in pcs.); kAKS1 – value, meaning the average financial costs for
road corridor widening minor improvements to make on the route (in RFU); kAKS2 – value,
meaning the average financial costs for road corridor widening major works to make on the
route (in RFU); kAKS3 – value, equal to the infinity. It means that when critical area (such as
curve, bridge, throughputs, etc.) appears on the road route, which cannot be removed,
bypassed, rebuilt or otherwise avoided, in such a case the freight transportation capability on
this route is not analysed.
4. Barriers due to bridges/dams along the route. Insufficient bridge carrying capacity – this
criterion has five senses: the bridge carrying capacity meets the conditions of freight
transportation (in this case the criterion weighting value is equal to 0); there is required a metal
ramp or bridge reinforcement; it is necessary to install a throughput; there is required a new
bridge construction; it is impossible to rationally solve the problem – it is necessary to use
other mode of transport.
FAT  x AT 1  k AT 1  x AT 2  k AT 2  x AT 3  k AT 3  x AT 4  k AT 4 , (11)
whereas: FAT – total value of the financial costs for HOF transportation via bridges or
throughputs along the route to ensure the conditions (in RFU); xAT1 – number of places along
49
the road route, where it is appropriate to install a metal ramp ensuring safe HOF movement
through the bridges or throughputs along the route (in pcs.); xAT2 – number of new throughputs
along the road route (in pcs.); xAT3 – number of bridges necessary to newly build along the road
route (in pcs.); xAT4 – number of impassable sections along the road route (in pcs.); kAT1 – value,
meaning the average financial costs of installing/dismantling the metal ramp (in RFU); kAT2 –
size, meaning the average financial costs of installing throughput on the route (in RFU); kAT3 –
value, meaning the average financial costs of building a new bridge on the road route (in RFU);
kAT4 – value, equal to the infinity. It means that when critical area appears on the road route due
to insufficient carrying capacity of bridge or throughput, which cannot be removed, bypassed,
rebuilt or otherwise avoided, in such a case the freight transportation possibility on this route is
not analysed.
 Construction of throughput – value of this criterion weight is calculated considering
throughput design and construction time and value of bypass road construction:
K th  k k  t c  d   lth  ath  kth   t th  d  , (12)
whereas: Kth – numeric value of criterion “construction of throughput” (in RFU); kk –

construction cost of one kilometre of road (in RFU); tc – road/object construction time (in
months); d – value of possible financial costs (losses) (in RFU); lth – length of constructed
throughput (in meters); ath – width of constructed throughput (in meters); kth – cost of one
square meter throughput construction (in RFU); tth – throughput construction time (in
months).
 Metal ramp – this is a temporary tool of route application for HOFs transportation. This
is metal construction, covering too weak small bridges or throughputs. In that case, the
ramp helps to ensure the necessary carrying capacity of structure. Ramp’s mounting is
related to the necessity of bypass road construction. This measure is applied only for road
routes adaptation for the HOF transportation.
With the possibility of using metal ramp, numeric value of criterion can be calculated as
follows:
Kramp  tR  d   Knr , (13)
whereas: Kramp – numeric value of criterion “required metal ramp”, (in RFU); tR – ramp
installation time (in months); d – value of possible financial costs (losses) (in RFU); Knr –
construction of new local road (in RFU).
 Numeric value of Kramp on road transport depends on the ramp’s acquisition/delivery time
t1, construction/installation time t2ramp.
t R  t1  t2 ramp , (14)
whereas: tR – total ramp installation time (in months); t1 – ramp’s acquisition/delivery
time (in months); t2ramp – ramp assembly/disassembly time (in months).
The numerical value of the criterion "construction of a new bridge" is calculated
according to the formula:
KT  lT  aT  kT   tT  d  , (15)
5. Total length of the route – this criterion belongs to the evaluation category of consistently
changing factors. The route length is assessed in kilometres. Weights of this criterion of
different transport modes are different because of different freight transportation speed and
transportation price per kilometre.
F L   li cikm , (16)
whereas: FƩL – total HOF transportation cost in case of multimodal (combined) transportation
(in RFU); li – distance of HOF transportation by i-th mode of transport (in km); cikm – HOF
transportation price per 1 km by i-th mode of transport (in RFU).
The weight of criterion evaluating total route length is assessed regarding to HOF
transportation time and costs due to social factors.
Klength  tK  d , (17)
50
whereas: Klength – numeric value of total route length criterion (in RFU); tK – planned freight
transportation time (in months); d – value of possible financial costs (losses) (in RFU).
6. Demand for installation of transhipment sites. When transporting HOF by vehicles, the value
of total costs required for such transhipment site installation, is calculated as follows:
FAP  x AP1  k AP1 , (18)
whereas: FAP – installation costs of transhipment site on road route (in RFU); xAP1 – number of
HOF transhipment sites to another mode of transport required to be installed along the road
route (in pcs.); kAP1 – value, meaning the average financial costs required to HOF transhipment
site to another mode of transport along the road route (in RFU).
There are following criteria evaluating the demand of transshipment site and freight storage
area installation:
 Transhipment site – this may be a temporary freight unloading from one mode of
transport to another area or temporary pier.
Ktranship  ttr  d  ktr , (19)
whereas: Ktranship – numeric value of criterion “transhipment site demand” (in RFU); ttr –
transhipment site installation time (in months); ktr – transhipment site installation cost (in
RFU); d – value of possible financial costs (losses) (in RFU).
7. Demand of freight storage areas installation. The storage areas’ installation costs along the
road route are evaluated as follows:
FAY  x AY 1  k AY 1 , (20)
whereas: FAY – costs of storage area installation along the road route (in RFU); xAY1 – number
of the HOF storage areas, required to install along the road route (in pcs.); kAY1 – value,
meaning the average financial costs installing HOF storage area along the road route (in RFU).
8. Obstacles due to the legal (including environmental) requirements. Assessing the road route, the
criterion has following alternatives: the demand of passage through towns/villages – this criterion
is related to the oversized and heavy freight transportation rules through the major urban areas, or
to the needs concerning route improvement on those areas; demand of protected areas’ passage
means the evaluation of obstacles appearing a demand of route development in those areas.
FAJ  x AJ 1  k AJ 1  x AJ 2  k AJ 2  x AJ 3  k AJ 3  x AJ 4  k AJ 4 , (21)
whereas: FAJ – total costs incurred in removing the obstacles due to the legal restrictions on
road route (in RFU); xAJ1 – number of towns / settlements along the road route (in pcs.); xAJ2 –
number of protected areas along the route (in pcs.); xAJ3 – number of permits required to install
transhipment site along the road route (in pcs.); xAJ4 – number of permits required to install a
temporary HOF storage area along the road route (in pcs.); kAJ1 – coefficient to compensate
costs for social losses (in RFU); kAJ2 – factor between price obtaining a permit and time spent
(in RFU); kAJ3 – cost coefficient for obtaining a permit to install a transhipment site (in RFU);
kAJ4 – cost factor for obtaining a permit to install a temporary storage area (in RFU).
9. Time losses due to the traditional transport intensity on analysed section of the road – this
criterion has three common alternatives to all the modes of transport and thus has the an
objective to assess the social consequences of oversized and heavy freight transportation.
Impact assessment of traditional transport intensity – transport disruption on moderate or high
intensity sections can lead to social consequences, which raises the risk of unpredictable
disrupting the transportation process. Local authorities, due to the negative reaction of the
population to the HOF transportation may refuse to issue permits to carry this type of freight
through their territory, or to delay the permit delivery. In that case, the risk of increasing costs
expands significantly. Time losses due to traffic intensity along the road route are evaluated
according to the following formula:
S AI  x AI 1  k AI 1  x AI 2  k AI 2  x AI 3  k AI 3 , (22)
whereas: SAI – total costs of additional time during HOF transportation on the automobile road
(in months); xAI1 – low intensity road section length on the route (in km); xAI2 – moderate
intensity road section length on the route (in km); xAI3 – high intensity road section length on
the route (in km); kAI1 – size, evaluating the freight train speed on low intensity road section (in
51
1/ km/h); kAI2 – size, evaluating the freight train speed on moderate intensity road section (in 1/
km/h); kAI3 – size, evaluating the freight train speed on high intensity road section (in 1/ km/h).
Barriers to legal requirements are calculated:
Klegal  t L  d , (23)
whereas: Klegal – criterion "Barriers to the numerical value of legal requirements (SFV); tL –
time for receipt of permission to transport in the territory (in months); d – the amount of
potential financial losses (losses).
10. Seasonality influence to the possibility of freight transportation – this criterion evaluates the
seasonality of transport mode. Numeric value of factor depends on the time period during
which it is possible to transport HOF. The influence of this criterion depends on the geographic
area in which the freight transportation is planned and on the technical characteristics of the
specific road. The weight of this criterion is the square dependence on the number of months
per year when the freight is not transported. Numeric value of criterion is calculated by the
following formula:
KSE  kseas  xs2  d , (24)
whereas: KSE – numeric value of the “seasonality” criterion (in RFU); kseas – coefficient
evaluating seasonality of the HOF transportation (according to multiannual experience in the
Republic of Lithuania it is 0.5); xs – time period – number of months when HOF is not
transported (in pcs.).
11. Existing experience in the transportation of HOF – this criterion evaluates existing experience
for each mode of transport during HOF transportation. Regular HOF transportation during more
than one year is considered as big experience. Experience for at least one or more freight
transportation is considered as little experience. This criterion is also associated with the
transportation risk assessment. The risk increases with the absence of experience in HOF
transportation. It is very important to evaluate the existing experience of HOF carriers in this area.
 N , N  0; 3 ,

Eij  M , M  3; 20 ,
(25)
D, D  20, ..., n  ,
 v
whereas: Eij – factor evaluating the existing experience in the HOF transportation agent by i-th
mode of transport in j-th road section; N – there is no experience in HOF transportation; M –
there is a little experience in HOF transportation; D – there is a big experience in HOF
transportation; nv – number of cases concerning HOF transportation.
Transport risk appraisers typically are using the following classification of existing experience:
no experience – when the carrier has performed no more than 3 HOF transportations; little
experience – when the carrier has performed from 3 to 20 HOF transportations; big experience
– when the carrier has performed more than 20 HOF transportations.
Number of carried equal freights – when it is necessary to deliver not a single HOF, but the
greater lot as a group of freight related to a specific technology (oil refineries, wind power-
plants etc.), it is appropriate to assess the number of identical freight, because it facilitates the
organization of transportation.
4. Discussion
Hypothetical Lithuanian territory was chosen check the criteria system efficiency, where there is
traditional road network of different modes of transport, natural geographic barriers restricting the
possibilities of HOF transportation and specific economic-financial conditions (Figure 1). The task is to
evaluate the options of HOF transportation, on this hypothetical territory, by selecting the most suitable
route for HOF transportation when using the universal HOF routing and evaluation system, created and
adapted for road transport.
A hypothetical task is formulated for system verification: distance of HOF transportation from
point “A” to point “B” (between the points’ geographical coordinates) is equal to 400 kilometres. It is
possible to carry the HOF in three modes: by road, by rail and by inland water transport. Choosing the
third HOF transportation alternative, there is no possibility to bring freight directly to the final destination
52
(because there is no waterway to reach the final destination), so in order to use the inland waterway
alternative it would require applying a multimodal (combined) transport option. Maximum HOF weight is
150 tons. Maximum HOF height is 6 m, width – 5 m, length – 15 m. It is intended to carry HOF once on
this route and there is no presumption that this route will be used many times.
Figure 1. The scheme of hypothetic territory
On a hypothetical route, we will analyze the possibility of HOF transportation by road. Therefore,
it is important to analyze and identify the critical points on the route that could restrict the HOF
transportation (calculated based on road construction and rebuild prices in Lithuania).
After the analysis of freight transportation up to 150 tons on road transport, there is defined
following:
 length of the road with asphalt pavement is 305 km, length with a gravel pavement is 95 km.
the HOF is transported on asphalt pavement in average of 18 km/h, on gravel pavement –
10 km/h;
 It is necessary to gravel 800 metres of road in a country, to dismantle 25 road signs, to
disassemble 2 hedges on the route;
 it is necessary to widen 8 turnings on the route;
 it is necessary to remove the 3 trees providing the necessary freight corridor width (removal of
trees is attributed to the minor works category);
 there are 3 viaducts on the route, which do not meet the parameters of freight transportation
height, so it is necessary to bypass these viaducts;
 2 throughputs and one 14 meter long bridge must be rebuilt on the route;
 the route passes through 15 villages;
 traffic intensity is high on 100 km route (it requires 6 hours to overcome this route), is average
on 100 km route (it requires 6 hours to overcome this route), and is low on 200 km route (it
requires 20 hours to overcome this route);
 it is possible to carry the HOF on the predicted route can carry 6 months per year.
The criteria weighting calculations on the road transport route:
 Whereas the use of existing road network is in the territory, the distance between point “A” and
point “B” is not a straight line, so we get that HOF transportation will comprise 350 kilometres
on the asphalt pavement and 110 kilometres on gravel pavement. Thus, the weight of criterion
evaluating the road distance is calculated as follows (see Formulas 2, 3):
1 1
k AD    0.00008;
b  vaver 720  18
53
1 1
k AZ    0.0001.
b  vaver 720  10
Total HOF transportation time by road on the route will be as follows (see Formula 1):
S AD  xA1  k AD  xA2  k AZ  350  0.00008 110  0.0001 0.039 (in months)
Increasing the route length, HOF transportation costs increases accordingly; the risk of
negative transportation effects also increases. It is assumed that HOF transportation price for 1
t/km is 500 RFU/km. Therefore, estimating the influence of freight transportation distance, it is
attributed a value of 500 units of the criterion weighting.
 Before starting to calculate the weight of criterion “road graveling”, it is necessary to calculate
the coefficient value d of potential financial costs (losses) (see Formula 6). Assuming that the
annual amount of credit for HOF transportation project implementation is 30 million RFU, and
the bank’s annual interest rate is 7.5 %. In such a case, it is considered that salary is included in
the total amount of credit, but its changes do not affect the period of the credit use.
Ek  Pb 30000000 0.075
d   187500 (RFU).
12 12
Under specific conditions, after calculating the value of size d, the size SAD can be expressed in
relative financial units (RFU):
S AD  0.039  187500 7312.5 (RFU).
1 km of road graveling costs 205 thousand RFU, it takes in average one working week. Since it
is necessary to graveling 800 meters of road, the value of the criterion is 8. Criterion “road
graveling” is attributed to a minor works category and is equal to (see Formula 9):
K graveling  kgrv  t grv  d   20500 0.023 187500  24813(RFU).
The average cost of destruction and rebuild of 100 metres of kerbs, road hedges or single road
sign, taking all the costs of coordination with the necessary institutions, is 6000 RFU. This is
attributable to the category of minor improvement works on the route.
 Increasing the radius of turning curvature corresponds to the construction of 100 metres of new
road (cost of construction of 1 kilometre of asphalt local road is ~897000 RFU and designing
time is ~2 months). The weighting value of criterion is calculated as follows:
K nk  kk  ts  d   89700  (2  187500)  464700(RFU).
 Taking into account the throughput designing and construction time (see Formula (12))
together with bypass road, financial expenses reach ~3 million RFU (throughput installation
time is ~2 months, the time for receiving necessary permits is ~2 months).
Kth  kk  tc  d   lth  ath  kth   tth  d  
464700  6  12  25000  4  187500  3014700(RFU).
Numeric value of criterion weight, evaluating the construction of new 14 meter long bridge is
calculated as follows (see Formula 15):
KT  lT  aT  kT   tT  d   14  12  27000 10  187500 6411000(RFU).
Under current building regulations, the resistance limit for road bridges in hypothetical
territory is 250 tons. Therefore, considering the impact of freight weight on the final result, it is
assigned a weight value of new bridge construction criterion.
 To carry the HOF through the villages or towns, such transportation must be coordinated with
the local authorities. Coordination takes approximately 1 month. The weighting value of
criterion is calculated as follows:
Klegal  tL  d  1 187500 187500(RFU).
 Transporting HOF may encounter a regular disturbance in public life. The European Union has
approved the methodology for assessment the latency of one man, moving by car, in public
level [40]. This methodology is applied to motivate the investments in roads. According to this
methodology, the criterion weight is calculated in high, medium and low traffic intensity road
sections. Weighting of the criterion, considered under this methodology, is estimated by rate of
1 hour loss to the public (on low traffic intensity, losses comprise 606 RFU/h, on average
intensity – 4038 RFU/h, on high traffic intensity – 9075 RFU/h).
54
 Seasonality criterion weight is equal to the seasonality coefficient multiplied by the size d
concerning financial costs (losses). The value of criterion concerning seasonality influence is
calculated as follows (see Formula 24):
KSE  kseas  xs2  d  0.5  62  187500 3375000(RFU).
Results of criterion evaluation concerning road transport route are shown in Table 1.
Table 1. Results of road transport criterion evaluation

Road transport
No. Criterion Criterion Criterion
Result
value weight
Influence of road section pavement to the freight transportation speed,
7312.5
SAD
Physical quality of road pavement during assessment, FAQ
1.
Quality is appropriate × ×
Minor improvements × ×
Major works 3 6411000 19233000
Road turnings of small radius, FAS
Radius of curvature meets requirement × 0
2.
Minor improvements are necessary 35 24813 868455
Major works are necessary 8 46400 371200
Freight transportation corridor on the road section is too narrow, FAKS
Corridor width is appropriate × ×
Minor improvements are necessary 3 24813 74439
Major works are necessary × ×
Rational solution of problem is impossible × ×
3.
Freight transportation corridor on the road section is too low, FAKZ
Corridor height is appropriate × ×
Minor improvements are necessary × ×
Major works are necessary × ×
Rational solution of problem is impossible × ×
Barriers due to bridges/dams on the route. Insufficient bridge carrying
capacity, FAT
Bridge carrying capacity meets freight
× ×
transportation conditions
Metal ramp is necessary × ×
Throughput should be installed 2 3014700 6029400
4.
Construction of a new bridge/quay is necessary:
Construction of bridge up to 42 m length × ×
Construction of bridge up to 14 m length 1 6411000 6411000
Rational solution of problem is impossible
Maximum weight of carried freight, k sv
Freight up to 100 t ×
5. Freight from 100 t up to 250 t 1 6411000 6411000
Freight from 250 t up to 550 t × ×
Freight over 550 t × ×
6. Total length of route, FƩL 460 500 230000
Demand for installation of transhipment sites, FAP
7. No need for installation of transhipment site × ×
Number of transhipment sites necessary to install × ×
Demand for installation of freight storage areas (temporary), FAY
8. No need for installation of storage areas × ×
Number of freight storage areas necessary to install × ×
Obstacles due to the legal (including environmental) requirements, FAJ
Demand for passage through towns/ villages (number of villages along the
9. 15 187500 2812500
route, distance through villages)
demand for passage of protected areas × ×
Traditional transport intensity on the analysed route, SAI
Low 6 606 3636
10.
Moderate 6 4038 4044
High 20 9075 181500
Seasonality influence to the possibility of freight transportation (in
11. 62 93750 3375000
months per year), KSE
Total: 46012486.5
Note: × – in this case criterion practically has no influence.
55
5. Conclusions
In scientific literature is a lack of methodology concerning the performance of calculations and

choice of elements, what must be assessed, during HOF transportation. In most cases, scientists consider
only individual items or groups: for example, influence of HOF transportation to the road pavement;
various structures and buildings along the transportation section etc.
In this article was formulated a set of 11 criteria with individual sub criteria, that defines HOF
transportation processes, evaluates parameters of the route, the type of transportation and transportation
vehicles. This allows to assess in a common system all the HOF transportation processes applying the
comparative method.
HOF routes and processes evaluation criteria can be grouped into two groups, in one group all
criteria have the time dimension, and in another – the monetary dimension. This allows the formation of
the evaluation system ensuring an objective assessment of the HOF transportation processes by
comparing different types of transportation, route segments, technologies of transportation and freight
transhipment.
References
1. Abulizi, N., Kawamura, A., Tomiyama, K., Fujita, S. (2016) Measuring and evaluating of road
roughness conditions with a compact road profiler and ArcGIS. Journal of Traffic and
Transportation Engineering, 3(5), 398–411. doi:10.1016/j.jtte.2016.09.004
2. Andrés, L., Padilla, E. (2015) Energy intensity in road freight transport of heavy goods vehicles in
Spain. Energy Policy, 85, 309–321. doi:10.1016/j.enpol.2015.06.018
3. Armstrong, M., Taylor, J. Regional Economics and Policy. Wiley-Blackwell; 2000. 448 p.
4. Bartuška, L., Biba, V., Jeřábek, K. (2016) Verification of methodical procedure for determining the
traffic volumes using short-term traffic surveys. Procedia Engineering, 161, 275–281.
doi:10.1016/j.proeng.2016.08.553
5. Bazaras, D., Batarlienė, B., Palšaitis, R., Petraška, A. (2013) Optimal road route selection criteria
system for oversize goods transportation. The Baltic Journal of Road and Bridge Engineering, 8(1),
19–24. doi:10.3846/bjrbe.2013.03
6. Benekos, I., Diamantidis, D. (2017) On risk assessment and risk acceptance of dangerous goods
transportation through road tunnels in Greece. Safety Science, 91, 1–10.
doi:10.1016/j.ssci.2016.07.013
7. Bickel, P., Friedrich, R., Burgess, A., Fagiani, P., Hunt, A., De Jong, G., Laird, J., Lieb, C.,
Lindberg, G., Mackie, P., Navrud, S., Odgaard, T., Ricci, A., Shires, J., Tavasszy, L. (2006)
HEATCO: Developing Harmonised European Approaches for Transport Costing and Project
Assessment. Contract No. FP6-2002-SSP-1/502481. IER, Germany, from http://heatco.ier.uni-
stuttgart.de/HEATCO_D5.pdf
8. Bula, G.A., Prodhon, C., Gonzalez, F.A., Afsar, H.M.;Velasco, N. (2017) Variable neighborhood
search to solve the vehicle routing problem for hazardous materials transportation. Journal of
Hazardous Materials, 324(Part B), 472–480. doi:10.1016/j.jhazmat.2016.11.015
9. Carrara, S., Longden, T. (2017) Freight futures: The potential impact of road freight on climate
policy. Transportation Research Part D: Transport and Environment, 55, 359–372.
doi:10.1016/j.trd.2016.10.007
10. Conca, A., Ridella, C., Sapori, E. (2016) A risk assessment for road transportation of dangerous
goods: A routing solution. Transportation Research Procedia, 14, 2890–2899.
doi:10.1016/j.trpro.2016.05.407
11. Datla, S.K., Moorthy, R.S., Krishna Rao, K.V. (2004) A GIS for routing of oversized and hazardous
material carrying vehicles. Map Asia Conference 2004, Beijing, China, 26–29 August 2004, 1–14.
12. Dell’Acqua, G., De Luca, M., Prato, C.G., Prentkovskis, O., Junevičius, R. (2016) The impact of
vehicle movement on exploitation parameters of roads and runways: a short review of the special
issue. Transport, 31(2), 127–132. doi:10.3846/16484142.2016.1201912
13. Doumiati, M., Erhart, S., Martinez, J., Sename, O., Dugard, L. (2014) Adaptive control scheme for
road profile estimation: application to vehicle dynamics. IFAC Proceedings Volumes, 47(3), 8445–
8450.
14. Durham, G., Faghri, A. (2002) GIS Application for Oversize Vehicle Routing through a Virtual
Road and Bridge Network. University of Delaware, National Science Foundation, Research
Experiences for Undergraduates in Bridge Engineering, USA. 29 p.
56
15. ECMT. (2006) Improving Transport Accessibility for All: Guide to Good Practice. European
Conference of Ministers of Transport (ECMT). OECD publications service. 161 p., from
https://www.itf-oecd.org/sites/default/files/docs/06tphguide.pdf
16. El-Rashidy, R.A., Grant-Muller, S. (2016) The evaluation of redundancy for road traffic networks.
Transport, 31(4), 427–439. doi:10.3846/16484142.2016.1255913
17. Gardziejczyk, W., Zabicki, P. (2017) Normalization and variant assessment methods in selection of
road alignment variants – case study. Journal of Civil Engineering and Management, 23(4), 510–
523. doi:10.3846/13923730.2016.1210223
18. Hall, J.W., Sullivan, E.C., Reilly, W.R., Homburger W.S. (2007) Fundamentals of Traffic
Engineering. University of California. 464 p.
19. Iwański, M., Buczyński, P., Mazurek, G. (2016) Optimization of the road binder used in the base
layer in the road construction. Construction and Building Materials, 125, 1044–1054.
doi:10.1016/j.conbuildmat.2016.08.112
20. Kombe, T., Ele, P., Offole, F., Miasse, H. O. (2017) Modelling an interactive road signs system,
using Petri nets. Transport and Telecommunication Journal, 18(1), 34–44. doi:10.1515/ttj-2017-
0004
21. Li, X.-H., Han, Y.-H., Tan, Y.-L. (2014) Analysis of the effect of heavy duty freight cars loaded
overweight goods on the track structure. Procedia – Social and Behavioral Sciences, 138, 45–53.
doi:10.1016/j.sbspro.2014.07.180
22. Newnam, S., Goode, N., Salmon, P., Stevenson, M. (2017) Reforming the road freight transportation
system using systems thinking: An investigation of Coronial inquests in Australia. Accident Analysis
& Prevention, 101, 28–36. doi:10.1016/j.aap.2017.01.016
23. Palšaitis, R., Petraška, A. (2012) Heavyweight and oversized cargo transportation risk management.
Transport and Telecommunication, 13(1), 51–56. doi:10.2478/v10244-012-0005-9
24. Park, S.W. (2004) Modelling of deformational characteristics in unbound granular geomaterials.
KSCE Journal of Civil Engineering, 8(3), 281–285. doi:10.1007/BF02836010
25. Pauer, G. (2017) Development potentials and strategic objectives of intelligent transport systems
improving road. Transport and Telecommunication Journal, 18(1), 15–24. doi:10.1515/ttj-2017-
0002
26. Petraška, A., Čižiūnienė, K., Jarašūnienė, A., Maruschak, P., Prentkovskis, O. (2017) Algorithm for
the assessment of heavyweight and oversize cargo transportation routes. Journal of Business
Economics and Management, 18(6), 1098–1114. doi:10.3846/16111699.2017.1334229
27. Petraška, A., Palšaitis, R. (2012) Evaluation criteria and a route selection system for transportating
oversize and heavyweight cargoes. Transport, 27(3), 327–334.
doi.org:10.3846/16484142.2012.721133
28. Podvezko V., Sivilevičius, H. (2013) The use of AHP and rank correlation methods for determining
the significance of the interaction between the elements of a transport system having a strong
influence on traffic safety. Transport, 28(4), 389–403. doi:10.3846/16484142.2013.866980
29. Pronello, C., Camusso, C., Valentina, R. (2017) Last mile freight distribution and transport
operators’ needs: which targets and challenges? Transportation Research Procedia, 25, 888–899.
doi:10.1016/j.trpro.2017.05.464
30. Sánchez-Díaz, I. (2017) Modeling urban freight generation: A study of commercial establishments’
freight needs. Transportation Research Part A: Policy and Practice, 102, 3–17.
doi:10.1016/j.tra.2016.06.035
31. Sarker, B.R., Baylot, E.A., Green, J.G., Biswas, P. (2012) Convoy movement: consideration of
turning geometrics for selected vehicle shapes. Journal of Transportation Engineering, 138(5).
doi:10.1061/(ASCE)TE.1943-5436.0000359
32. Sivilevičius, H. (2011) Modelling the interaction of transport system elements. Transport, 26(1),
20–34. doi:10.3846/16484142.2011.560366
33. Sivilevičius, H., Šukevičius, Š. (2007) Dynamics of vehicle loads on the asphalt pavement of
european roads which cross Lithuania. The Baltic Journal of Road and Bridge Engineering, 2(4),
147–154.
34. Sivilevičius, H., Vislavičius, K., Bražiūnas, J. (2017) Technological and economic design of asphalt
mixture composition based on optimization methods. Journal of Civil Engineering and
Management, 23(4), 627–648. doi:10.3846/13923730.2016.1210223
35. Sousa, H., Costa, B.J.A., Henriques, A.A., Bento, J., Figueiras, J.A. (2016) Assessment of traffic
load events and structural effects on road bridges based on strain measurements. Journal of Civil
Engineering and Management, 22(4), 457–469. doi:10.3846/13923730.2014.897991
57
36. Ušpalytė-Vitkūnienė, R., Laureshyn, A. (2017) Perspectives for surrogate safety studies in East-
European countries. The Baltic Journal of Road and Bridge Engineering, 12(3), 161–166.
doi:10.3846/bjrbe.2017.19
37. Vaitkus, A.,Strumskys, M., Jasiūnienė, V., Jateikienė, L., Andriejauskas, T., Skrodenis, D. (2016)
Effect of intelligent transport systems on traffic safety. The Baltic Journal of Road and Bridge
Engineering, 11(2), 136–143. doi:10.3846/bjrbe.2016.16
38. Van Damme, O., Van Geelen, H., Courange, P. (2016) The Evaluation of Road Infrastructure
Development Projects. Transportation Research Procedia, 14, 467–473.
doi:10.1016/j.trpro.2016.05.099
39. Wang, K.C.P., Li, Q., Hall, K.D., Nguyen, V., Xiao, D.X. (2011) Development of truck loading
groups for the mechanistic-empirical pavement design guide. Journal of Transportation
Engineering, 137(12). doi:10.1061/(ASCE)TE.1943-5436.0000277
40. Xu, C., Hoel, L.A. (2001) A Methodology for Oversized Vehicle Trip Scheduling. Research Report
No. UVACTS-14-13-23. Center for Transportation Studies, University of Virginia, USA, 85 p.
58
319
HEALING OR HURTING: DEVELOPMENT OF HIGHWAY PUBLIC

TRANSPORTATION TECHNOLOGY
Nurul Qamar and Aan Aswari
Faculty of Law, University of Muslim Indonesia - Indonesia
E-mail: nurulqamar@umi.ac.id
Abstract
A collaboration of information technology with public transportation has brought an impact on the
improvement of public service in terms of public transportation online due to the government inca-
pability to provide good service. This study illustrates that there is an expectation to be built by the
concept of law namely through a legal approach that serves as a cure from the perceived failure re-
sulting from the lack of law in accommodating the needs of society for providing maximum service
on public transport. Meanwhile, the emergence of regulations that provide the legality of an action
in fact exacerbates the wound. This paper is not intended to reject the collaboration of technology.
However, it may fail to cope with human needs but not in conjunction with new technology. In con-
clusion, the law should easily materialize its goal of public transportation innovation and becomes a
healer from previous product failures.
Keywords: technology, transportation, online, street
Abstrak
Kolaborasi teknologi informasi dengan alat transportasi umum telah memberikan dampak pada
peningkatan pelayanan masyarakat dalam bentuk transportasi umum online, akibat ketidakmampuan
pemerintah untuk menghadirkan sebuah bentuk pelayanan yang ideal. Kajian ini memberi gambaran
bahwa terdapat sebuah harapan yang ingin dibangun oleh konsep hukum, bahwa sesungguhnya mela-
lui pendekatan hukum maka dapat digunakan sebagai obat menyembuhkan luka dari anggapan ke-
gagalan yang timbul akibat kekurangan hukum dalam mengakomodir kebutuhan masyarakat dalam
memberikan pelayanan maksimal pada angkutan jalan, sedangkan timbulnya regulasi yang memberi-
kan legalitas terhadap sebuah tindakan, realitasnya malah membuat sebuah luka yang semakin kro-
nis, dan di dalam tulisan ini tidak menolak teknologi untuk berkolaborasi dengan produk konven-
sional, akan tetapi idealnya produk terbarukan tersebut boleh saja mengalami kegagalan terhadap
mengatasi kebutuhan manusia, namun seharusnya kegagalan tersebut tidak beriringan dengan hadir-
nya teknologi terbarukan. Kesimpulan, hukum seharusnya mudah mewujudkan tujuannya pada inovasi
transportasi umum dan menjadi pengobat luka yang lama tak sembuh dari kegagalan produk sebelum-
nya.
Kata kunci: teknologi, transportasi, online, jalanan
Introduction Technology keeps developing from con-

Development in technology keeps provid- ventional to modern. Even nowadays public
ing advantages for the new ideas growth despite transportation such as city bus, taxi, and con-
its double-eyed blade effect in its usage,1 and it ventional taxi bike have transformed them-
is open for development and solution. The ma- selves after being collaborated with electronic
king of new law in society keeps developing for system including Gojek, Grab, and Uber. By
the sake of creating business effectively and ef- technology, public transportation becomes more
ficiently which sometimes cause violations. precise and effective in time,2 minimum busi-
2
Fania Darma Amajida, “Kreativitas Digital Dalam Masya-
1
Flourensia Sapty Rahayu, “Cyberbullying Sebagai Dam- rakat Risiko Perkotaan: Studi Tentang Ojek Online “Go-
pak Negatif Penggunaan Teknologi Informasi”, Journal Jek” Di Jakarta”, Jurnal Informasi, Vol. 46 No. 1, June
Of Information Systems, Vol. 8 No. 1, April 2012, p. 22. 2014, p. 117
320 Jurnal Dinamika Hukum
Vol. 18 No. 3, September 2018
ness capital, and effective in making the supply several problems which somehow dim the law, 8
and desirable price. In contrast, it seemingly by doing fictional obligation for the concrete
hurts other parties in the management of public right which is driver’s obligation implementing
transportation in terms of clashes between on- passenger or things movement for real as their
line driver and conventional driver several ti- responsibility.9 However it has now become
mes. commodity for them who violate the science to
Law substance with facilities, as well as get profit in every chance, that is the fulfill-
law enforcers who uphold legal supremacy, sol- ment of fake obligation for the sake of real
ve the traffic issues through the attempt in uti- right through reward on driver, the organizer of
lizing electronic system so the society prove the electronic agent10 is supposed to be responsible
thought of a put-forward statement. Thus, it upon the loss of other parties.
creates technology of public transportation Considering the technology impacts in so-
combined with electronic system which forces cial needs, this study is expected to offer solu-
the law to convergence to achieve efficiency, 3 tion as the answer of the problem on what po-
with the expectation of the supervision and the tentials needed to be noted to make people
implementation of direct reward and punish- aware of the new impacts in the application-
ment when the online drivers violate the law. based public transportation system. That it as
The improvement of online driver’s be- the cure of the wound from the failure in giving
havior on the street and how they prioritize the maximum service that should be there in the
passenger’s safety protected by the insurance, 4 product of public transportation, or improve-
as the guarantee of human rights to be protect- ment through the touch of technology as what
ted.5 However it is worsened by the passengers society expect in the implementation of tech-
who agree to break the law, such as violating nology as it should be.
the traffic law or not obeying the traffic rules. 6
In fact, the driver as the information and digital Discussion
society have passed through the recruitment The Presence of Healing Innovative Products
process upon the driver recruitment.7 The presence of electronic system-based
A profit-oriented behavior makes the on- product used by many people is the actual so-
line drivers ‘clone’ themselves which enable cial expectation of society to improve the ef-
them to run more than one vehicle simulta- ficiency11 in the utilization as well as covering
neously. If an old fiction telling about keeping the existing weakness as the cause of the fail-
the candle lit to get money by stealing, this is ure in the maximum use of product. This inno-
the time where keeping the internet connected
can make money by the onrechmatigheid, for 8
Nurul Qamar; Hikmawati Mustamin; Aan Aswari, “Local
real. The era of technology still brings about Wisdom Culture Of Bugis-Makassar In Legal Perspecti-
ve”, Adri International Journal Of Law And Social Scien-
ce, Vol. 1 No. 1, October 2017, p. 39; Ari Sandita Murti,
January 31th, 2018, Raup Rp.600 Juta Dari Order Fiktif
3
D Budhijanto, 2010, Hukum telekomunikasi, penyiaran, Tuyul Grab Dicokok Polisi, Available on: https://metro.
dan teknologi informasi: regulasi dan konvergensi. Ban- sindonews.com/read/1278361/170/raup-rp600-juta-
dung Refika Aditama. p. 97. dari-order-fiktif-tuyul-grab-dicokok-polisi-1517410196
4
Look at Victim’s Right on Regulation Number 22 Year Accessed on February 27th, 2018.
9
2009 About Traffic and Street Transport, Article 240. Grab, Kode Etik Mitra, Available on website: https://
5
Ratna Kumala Sari, Et. Al, 2018, Pertanggungjawaban www.grab.com/id/kodeetik/, Accessed on March 3rd,
Pidana Korporasi Gojek atas Tindak Pidana Kecelakaan 2018; Rules of Ministry on the Implementation of Social
Lalu Lintas oleh Pelaku Driver Ojek Online Di Yogyakar- Trans-port by Public Vehicle Not in Indonesian Route,
ta, Thesis, Yogyakarta: Universitas Islam Indonesia, p. regul-ated in PM 108 Year 2017
10
84. Article 1 verse (4) Governmental rule of Republik Indo-
6
Denico Doly, Penegakan Hukum Terhadap Undang-Un- nesia Number 82 year 2012 about Implementation of
dang Nomor 22 Tahun 2009 Tentang Lalu Lintas Dan Ang- Electronic System and Transaction.
11
kutan Jalan: Tantangan Dan Prospek. Jurnal Kajian, Vol. Manuela Manueke, Gustaaff Budi Tampi, Very Y Londa,
20 No. 3, September 2015, p. 232. “Persepsi Masyarakat Tentang Jasa Transportasi Berbasis
7
Yasraf Amir Pilliang, “Masyarakat Informasi dan Digital: Aplikasi Online Di Kota Manado (Studi Kasus Di Pt. Go-
Teknologi Informasi dan Perubahan Sosial”. Jurnal So- Jek). Jurnal Administrasi Publik, Vol. 4 No. 51, April
sioteknologi, Vol. 11 No. 27, 2012, p. 148. 2018, p. 2
Healing or Hurting: Development of Highway Public Transportation Technology 321
vative product improves many human needs by tional insurance, even flexibility for the driver
giving satisfaction through the abstract concept because they still can do other things,17 and al-
of legal idea that always tries to make a favor- so driver’s pattern on service innovation with
able environment.12 It is pioneered by technolo- other excesses that is not found in the conven-
gical product that is able to predict the re- tional public transportation.
quired work time more precisely. It is the public The improvement of discipline in traffic is
transportation as the tools to give more time to the effect of electronic system which enable
prepare or do many things in the middle of the the user to give feedback in form of information
moving process to the destination and improve so the users can choose needed action in antici-
the success.13 It is also the same as the techno- pating the potential problems. As an example,
logy of safety in application-based public trans- the speed of vehicle from the pick-up spot to
portation. the destination can be predicted. Driver’s and
A form of government protection from passenger’s decision is more accurate to the
many evolving threats to keep being able in destination because of the information system
going through the existed hole at the safe con- in the traffic condition on how it affects the
dition and breaking formed system in the pro- maximum limit of vehicle’s speed. Thus, the
cess of safe condition evolution is the evolution discipline is improved from the information to
coming from the legal idea existed for human in all parties in the moving process done by inno-
fulfilling the needs instinctively and the will of vative public transportation model, so the solu-
having comfortable life in the quiet and dis- tion towards the obstacles is solved by the role
cipline condition.14 That threat just weakens of technology in form of information.
the ideal system and makes failure of a product Next affordability aspects include: price
even more because application-based public affordability and service affordability. The price
transportation consists of a circle of difficult affordability through information in the imple-
system.15 mentation on innovative public transportation
Technology of public transportation is provides concrete image on the reasonable
now offering consumer comfort,16 which prio- price for the candidate of users. It means a
ritize consumer and driver interest. Both drivers prior agreement is made before giving some-
and passengers can explore and utilize the tech- thing, doing something, or not doing some-
nology facilities which then potentially avoid thing,18 the potential of the continuous agree-
traffic jam, time saving, location precision, le- ment in the alliance is surely inevitable. It is
vel of privacy and security, the supply of addi- different from the earlier public transportation
where the route is limited19 and the determined
12
Ahmad Jazuli, “Penegakan Hukum Penataan Ruang da-
price ignores the distance, for example in the
lam Rangka Mewujudkan Pembangunan Berkelanjut-
an”, Jurnal Rechts Vinding: Media Pembinaan Hukum taxi argo which is worked under the time in the
Nasional, Vol. 6 No. 2, August 2017, p. 278.
13 cab, or micro buses that takes far but the price
Raffi Wahyu Kusuma, Et. Al., “Perancangan Sistem In-
formasi Waktu Operasional Mengemudi Guna Menunjang is the same, or the conventional taxi bike in
Keselamatan Berkendara Dan Mengurangi Kecelakaan
which they have relatively high price.
Dalam Berlalu Lintas”, Semnasteknomedia Online, Vol. 3
No. 1, 2015. p. 391. This innovation of transportation innova-
14
Maulidi, “Paradigma Progresif dan Maqashid Syariah:
tion has made people feel secured legally. It
Manhaj Baru Menemukan Hukum Responsif”, Asy-Syir'ah
Jurnal Ilmu Syari'ah dan Hukum, Vol. 49 No. 2, Dec-
ember 2015, p. 253
15 17
Aan Aswari, Et. Al. “Legal Security On Cellphone Trading Noor Rahamah Haji Abu Bakar, “Masa Kerja Fleksibel da-
Through Electronic Media In Indonesia”. Jurnal Dinamika lam Sektor Formal di Bandar: Mengimbangi Kerjaya dan
Hukum, Vol. 17 No. 2, May 2017, p.184 Tanggungjawab Keluarga (Flexible Working Hours in the
16
Wiratri Anindhita, Melisa Arisanty, Devie Rahmawati. Urban Formal Sector: Balancing Career and Family Re-
"Analisis Penerapan Teknologi Komunikasi Tepat Guna sponsibilities)”, Akademika, Universiti Kebangsaan Ma-
pada Bisnis Transportasi Ojek Online (Studi pada Bisnis laysia, Vol. 83 No. 1, 2013, p. 78.
18
Gojek dan Grab Bike dalam Penggunaan Teknologi Ko- See Article 1234 Code of Civil Law Book III of Engage-
muniasi Tepat Guna untuk Mengembangkan Bisnis Trans- ment
19
portasi)." Prosiding Seminar Nasional Indocompac. May See Article 140 and Article 143 Law No. 22 Year 2009 of
2016, p.728 Traffic and Public Transport
means the community is closer to legal cer- using their free time to earn money which co-
tainty as one of the objectives of law. Starting mes from various levels of society and finally
from the evidence development, it develops in- makes a form of equal value of work which is
to the tools that humans use in their daily activ- considered by the community to be of the same
ity needs. It is where the paradigm related to quality as the work that was once considered
electronic evidence can be used as an accept- prestigious. Some people still consider this low
able proof by law enforcers 20 and people in but it seems ambiguous because in the digital
their daily life in doing contractual relations. information era many forms of multiple person-
Legal certainty is actualized by tracking alities are found.25
data contained in the data center to prove the Law accommodates all kinds of communi-
dispute between passengers, drivers and service ty interests and accommodates the develop-
providers. Thus the application of sanctions due ment of people needs in the implementation of
to such errors can be implemented as a reflec- online transportation which can be seen from
tion of the legal certainty. In relation to the ap- the various facilities offered by this technology.
plication of witnesses to application-based pub- This easiness makes many people prefer to use
lic transportation, it is initiated by data sources it and feel accommodating in present. There-
obtained from electronic information 21 which fore, the regulations that are formed to legalize
tend to be judged to provide objective inform- these activities will become healers of the
ation rather than information from the parties weaknesses found in conventional public tran-
solely related to the dispute resolution. sportation. The things stated above can be seen
Legality given by law to the implementa- from the diagram description below which
tion of product from technology impact in pub- shows the existence of the community's wealth
lic transportation raises the image 22 of work and accommodated in a renewable product.
opens up job vacancy widely to reduce number
of unemployment significantly.23 The image of a
job goes from the base and has a low rating in
community’s perspective. It takes effort to in-
crease people interest by offering a variety of
typical programs to be worthy. Since something
valuable tends to have high value and vice ver-
sa, something of high value is not necessarily
valuable.24 The image of ojek (motorcycle pub-
lic transportation) is now a job opportunity that Diagram 1. Reasons Why Consumers Choose
Online Transportation.26
can be utilized by various levels of society by
20
The law encourages the efforts to provide
Aradhya Sethia, “Rethinking admissibility of electronic
evidence. International Journal of Law and Information protection and law enforcement for the com-
Technology”, Vol. 24 No. 3, September 2016, p. 229- munity. This innovative product is an effort to
250.
21
See Article 1 Number 1 Law No. 19 Year 2016, Change stimulate legal products that support the imple-
on Law Number 11 Year 2008 of Electronic Information mentation of public transportation models and
and Transaction
22
Aan Aswari, Et. Al., “Harmonisasi Hukum Hak untuk Di- always accommodate all people interest over
lupakan bagi Koran Digital terhadap Calon Mahasiswa di
Makassar”, Kanun: Jurnal Ilmu Hukum, Vol. 20 No. 1,
April 2018, p. 43.
23
Geistiar Yoga Pratama, Et. Al., “Perlindungan Hukum
25
Terhadap Data Pribadi Pengguna Jasa Transportasi On- Rini Fidiyani, Dewi Sulistianingsih, Pujiono Pujiono. "Law
line Dari Tindakan Penyalahgunaan Pihak Penyedia Jasa and Ethics of Communication in Social Media." Jurnal
Berdasarkan Undang-Undang Nomor 8 Tahun 1999 Ten- Dinamika Hukum Vol. 17 No. 3, September 2017, p. 262
26
tang Perlindungan Konsumen”, Diponegoro Law Journal, Researcher team of YLKI, July 2018, Warta Konsumen
Vol. 5 No. 3, June 2016, p. 3. Transportasi Online Kawan Atau Lawan, Available on:
24
Nurul Qamar, Et. Al, 2018. Menguak Nilai Kearifan Lokal https://ylki.or.id/2017/07/warta-konsumen-transporta
Bugis Makassar, Perspektif Hukum Dan Pemerintahan. si-online-kawan-atau-lawan/, Accessed on March 14th,
Makassar, Cv.Social Politic Genius (Sign), p. 16 2018.
time.27 It is a solution expected by the commu- ventional public transportation systems, but the
nity as a cure of legal failure wounds in manipu- new community seems to be successful in colla-
lating humans to improve their quality through borating with these services with the touch of
the development of advanced products and desirable information technology today.
quality laws. Public transportation transformation29
sparks clashes between online drivers and con-
Legal Unpreparedness of Innovative Products ventional drivers. The reality raises new con-
Remains Hurting flicts since the existence of technology in public
The emergence of application-based pub- transportation causes problems at the empirical
lic transportation services is a form of response level. The response of the conventional public
from the government's inability to present a transportation providers hardly accept the real-
form of appropriate public transportation. The ity30 of the presence of technology widely used
impacts in Indonesia are several service com- by digital natives.31 It is seen that the new com-
panies that open services to the public in pre- munity try to shift the old which potentially
senting people dreams in the form of applica- raises clashes even though it is necessary for
tion-based public transportation, such as a des- the community to realize that the era of utili-
cription of consumer perceptions towards tran- zing this technology improved market quality.
sportation services below. Yet its presence ac- Yet, the efficiency that supported market quali-
tually creates new problems that go hand in ty is reduced due to excessive government reg-
hand with an innovative product previously ex- ulations and lack of competition market.32
pected to solve the problem of community The clashes are initiated by two groups:
needs and can actually treat old wound for a the conventional public transportation provi-
while. ders, and the application-based public transport
providers. The first group speaks up about the
legality of the system run by those who use the
application, and the second group speaks up
their feasibility to serve the public interest that
must continue to innovate through the presence
of products that can be utilized by more peo-
ple.
Legality of actions rises from these inno-
vative products. It becomes a gap considered in-
Diagram 2. Consumer Perception Towards On- appropriate by online drivers because they do
line Transportation Services.28 not have regulations related to their imple-
mentation, where their responsibilities have not
Perception based on the above data
been regulated in various laws and regulations
shows that a new power was born and begins to
that disrupt the given law. The confusion cau-
disturb the ideal situation that have been pre-
sed by this imperfect arrangement often occurs
viously enjoyed by the community that have
in every innovative product that seems to try to
mastered and carried out the public transport-
get rid of those who have enjoyed the benefits
ation system. The strength of the new commun-
for a long time, even though in fact there is a
ity brings the concept of services that should be
continuously improved by the organizers of con-
29
Aan Aswari, Et. Al. 2018, op, cit, p. 45.
30
Manuela Manueke, Gustaaf Budi Tampi, Londa, Very,
Op.Cit, April 2018, p. 5.
27 31
Lucchi, N., IHH, R.o.R., Högskolan i Jönköping & Inter- Eri Maryani. "Peningkatan Kualitas Pelayanan Perpusta-
nationella Handelshögskolan, "Internet content govern- kaan Untuk Digital Native Generation (Perspektif Peru-
ance and human rights", Vanderbilt Journal of Enter- bahan Karakter Pemustaka di Era Digital)." Prosiding Ko-
tainment and Technology Law, Vol. 16 No. 4, June 2014, munikasi Vol.1 No. 1. 2016, p. 214.
32
p. 809. Elizabeth Spahn. "Nobody gets hurt." Geo. J. Int'l L. 41,
28
Researchers Team of YLKI, op.cit 2009, p. 865.
role of the smart system in public transportati- ment of patterns that need to be improved, so
on. Yet, the regulation seems to be less respon- as to make this community more competitively
sive to immediately improve reality like trans- healthy through innovative findings in gaining
portation conventional has been clearly regula- passenger interest through quality service. 35 The
ted about rights and obligations. pattern of maintaining territories is often resol-
The feasibility offered by the renewable ved illegally, enriching and illegal access.
model really has advantages that can be used This technology product is an example
compared to the previous ones. It is often voi- that in fact there is a new weakness as a failure
ced in demonstrations initiated by the commu- of the form of the human product of in its de-
nity of application-based public transport users. velopment to meet the human needs today. Ap-
However it can also be used as a threat to mo- parently, this clarifies the weaknesses in con-
dern human, including: Making a gap between ventional systems that are deemed necessary to
the two communities, Violating the signposts, be renewed so as to heal wounds and even in
Increasing defaults, and Stimulating scientific the new system it appears to open new wounds
abuse accommodated by electronic agent pro- too, where actually the failure is supposed to
viders. not be felt immediately, when the role of vi-
Affordability in the area of public tran- sionary legal products and quality technology
sportation (accessibility)33 should be able to be products is present to accommodate community
fully achieved through development, but it be- interests. The picture shows just a form of inno-
comes a problem in accessibility when there are vation that is applied without a long considera-
often limitations in entering certain routes by tion, both in legal products that are always
application-based public transport when enter- expected to be hobbled to follow the changing
ing the area restricted by conventional public times, as well as human products that are ap-
transport users, for example has long been used plied but have a fairly high level of vulnerabili-
by the motorcycle taxi base community. Dis- ty. It becomes the author question, whether the
putes often occur when it is known that the dri- product from the study of science is only form-
ver has entered the conventional community ed without considering the humanism in its for-
area,34 by which it can be seen that the actual mation by ignoring the legal study as a whole.
affordability of the innovative system only ma- The information community are required
kes minimal accessibility. to improve the use of technology, but often not
An area of public facilities held by both accompanied by improvements in quality as
government and private sectors also become a what is seen in the reality of driving based on
contested area to be controlled by both com- the recruitment problem of drivers as human
munities such as airport, terminal, mall, tourist beings who are expected to bring goodness in
attraction etc. The assertion of certain areas of the implementation of this online transportation
power makes one community feel pressured and technology product. Although it is known that it
focuses on maintaining the territory by ignoring no longer uses a high speed and scrambles to
the optimal service from public transportation find passengers, it still does not adhere to traf-
that must be given to the public as users to ex- fic regulations, which is known that one of the
pand the coverage area, as well as the develop- causes of accidents is traffic rules violation.
Still, it is often seen that there are similarities
33 between online drivers and conventional drivers
Dian Maria Andriani, Nany Yuliastuti, “Penilaian Sistem
Transportasi yang Mengarah Pada Green Transportasi Di
35
Kota Surakarta, Jurnal Pembangunan Wilayah & Kota, Mamang Etta Sangaji. Sopiah, 2013, Perilaku Konsumen;
Vol. 9 No. 2, June 2013, p.185 Pendekatan Praktis disertai Himpunan Jurnal Peneli-
34
Ya’cob Billiocta, Padmasari, SI, September 29th 2017, tian. Yogyakarta: Penerbit Andi, p. 100; Stepanus S, Sa-
Sopir Konvensional Di Makassar Tebar Ancaman Hingga lonten S, & Supiyan S, “Analisis Persepsi Masyarakat
Bakar Atribut Ojek Online, Avaiable on: https://www. Terhadap Tingkat Pelayanan Angkutan Umum Kota Pa-
merdeka.com/peristiwa/sopir-konvensional-di-makas langka Raya (Studi Kasus Angkot Pada Jalan G Obos Dan
sar-tebar-ancaman-hingga-bakar-atribut-ojek-online. Jalan Rajawali)”. PROTEKSI journal (Proyeksi Teknik Si-
html, Accessed on January 8th, 2018. pil), Vol. 1 No. 2, July 2015, p. 201.
which should have significant differences, this transportation are almost resolved. Yet the pre-
shows that the technology must be supported by sence of new products actually opens up a great
its human resources. opportunity for some people to misuse their
The given regulations have accommoda- knowledge in achieving great profits. Now, fic-
ted the use of technology in public transporta- titious orders adorn the journey of improving
tion,36 but failure after failure continues to ac- public transportation, as if they have carried
company this renewable product from practical out their obligations but demand their rights
business held by venture capital owners in an perfectly. This condition significantly ignores
effort to avoid its interests hindered by regula- the orientation towards customers to develop
tions through bribery. It makes government offi- their long-term relationship in the form of the
cials accept bribes return the favor by reducing existence of technology public transportation
the barriers to conventional public transporta- products that are expected to provide maximum
tion that come from the regulation through its service when this is for people who can sustain
authority to make policies or maintain existence profits.39 The presented data for the govern-
by building strength through the formation of ment shows an increase in the workforce of the
strict rules and complicated procedures. This community, but the actual fictional data of the
bribery effort is like eliminating garbage car- community are in a stagnant state. Fictitious in-
rying health problems in human life and not formation presented by fictitious electronic da-
wanting to see it again. ta has the potential to produce policies that are
The situation is increasingly ironic when not on target.
people assume that the potential of work can Despite the consequences of a form of
be carried out easily. Moreover, open large- change, it cannot be proceeded since the law is
scale employment opportunities looks very pro- believed to always be able to provide strength
mising and considered to improve welfare le- in solving the arising problems and as a driving
vels.37 Yet, since this application is used, there force to realize the legal ideal of society.40 If
are many defaults occurring in various sectors the fictional story that candles are used as a
of financial institutions, and neglect of traffic source of information to steal others property,
regulations until the threat of life safety 38 then it should be used as an inspiration to ac-
shows the poor system that is considered pro- commodate the problem of maintaining an in-
mising, which finally forces people who have a ternet connection to be able to produce a level
higher level of knowledge to find a way to of welfare in an illegal way, which is to carry
achieve a variety of financing institutions, in out fictitious orders.
general both with legal patterns and vice versa. Fictitious orders can be detrimental to
There are so many failures that occur, the hope many parties impacting on determining govern-
of the people to reach their ideals is difficult to ment policies that miss the target which en-
materialize and this finding in a new product courage undesirable acts as a result of collabo-
with this technology is not ultimately conside- rative information technology and conventional
red as the best solution in realizing the govern- products. It would suffice when Indonesia is
ment's goal in encouraging the improvement of blacklisted in world trade for the online tran-
the people welfare. sactions before the 2000s in the misuse of credit
Threats appear to be increasingly large cards. This experience can be used as a refer-
opened and make wounds in renewable pro- ence to anticipate the existence of new unlaw-
ducts when the problems in conventional public
39
Leslier M. Valenzuela, Jay P. Mulki, and Jorge Fernando
Jaramillo. "Impact of customer orientation, inducements
36
PM 108 Year 2017, op.cit. and ethics on loyalty to the firm: Customers’ perspec-
37
Agung Hadi Pranoto. Siasat Manipulatif Dalam Kompetisi tive."Journal of Business Ethics” Vol. 93 No. 2, May
Pengemudi Gojek di Kota Surabaya. Jurnal Thesis De- 2010, p. 282.
40
partemen Antropologi, FISIP Universitas Airlangga, Dec- Nainggolan, 2015, Energi hukum sebagai faktor pendo-
ember 13th 2017, p. 2. rong efektivitas hukum, Bandung: Refika Aditama, p.
38
Denico Doly. op.cit, September 2015, p. 231. 170.
ful acts on new products born to the maximum vements are made in the form of evolution in
utilization of information technology. conventional public transportation.
Conclusion References
The evolution of ideal conditions initiated Amajida, Fania Darma. “Kreativitas Digital da-
by legal objectives is often along with the evo- lam Masyarakat Risiko Perkotaan: Studi
lution of threats to keep trying to maintain legal tentang Ojek Online “Go-Jek” di Jakar-
ta”. Jurnal Informasi, 2014, Vol. 46 No. 1.
failure to achieve legal objectives. Through ef-
June 2014: Pp. 115-128. DOI: 10.218
forts to produce a concrete picture of the legal 31/Informasi.V46i1.9657;
objectives in the implementation of an online
Andriani, Dian Maria; Nany Yuliastuti. “Penilaian
public transportation system, the touch of tech- Sistem Transportasi yang Mengarah pada
nology in terms of innovative products has given Green Transportasi di Kota Surakar-
rise to many forms of service from the develop- ta”. Jurnal Pembangunan Wilayah & Ko-
ment of accessibility, so that smooth informa- ta. Vol. 9 No. 2. June 2013 Pp. 183-193.
DOI: 10.14710/pwk.v9i2.6535;
tion can influence the decision making. Indeed,
information technology provides a lot of con- Anindhita, Wiratri. Melisa Arisanty. Devie Rah-
mawati. "Analisis Penerapan Teknologi
venience for the law in realizing its purpose in Komunikasi Tepat Guna pada Bisnis Trans-
this public transportation innovation so that it portasi Ojek Online (Studi pada Bisnis Go-
becomes a long-time wound healing which does jek dan Grab Bike dalam Penggunaan Tek-
not recover from previous product failures, due nologi Komuniasi Tepat Guna untuk Me-
to its limitations which are actually constructed ngembangkan Bisnis Transportasi)." Prosi-
ding Seminar Nasional Indocompac. May
by humans themselves. Now, failures due to
2016, Pp. 712-729;
previous limitations can be cured by the touch
Aswari, Aan. Et. Al. “Legal Security On Cellpho-
of information technology as a result of human ne Trading Through Electronic Media In
work and felt by the majority of the communi- Indonesia”. Jurnal Dinamika Hukum, Vol.
ty. Yet in fact it is predictable that in the futu- 17. No. 2. May 2017. Pp. 181-187. DOI:
re there will be legal limitations and innovative 10.20884/1.jdh.2017.17.2.800;
products when the existing technology starts to Aswari, Aan. Et. Al. “Harmonisasi Hukum Hak
be obsolete. If now the improvement of accessi- untuk Dilupakan bagi Koran Digital terha-
dap Calon Mahasiswa di Makassar”. Ka-
bility is initiated by the concept of borderless
nun: Jurnal Ilmu Hukum. Vol. 20 No. 1.
world, then in practice there is now a widening April 2018, Pp.39-62. DOI: 10.24815/ka
gap that limits territory in finding passengers nun.v20i1.9828;
for both communities, apparently borderless Bakar, Noor Rahamah Hj Abu. “Masa Kerja Flek-
world still has not been accepted by all people sibel dalam Sektor Formal di Bandar: Me-
who use technological tools in their daily lives. ngimbangi Kerjaya dan Tanggungjawab
Keluarga (Flexible Working Hours in the
Urban Formal Sector: Balancing Career
Suggestion
and Family Responsibilities)”. Akademika,
Conventional public transportation should Vol. 83 No. 1. 2013. Pp. 77-80;
be attached to the system run by the tools used Billiocta, Ya’cob. Padmasari SI. September 29th
by the application-based public transportation 2017. Sopir Konvensional Di Makassar Te-
for optimal performance to avoid clash between bar Ancaman Hingga Bakar Atribut Ojek
online and conventional providers in running a Online. Available on website: https://
good system. It suggests that these two com- www.merdeka.com/peristiwa/sopir-kon
vensional-di-makassar-tebar-ancaman-
munities can collaborate well so that there is no hingga-bakar-atribut-ojek-online.html.
need to make rules that explicitly separate the Accessed on January 8th, 2018.
two. By doing so, there are no new failures that Budhijanto, D. 2010. Hukum telekomunikasi,
are seen as incurable old wounds when impro- penyiaran, dan teknologi informasi: regu-
lasi dan konvergensi. Bandung: Refika December 2015. Pp. 251-264. DOI: 10.14
Aditama; 421/asy-syir'ah.2015.%25x;
Doly, Denico. “Penegakan Hukum Terhadap Un- Murti, Ari Sandita. Januari 31th. 2018. Raup Rp.
dang-Undang Nomor 22 Tahun 2009 Ten- 600 Juta dari Order Fiktif. 'Tuyul' Grab Di-
tang Lalu Lintas Dan Angkutan Jalan: Tan- cokok Polisi. Available on https://metro.
tangan Dan Prospek”. Jurnal Kajian, Vol. sindonews.com/read/1278361/170/raup-
20 No. 3. September 2015. Pp. 219-240; rp600-juta-dari-order-fiktif-tuyul-grab-
Fidiyani, Rini, Dewi Sulistianingsih and Pujiono. dicokok-polisi-1517410196. accessed on
“Law and Ethics of Communication in So- February 27th. 2018;
cial Media." Jurnal Dinamika Hukum. Vol. Nainggolan, J. 2015. Energi Hukum Sebagai Fak-
17 No. 3. September 2017. Pp. 258-265. tor Pendorong Efektivitas Hukum. Ban-
DOI: 10.20884/1.jdh.2017.17.3.1665; dung: Refika Aditama;
Grab, Kode Etik Mitra, Available on website: Pilliang, Yasraf Amir. “Masyarakat Informasi dan
https://www.grab.com/id/kodeetik/, Digital: Teknologi Informasi dan Perubah-
accessed on March 3rd, 2018; an Sosial”. Jurnal Sosioteknologi, Vol. 11
Jazuli, Ahmad. “Penegakan Hukum Penataan No. 27. 2012, Pp. 143-155;
Ruang Dalam Rangka Mewujudkan Pemba- Pranoto, Agung Hadi. “Siasat Manipulatif Dalam
ngunan Berkelanjutan”. Jurnal Rechts Kompetisi Pengemudi Gojek di Kota Sura-
Vinding: Media Pembinaan Hukum Nasio- baya”. Jurnal Thesis Departemen Antro-
nal, Vol. 6 No. 2. August 2017. Pp. 263- pologi. December 13th 2017. Pp. 1-10;
282; Pratama, Geistiar Yoga, Et. Al. “Perlindungan
Kusuma, Raffi Wahyu. Et. Al. “Perancangan Sis- Hukum terhadap Data Pribadi Pengguna
tem Informasi Waktu Operasional Menge- Jasa Transportasi Online dari Tindakan
mudi Guna Menunjang Keselamatan Ber- Penyalahgunaan Pihak Penyedia Jasa Ber-
kendara Dan Mengurangi Kecelakaan Da- dasarkan Undang-Undang Nomor 8 Tahun
lam Berlalu Lintas”. Semnasteknomedia 1999 Tentang Perlindungan Konsumen”.
Online. Vol. 3 No. 1. 2015. Pp. 391-396; Diponegoro Law Journal. Vol. 5 No. 3.
Lucchi, N., IHH, R.o.R., Högskolan i Jönköping & 2016. Pp.1-19;
Internationella Handelshögskolan, "Inter- Qamar, Nurul et al, 2018. Menguak Nilai Kea-
net content governance and human rifan Lokal Bugis Makassar, Perspektif
rights", Vanderbilt Journal of Entertain- Hukum Dan Pemerintahan. Makassar: So-
ment and Technology Law. Vol. 16 No. 4 cial Politic Genius Sign;
June 2014. Pp. 809: Qamar, Nurul. Hikmawati Mustamin. Aan Aswa-
Manueke, Manuela, Gustaaf Budi Tampi, Very ri. “Local Wisdom Culture Of Bugis-Makas-
Londa. “Persepsi Masyarakat Tentang Ja- sar In Legal Perspective”. Adri Interna-
sa Transportasi Berbasis Aplikasi Online Di tional Journal Of Law And Social Science.
Kota Manado (Studi Kasus Di Pt. Go-Jek)”. Vol. 1 No. 1. October 2017. Pp. 35-41;
Jurnal Administrasi Publik. Vol. 4 No. 51. Rahayu, Flourensia Sapty. “Cyberbullying Seba-
April 2018. Pp. 1-6: gai Dampak Negatif Penggunaan Teknologi
Manueke, Manuela; Tampi, Gustaaf Budi; Londa, Informasi”. Journal Of Information Sys-
Very. “Persepsi Masyarakat Tentang Jasa tems. Vol. 8 No. 1. April 2012. Pp. 22-31.
Transportasi Berbasis Aplikasi Online Di DOI: 10.21609/Jsi.V8i1.321
Kota Manado (Studi Kasus Di Pt. Go-Jek)”. Sangaji, Mamang Etta. Sopiah. 2013 Perilaku
Jurnal Administrasi Publik, Vol. 4 No. 51. Konsumen; Pendekatan Praktis disertai
2018. Pp 1-6; Himpunan Jurnal Penelitian. Yogyakarta:
Maryani, Eri. "Peningkatan Kualitas Pelayanan Penerbit Andi;
Perpustakaan Untuk Digital Native Gener- Sari, Ratna Kumala. Et. Al. 2018. Pertanggung-
ation (Perspektif Perubahan Karakter Pe- jawaban Pidana Korporasi Gojek Atas Tin-
mustaka di Era Digital)." Prosiding Komu- dak Pidana Kecelakaan Lalu Lintas Oleh
nikasi. Vol. 1 No. 1. 2016. Pp. 212-220; Pelaku Driver Ojek Online Di Yogyakarta.
Maulidi, Maulidi. “Paradigma Progresif dan Ma- Skripsi. Yogyakarta: University of Islam
qashid Syariah: Manhaj Baru Menemukan Indonesia;
Hukum Responsif”. Asy-Syir'ah Jurnal Il- Sethia, A. 2016. “Rethinking admissibility of e-
mu Syari'ah dan Hukum, Vol 49 No. 2. lectronic evidence”. International Jour-
nal of Law and Information Technol-

ogy. Vol. 24 No. 3. September 2016. Pp.
229-250. DOI: 10.1093/ijlit/eaw005;
Spahn, Elizabeth. "Nobody gets hurt." Geo. J.
Int'l L. No. 41. 2009. Pp. 861-1043;
Stepanus, S. S Salonten. & S Supiyan. “Analisis
Persepsi Masyarakat Terhadap Tingkat Pe-
layanan Angkutan Umum Kota Palangka
Raya (Studi Kasus Angkot Pada Jalan G
Obos Dan Jalan Rajawali)”. Jurnal PRO-
TEKSI (Proyeksi Teknik Sipil). Vol. 1 No.
2. July 2015. Pp. 198-206;
Tim Peneliti YLKI, Juli 2017, Available on
website: https://ylki.or.id/2017/07/war
ta-konsumen-transportasi-online-kawan-
atau-lawan/, accessed on March 14th,
2018
Valenzuela, Leslier M. Mulki Jay P and Jorge
Fernando Jaramillo. "Impact of customer
orientation, inducements and ethics on
loyalty to the firm: Customers’ perspec-
tive." Journal of Business Ethics Vol. 93
No. 2. May 2010. Pp. 277-291.
Yu Shen, Hongmou Zhang and Jinhua Zhao
Yu Shen, PhD
Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji
University, Shanghai, China
Future Urban Mobility IRG, Singapore-MIT Alliance for Research and Technology Centre,
Singapore
Hongmou Zhang
Massachusetts Institute of Technology, Cambridge, MA
Jinhua Zhao, PhD (Corresponding Author)

Edward H. and Joyce Linde Associate Professor
Massachusetts Institute of Technology
77 Massachusetts Ave, Cambridge, MA 02139
Email: jinhua@mit.edu
Abstract
This paper proposes and simulates an integrated autonomous vehicle (AV) and public
transportation (PT) system. After discussing the attributes of and the interaction among the
prospective stakeholders in the system, we identify opportunities for synergy between AVs and
the PT system based on Singapore’s organizational structure and demand characteristics.
Envisioning an integrated system in the context of the first-mile problem during morning peak
hours, we propose to preserve high demand bus routes while repurposing low-demand bus
routes and using shared AVs as an alternative. An agent-based supply-side simulation is built
to assess the performance of the proposed service in fifty-two scenarios with different fleet
sizes and ridesharing preferences. Under a set of assumptions on AV operation costs and
dispatching algorithms, the results show that the integrated system has the potential of
enhancing service quality, occupying fewer road resources, being financially sustainable, and
utilizing bus services more efficiently.
Keywords:
Autonomous vehicle; Public transportation; Agent-based model; First-mile problem; Mobility-
on-demand
1
1. Introduction
Autonomous vehicles (AVs) are poised to represent a revolutionary future for urban mobility
(Silberg and Wallace, 2012). Recent literature on the potential operation of AVs primarily
regards them as an upgrade to conventional personal vehicles with the essential characteristics
of demand responsiveness, fleet repositioning, and shareability (Alonso-Mora et al., 2017;
Correia and van Arem, 2016; Fagnant et al., 2015; Fagnant and Kockelman, 2014). However,
if we merely deploy AVs as upgraded versions of human-driven vehicles, we may not derive
the optimal benefits of the new technology, especially in large metropolitan areas with high
population densities and limited road resources. In cities where public transportation (PT) plays
a critical role, the relationship between AVs and the PT system should not be ignored. Most
studies, however, do not take into account such relationships. Only a few offer limited insight
into the AV-PT interaction and the PT is mostly pitted as a competitor (Chen and Kockelman,
2016 and Mendes et al., 2017). Discussions regarding AV and PT as complementary and
integrated are scarce. Liang et al. (2016) compared the service offered by automated taxi
systems with those provided by human-driven taxis over the last mile to train service; however,
improvements in the PT performance was not the focus of the study. Vakayil et al. (2017)
explored a hybrid transit system with on-demand AVs as an additional service to improve metro
connectivity; nevertheless, the relationship between AV and bus networks was neglected. Lenz
and Fraedrich (2016) discussed, however conceptually, the possibilities of hybridizing AVs
with PT, including improvements in inter-modality and individualization of the transit service.
To fill the research gap, we examine the attributes of and the interaction among the prospective
players in an integrated system (AV operators, PT operators, riders, public authorities, and
automakers) and explore the opportunities that AVs can provide when integrated into PT
systems. We envision a scenario in which AVs provide a complementary on-demand service to
conventional fixed-schedule fixed-route buses for the first/last mile and assess whether the new
integrated service improves the performance of the overall system. This study is based on
empirical travel demand and transit operation details derived from the smart card data in
Singapore.
The remainder of this paper is organized as follows. Section 2 discusses the integrated AV and
PT framework and highlights a list of characteristics of AVs that distinguishes the new system
from the traditional hybrid public transit system. Section 3 presents the case of Singapore. After
analyzing the current PT travel demand focusing on the first-/last-mile trips, we propose a
design where we preserve the high demand bus routes, repurpose low-demand bus routes
introduce AVs service. Next, we describe the agent-based model and simulate 52 integrated AV
and PT scenarios for first-mile trips during morning peak hours. We also simulate the status-
quo bus operations as the benchmark, again which we evaluate whether the integrated AV-PT
system is well-suited to improve the quality of service, occupy fewer road resources, and be
financially sustainable, and utilize buses more efficiently. Section 4 concludes this paper.
2. Toward a framework for integrated AV and PT system

There have been decades of efforts to design and to operate an integrated PT system with on-
demand flexibly-routed service since Daganzo (1978) and Wilson and Hendrickson (1980).
Taking advantages of both operating styles—fixed-route fixed-schedule service for corridors
2
of high demand and density, and demand-responsive service for areas with low ridership, low
density and scattered demand (Adebisi and Hurdle, 1982; Chang and Schonfeld, 1991a),
several integrated PT systems have been proposed (Aldaihani et al., 2004; Chang and
Schonfeld, 1991b; Li and Quadrifoglio, 2009). Various conceptual models evaluating the fixed-
route and demand-responsive transit services have also been presented (Diana et al., 2009; Li
and Quadrifoglio, 2010; Qiu et al., 2015). Constrained by the 20th-century technology, the
integrated demand-responsive transit system faced critical challenges including high costs to
operate the service, difficulties to communicate with the riders and manage shared rides, and
problems to control drivers.
In recent years, the rapid development of Information and Communication Technology (ICT)
has led to the emergence of transportation network companies (TNCs), e.g., Uber and Lyft, and
revived interest in flexible on-demand systems. Online communication platforms manage
shared rides more efficiently by matching the real-time demand with dynamic fleet operation
strategies with lower price for the rides. Some operators have incorporated non-dedicated
vehicles into their service models to reduce capital costs, allowing the fleet size to vary
dynamically according to changes in demand (see the case of FlexDenmark).
When AVs become available, they may offer an opportunity to address many organizational
and technological challenges in the current system such as reducing labor cost, improving
compliance with planning and operation control, expanding service hours, avoiding erroneous
human-driving behavior, and optimizing the spatial and temporal allocation of the PT services.
Figure 1 summarizes the AV characteristics from the perspectives of operation (AV operator
and PT operator), governance (public authority), technology (AV producer), and consumption
(AV riders and PT riders), distinguishing the integrated AV-PT system from the traditional
hybrid transit system. In this study, we focus on the relationships between AV operators and PT
authority, and between AV operators and conventional transit operators. The rest, despite its
importance, is beyond the scope of this paper.
The AV operators offer a new travel option from the passengers’ point of view. Indicators of
conventional PT service quality also apply to the shared-AV service on demand (See the
comprehensive set of service indicators in Eboli and Mazzulla, 2011). AV service can be more
attractive if the system is able to provide a flexible door-to-door service, cheaper thanks to
lowering operating cost, and personalized to fit individual preferences and circumstances. The
service can also benefit the elderly or passengers with disabilities. There can also be drawbacks
associated with the AV service. The impact on the driver employment will have profound
implication on the labor relation, contracting, and social justice in general, which is part of the
broad discussion on the social impact of the automation. Without the appearance of human
drivers, the driverlessness may raise concerns from riders regarding safety (the maturing of
driverless technology), security (preventing crimes in the vehicle) and the quality of customer
service in general (e.g., human assistance). These concerns are partially demonstrated in Dong
et al. (2017).
3
Figure 1. Stakeholders around AV-PT operation and AV characteristics from the perspectives
of operation, governance, technology, and consumption
AV brings the attention of public authority to understanding the role and impacts of AVs on the
mobility system. A variety of organizational structures can be imagined with respect to the
ownership of the transit and AV operators, the interaction between the operators, and the degree
of regulation and intervention from the public authority. Possible organizational models of the
integrated AV and PT systems can be envisioned based on the existing transit governance
literature (Bruun, 2013; Costa, 1996; van de Velde, 1999; Wilson, 1991) and the emerging
experience in regulating the TNC firms in London, Singapore, and Boston. They include but
are not limited to 1) the laisse-faire structure; 2) the experience on regulating TNC firms (e.g.
in London); 3) the “on-demand paratransit pilot program” led by the Massachusetts Bay
Transportation Authority (MBTA); 4) the deregulation of PT in the UK (Wilson, 1991); 5) the
“Scandinavian” model (Costa, 1996; van de Velde, 1999); 6) the fully coordinated model.
Appendix A summarizes the key characteristics of the six models but does not get into the
details of the specific responsibilities and functions of the public authorities and other
stakeholders. For instance, within the Scandinavian model, the organizational structures in
Copenhagen (Denmark), Malmö (Sweden), and Adelaide (Australia) can be distinguished
further based on the level of central planning and the design of incentives to tendering parties
(van de Velde, 1999).
4
2.1. Assumptions of the AV-PT systems based on the organizational models in Singapore
Public transit in Singapore is highly regulated by the Land Transport Authority (LTA), which
is responsible for service integration including fares, information, and route design 1 . The
subway system, known as mass rapid transit (MRT), is planned and constructed by the LTA,
who then leases the operating licenses to transit operators. The LTA tenders out bus services
with a contract for five years plus a two-year extension based on service performance. The bus
operators compete for the market to win the contract to enter the market, while the bus routes
are still designed by the LTA.
Based on the characteristics and regulatory structure of the transport system in Singapore, we
identify the opportunity for synergy between AV and PT system and assume the following
characteristics of the integrated AV-PT system.
• Planning and regulation: the public authority LTA remains responsible for transit network
planning and the assignment of service areas and routes.
o On-demand AV operation is supported and regulated by the public authority.
o On-demand AV operation is provided in dedicated service areas and is limited to first-
and last-mile service in the current phase.
• Transit fare and subsidy: The transit fare is heavily regulated. The fare adjustment formula
is regularly reviewed to ensure that the fares are in line with the cost structure and
productivity achieved in the industry and that they are affordable to the public.
o The fare structure of the AV service is considered part of the transit fare review and
adjustment.
o The cost for an AV ride is calculated in Section 3 to test the financial viability of the
AV service.
• Coordination and competition: the current transit system in Singapore is a coordinated
system with the unified fare and information structure.
o AV operation is coordinated with the MRT system for feeder service. In the designated
area, the AV operation and the adjusted bus operation are also coordinated (see section
3 for the specific design).
• Fare, Ticketing and Information Integration: the AV service is integrated into the transit
system regarding:
o Ticketing: the transit smart card (or any other payment systems used by PT) can be
used for the AV service.
o Fare: the cost of a shared-AV ride is set to be close to the current transit fare.
o Information: the operating characteristics of both AV and conventional PT modes are
known to each other and available to the passengers.
3. An integrated AV-PT system for the first-mile service in Singapore

First, we examine the first-mile performance in the status quo PT system using the travel
records from the contactless e-purse application (CEPAS) card—the transit smart card in
Singapore. Second, based on the organizational structure assumed in Section 2, we propose one
feasible operational scenario of integrating AVs into the PT system, aiming at improving the
first-mile service during the a.m. peak. Finally, we assess the performance of the newly
integrated system via agent-based simulation and compare it with the current PT-only system.
3.1. Study area and data

We define the first-/last-mile trips in Singapore as the connecting trips to/from the MRT stations,
including walk and bus as the dominant modes. The CEPAS data used in this study covers all
1
TransitLink, the subsidiary of LTA, is in charge of the specific development of service integration and design.
5
PT trips in August 2013, with over 175 million travel records. Since the CEPAS requires the
tap-in and tap-out policy in both bus and MRT systems, each trip record archives the date and
time of every entry and exit activity and the boarding and alighting stops/stations. The fare
policy recognizes the temporally adjacent journey stages as one single journey, which may
contain a series of MRT and/or bus rides.2 The details on the bus trips made before and after
the MRT ride are clearly identified.
Appendix B discusses the first-/last-mile travel demand during the entire workdays, and we
focus here on the two-hour morning peak (7:00 to 9:00 a.m.). Figure 2 shows the volume of
passengers entering each MRT station during the morning peak and the mode shares by bus vs.
other modes. We highlight four stations with over 10,000 passengers and more than 50% of
bus mode share, among which, the Tampines area is chosen as the case study. Tampines has
one of the highest population densities in Singapore, with about 240,000 residents living in a
12 km2 area (HDB, 2015). The MRT travel demand in Tampines station is also one of the
highest: between 7 and 9 a.m. on a typical workday, more than 15,000 passengers use the MRT
service, including over 8,000 taking a bus to access the train station. Figure 3 ranks the 27 bus
routes (and directions) serving the MRT station by the average workday peak-hour bus
ridership. There is an obvious imbalance of patronage: the top five routes account for more
than 55% of the first-mile travel demand, and the bottom 11 routes account for less than 10%
of the demand. The footprints of the high-demand and low-demand bus routes are illustrated
in Figure B in Appendix C.
Figure 2. First-mile travel demand between 7 and 9 a.m.

(Calculated by the authors based on CEPAS smart card data)
2
According to the LTA, in a single journey, the transfer time should be less than 45 minutes while the total
travel time should be less than two hours.
6
Figure 3. Bus ridership to Tampines MRT by route
(Calculated by the authors based on CEPAS smart card data; “_1” and “_2” indicate the two
directions of the same bus route)
3.2. Design and assessment of the integrated AV-PT system

People in Singapore are highly dependent on the MRT system. The first-/last-mile connectivity
is one of the major issues in the current system of Singapore. Numerous solutions have been
designed and proposed (Lesh, 2013) to improve the connectivity: 1) by proactive planning,
such as enhancing walkability (Cervero, 2007), encouraging park-and-ride (Hamer, 2010), and
increasing intermodal coordination (Chien and Schonfeld, 1998; Sørensen and Longva, 2011);
and 2) by innovative travel modes, such as bike sharing (Cervero et al., 2013; DeMaio, 2009;
Shaheen et al., 2010), and electric scooters (Shaheen and Finson, 2003).
We propose an integrated AV-PT system by preserving the high demand bus routes and
repurposing low-demand bus routes while using the shared AV service on demand as an
alternative:
• The 16 busiest bus routes (shown in Figure 3) are kept to ensure that 90% of the travel
demand is served efficiently.
• The 11 low-demand routes are repurposed in four different ways depending on the specific
configuration of the routes: 1) rerouting to reduce the detour, 2) rerouting to bypass the
high traffic center, 3) maintaining the route but with larger stop spacing, and 4) shifting
the destination to the nearby Simei MRT station. The details of these repurposing options
are discussed in Appendix C.
• Only the remaining 10% of the demand previously serviced by the 11 low-demand buses
is to be served by the on-demand AV service.
7
We further assume the following additional AV operation features in the simulation:
On vehicle ownership and operation:
• Vehicle ownership: individually owned AVs are not taken into account in the model; and
the fleet-based AVs are rented from non-dedicated vehicle providers (e.g., car-rental
companies) whenever needed.
• Labor cost: there is no cost for human drivers. Service and maintenance cost of AVs are
accounted in the rental fees paid to the car rental companies.
• Operation Compliance: the AVs comply fully with the central controller and do not reject
service qualifying requests from customers.
• Fleet rebalancing: when the AV is idle, it is sent to a parking area3.
On communication and sharing:
• Mobility-on-demand: the AV service is requested on demand.
• Real-time shareability: The requests for ridesharing are processed and responded to in real-
time. Both the sharing and no-sharing scenarios are simulated and compared.
We built an agent-based model4, simulating the behaviors and interaction among three types of
agents in the system: the passenger, the AV, and the bus, shown in Figure 4.
Figure 4. Agent-based simulation of passengers, buses and AVs

(rounded rectangles refer to the agents, ellipses refer to locations, rectangles with solid line
refer to the movements of agents, rectangles with dashed line refer to the interaction between
agents, and diamonds refer to the decisions of agents.)
3
We do not introduce more sophisticated anticipatory rebalancing algorithms in this paper, but they can be
incorporated in the future research. See Wen et al. (2017). In the simulation vehicles simply return to the
parking lot of the Tampines MRT station, owned by public authority for both bus and shared AV parking. The
parking cost is thus not included in the model.
4
Both the analytical approaches and simulation approaches can be used to assess an integrated AV-PT system.
Analytical approaches are important, but the closed form solution may not be available in the context of the
dynamic and complex real world. They are often based on a hypothetical and small transit network (e.g.,
Aldaihani et al., 2004; Li and Quadrifoglio, 2009).
8
Passenger behaviors
This model focuses on the passengers who used to ride the bus to the MRT station. The
incidence of passenger trips based on the bus boarding in every 15 minutes during the peak
hours. A Voronoi diagram based on the location of bus stops is generated to assign travel
demand to each building5.
As AVs are not yet widely available to the public, current worldwide studies on AV preference
are hypothetical (e.g., Bansal et al., 2016; Krueger et al., 2016; Kyriakidis et al., 2015; Lu et
al., 2017; Payre et al., 2014; Shin et al., 2015; Yap et al., 2016) and the settings in the cases
from other countries may not transfer easily to the context of Singapore. Since the consumer
preference for AVs in Singapore is not the focus of this study, this paper does not model the
mode choice behavior. Instead the model assumes the fixed modal split between bus (90%) and
AV (10%) and focuses on the supply-side simulation. But it is a critical future research area to
have a dedicated AV preference study based on the local context in Singapore.
When the AV is chosen as the travel mode, the passengers start to call for the ride repeatedly
until the system has successfully booked a car for him/her. Once the vehicle is assigned, he/she
then moves to the pick-up location to wait for the AV. As the density of bus stops in Singapore
is very high, the pick-up location is set as the closest bus stop from the passenger. To prevent a
passenger from waiting forever when no AV is available, a maximum waiting time is set to 10
minutes in the study, beyond which he/she will forgo the AV request and travel by bus.
When the bus is chosen, the passenger will walk directly to the bus stop to wait for the next
bus. As the frequency of bus service is high (less than 10 min headway), the model assumes
that passenger arrival time is independent to the bus schedule (Abkowitz and Tozzi, 1987; Fan
and Machemehl, 2002; Frumin and Zhao, 2012). Thanks to the high service reliability of
Singapore’s bus system, all passengers get onto a bus within 10 minutes waiting time.
On-demand shared AV service

The concept of the mobility-on-demand system was proposed and demonstrated in the 1970s,
and was called Dial-a-Ride Transit (DART) (Wilson et al. 1976). Two decades later, Dial (1995)
proposed a fully automated DART system. More advanced transit systems integrating various
vehicle sharing services—including buses and taxis—were recently designed and tested (e.g.,
Atasoy et al., 2015; Djavadian and Chow, 2016; Jung and Jayakrishnan, 2016). In this study,
the on-demand AV service with ridesharing resembles a taxi-sharing system (Galland et al.,
2014; Martínez et al., 2015) but the service area is restricted to the Tampines town with only
the first/last mile service, consistent with the proposed usage of the AV prototype in Singapore
(Chong et al., 2011). The driving behaviors for all AVs are identical, and no booking requests
are rejected. The idle AVs are sent to the parking lot of the Tampines MRT station if there is no
demand for AV. Each AV allows a maximum of four passengers, the capacity of the AV
prototype, to share the ride. As discussed in Section 2.1, the AV operation is regulated by the
transit authority to be complementary to the bus. It does not compete against bus service for a
bigger market share.
The AV routing follows the shortest path. We assumed that a passenger prefers to ride alone if
possible even when he/she agrees to share the ride. Thus, when a passenger calls for a ride, the
system first scans all empty AVs. If there are empty AVs, the system assigns the closest
5
Due to the lack of demographic information at the building level, the population are assumed to be evenly
distributed to the buildings in the cell.
9
available AV to pick up the passenger. Once the AV and the passenger are matched, a notice is
sent to the passenger requesting a meet-up at the pick-up point. If there are no empty AVs, the
system searches for all occupied AVs, the passengers of which have agreed to share the ride,
based on pre-assigned passenger sharing preference. The criteria to match the shared rides are
as follows. For each shareable AV, after picking up the passenger at location i, the decision of
whether to pick up the next passenger at location j is determined by the following conditions:
• The vehicle must have enough available seats.
• Global detour constraint: For a passenger at location i in a shared trip, regardless of the
number of passengers in the AV, and regardless of the boarding order of passenger i, the
total travel time from pick-up location of passenger i to MRT station s, 𝑇"# , must be less
than the direct service time, 𝑡"# multiplied by an overall detour ratio, α:
Tis < a × tis , where a > 1 . (1)
• Incremental detour constraint: In each event of picking up an additional passenger at
location j, the deviated travel time for the previous passenger i due to picking up j (i.e.,
𝑡"% + 𝑡%# ) must be less than the direct service time, 𝑡"# , multiplied by the step-by-step
detour threshold, β.
tij + t js < b × tis , where b > 1. (2)
The pricing structure is purely distance-based, without a “base fare”. It also reflects the detour
ratio of each passenger. For ridesharing trips, a discount for passenger i is calculated based on
the actual detour ratio r. The discounted price, 𝑃" is computed as in Eq. (3).
𝑃" = 𝑝 ∙ 𝑑" ∙ (1 − 𝑟 0 ) (3)
where:
• p is the base AV fare per km;
• 𝑑" is the direct travel distance of passenger i in km;
• 𝐷" is the detoured actual distance;
3
• 𝑟 = 5 4 − 1, the detour ratio;
4
• 𝛾 is the degree of discount offered due to the detour.
In the simulation, we set 𝑝" to S$1 per kilometer which is at the same order of magnitude as
the base bus fare during peak hours6, α to 2, β to 1.2, and 𝛾 to 2. Please note that past literature
shows that given the bus pricing structure in Singapore, there is no price variation in the first
mile access by bus (Mo et al., 2018). We cannot find locally relevant willingness-to-pay
estimates from prior literature. Thus, we design the pricing discount and the distance detour to
roughly balance each other so that the fixed demand assumption remains reasonable.
Bus routing and dispatching

While the shared-AV service is on-demand, the bus service still operates on a fixed route and
fixed schedule basis. Two agents are introduced for bus routing and dispatching. The bus route
controller is a high-level agent that manages all bus routes, including the stop sequences to
follow and the headways of each bus route during different time periods. We assigned a bus
dispatcher for each bus route. Based on the stop lists and headways, the bus dispatchers
generate buses for the corresponding routes according to the actual bus schedule. The data for
bus routes and schedules are provided by the LTA. The routes used are the ones presented in
Figures 3, which intersect the Tampines area and stop at the MRT station.
6
The base bus fare is about 0.87 Singapore dollar.
10
Once a bus is generated by the dispatcher, the route is determined based on a sequential stop
list. The bus vehicle follows the route at a constant speed. Upon arrival at a bus stop, each bus
agent dwells for a certain amount of time to pick up all the passengers who are waiting at the
bus stop at the bus arrival time. The actual bus dwell time depends on the number of passengers
entering and exiting the bus, as well as the total number of passengers on the bus. However,
we simplify the dwell time to be fixed (30 seconds) in the simulation due to the lack of detailed
operational information and corresponding coefficients. The waiting time of each passenger is
archived, and the number of passengers onboard is logged. We also calculate the travel time
and travel kilometers for each passenger.
The model was coded in AnyLogic 7.3. We executed the simulation between 6:00 and 9:00 a.m.
(10,800 s) where one second in the simulation corresponds to one second in reality. The first
hour (6 to 7 a.m.) was regarded as the warm-up period; the outputs from the last two hours (7
to 9 a.m.) were recorded and analyzed.
3.3. Simulation results and analysis

Benchmark scenario: Bus-only service
We first simulated the status quo scenario with only the bus services, following the existing
bus routes, and used it as the benchmark to compare the new integrated AV-PT scenarios. The
simulation was run for 100 iterations. Figure 5 compares the observed numbers of boardings
in the CEPAS data with the simulated numbers in the benchmark model in every 15 minutes
interval. For most time intervals, the range of the simulated results captures the empirical
values.
Figure 5. Number of boardings comparison between simulation and CEPAS data
We also compared the share of ridership in each bus route between the simulated results and
the observation using the root-mean-square error (RMSE):
N R
åå ( rˆ - r nr )
2
nr
RMSE = n =1 r =1
, (4)
N ×R
where 𝜌89: is the simulated share of ridership of the bus route r at the n-th simulation; 𝜌9: is
the observed share of ridership of bus route r at the n-th simulation; N is the total number of
simulation runs; R is the total number of bus routes. The RMSE was 0.026, indicating that the
11
simulated demand in each bus route well fitted the actual patronage derived from the CEPAS
data7.
Integrated AV-PT scenarios

As proposed in Section 3.2, we simulated the scenarios where shared AVs would accommodate
the first-mile service as a replacement of the 11 low-demand bus routes. 90% of the passengers
would still choose the bus, as per usual, whereas the remaining 10% would hail an AV for the
ride to the MRT station. After repurposing the 11 bus routes, we evaluated the service
performance of different AV fleet sizes (from 10 to 35 vehicles) and with two extreme settings
of sharing preference of the AV riders: (1) no one is willing to share the ride, and (2) everyone
is willing to share. A total of 52 AV-PT scenarios were developed, and for each scenario, 100
times of simulations were run.
Among the many factors affecting system performance, we look into three perspectives to
evaluate the integrated AV-PT system:
• Passenger perspective (i.e., service quality): travel time, particularly the out-of-vehicle
travel times (OVT) including walking time and waiting time
• System perspective (i.e., road traffic): we use the passenger car unit kilometers (PCU-km)
as the approximation to the road resource usage for mixed-vehicle fleets (BITRE, 2015;
BTRE, 2007). The PCU-km is derived as the product of the vehicle-kilometers traveled
and the passenger car equivalent (PCE) factor. The empirical PCE factor for buses in
Singapore is approximately 2.75 (Fan, 1990; Yeung et al., 2015). As the AV is classified
as a personal car, its PCE factor is one.
• Business perspective (i.e., operating cost and revenue): whether the AV service can be
financially viable based on the proposed pricing model.
The simulation results of the 52 scenarios are plotted in Figure 6, where each dot represents
the average value of 100 simulation runs; the triangles represent the non-ridesharing setting,
whereas the circles represent the all-ridesharing setting. The dashed line shows the average
value of 100 simulation runs of the bus-only benchmark scenario.
Owing to the heat and humidity in Singapore, the outdoor walking time and the waiting time
are of high importance to service quality. The average OVT in the benchmark bus-only scenario
is approximately 420 seconds. We recorded in the simulation the total time spent by each
passenger from his/her incidence being generated to the time when he/she boards a bus or AV.
Figure 6(a) compares the average OVT of all passengers in the AV-PT scenarios with that in
the benchmark bus-only scenario. More AVs reduce the OVT until they saturate; and the OVT
is shorter when rides are shared for the AV fleet size less than 30, beyond which sharing no
longer affects the OVT. The first-mile travel demand can be served by only 17 AVs with
ridesharing for the average OVT to be on par with the benchmark scenario. However, without
sharing, we would need 22 AVs, an increase of fleet size by a third, to guarantee the same
service quality.
7
The visual comparison in Figure 5 and the RMSE are both aggregate measures, which do not capture the
variations in the number of passenger boardings at the bus stop level, resulting in an overestimation of the realism
of simulation.
12
Figure 6(b) indicates the total PCU-km in different AV-PT scenarios. By repurposing the 11
low-demand bus routes based on the strategies in Appendix C (mostly to reduce detours), we
can save a total of 860 PCU-km, which we consider as the acceptable “quota” for AVs in order
not to increase the total traffic in the study area. The simulation results indicate that about
18~19 AVs can be accommodated in the study area within the traffic “quota.” Since our
simulation model only uses the basic heuristic AV dispatching and assignment algorithms, the
AV-PT system performance can be further improved with more advanced algorithms.
Figure 6. Simulation results
Figure 6(c.1) and 6(c.2) plot the estimated revenue and cost per kilometer for various AV fleet
sizes. The operation cost of AVs is based on the car leasing plan between Uber and SMOVE8,
an emerging car-sharing company in Singapore. We assume that the AVs are owned by SMOVE,
or its equivalent, and rented to the AV operators using the same leasing plan: the costs being
calculated based on the usage of vehicles in terms of both time and distance. The simulation
results show that smaller fleet size has higher profits per kilometer. Sharing increases the profit
margin per kilometer. When the fleet size increases, the profit gap between sharing and non-
sharing scenarios shrinks. The integrated AV-PT service is financially viable in the tested range
of fleet size assuming the above cost structure and the pricing model.
8
See https://www.smove.sg/#pricing for more details.
13
4. Discussion
In addition to the 90%-10% scenario (90% by high demand bus routes and 10% by AV)
reported above, we also tested two alternative AV-PT scenarios: the first is an extreme case
where all first mile bus services are replaced by the AVs, and the second is to use AV to serve
45% of first-mile bus demand by preserving only the 5 busiest bus routes. Appendix D shows
that in both alternative scenarios, the system performance worsens, i.e., there will be far larger
PCU-km in the system to maintain the same level of service quality as in the benchmark case.
We should not take it for granted that an integrated AV-PT system will automatically improve
the status quo. Instead, it requires attentive research and innovative design to identify the
opportunity window where an optimal combination of AV and PT can enhance service quality,
utilize buses more efficiently, occupy fewer road resources, and be financially sustainable.
The Ministry of Transport (2017) of Singapore is planning for the pilot deployment of AVs in
several planning areas. The integration of AV into the PT system is one of the primary
objectives, including the exploration of operational characteristics and business models. Our
study proposes an integrated AV-PT system based on Singapore’s PT organizational structure
and demand characteristics. Specifically, we envision a synergic system in the context of the
first-mile service during morning peak hours, where high-demand bus routes are preserved,
low-demand routes are repurposed, and shared AVs are introduced as an alternative, and
evaluate the system from the passenger, business and system perspectives using agent-based
simulation.
The real world is a complex nonlinear system. There is a large room to improve the simulation
model in future research:
• The 90%-10% scenario is an intuitive design. A more systematic approach to optimize the
PT-AV split should be considered.
• The fixed AV demand is a strong assumption. A full multimodal demand model should be
developed based on consumer preference surveys conducted in Singapore for the
estimation of the mode choice between the bus, AVs and walking in the first-mile access
to MRT stations.
• A dynamic interaction between demand (modal choice) and supply (AV and bus fleets)
should be taken into consideration (See Wen et al. 2018).
• The dispatching of AV in the current simulation is based on simple heuristics. A more
efficient ridesharing model can be incorporated with more sophisticated algorithms to
optimize the ridesharing efficiency.
• Other sub-models such as traffic flow models and bus dwelling time models can also be
integrated to make the simulation closer to the real-world circumstances.
Finally, the case study presented in the paper is only one of many possible ways to integrate
the shared AVs into the PT system from a planning perspective. As in Figure 1, there are many
alternative schemata from the viewpoints of operation, governance, technology, and
consumption, e.g., pricing or ticketing integration, information sharing, institutional design, etc.
Private companies can also be potential AV fleet owners and operators. Various ownership
structures affect the interests and incentives of different players, leading to different behaviors
in the integrated mobility system. Each operator may have its own strategies toward its peers,
e.g., one may choose to compete against or to coordinate with the others.
14
Acknowledgement
The research is supported by the National Research Foundation (NRF), Prime Minister’s Office,
Singapore, under CREATE programme, Singapore-MIT Alliance for Research and Technology
(SMART) Centre, Future Urban Mobility (FM) IRG. The authors would like to thank the
anonymous reviewers and the members of MIT JTL Urban Mobility Lab and MIT Transit Lab
for their comments.
References
Abkowitz, M., Tozzi, J., 1987. Research contributions to managing transit service reliability. J. Adv.
Transp. 21, 47–65.
Adebisi, O., Hurdle, V.F., 1982. Comparing fixed-route and flexible-route strategies for intraurban
bus transit. Transp. Res. Rec. 854, 37–43.
Aldaihani, M.M., Quadrifoglio, L., Dessouky, M.M., Hall, R., 2004. Network design for a grid hybrid
transit service. Transp. Res. Part A Policy Pract. 38, 511–530.
Alonso-Mora, J., Samaranayake, S., Wallar, A., Frazzoli, E., Rus, D., 2017. On-demand high-capacity
ride-sharing via dynamic trip-vehicle assignment. Proc. Natl. Acad. Sci. U. S. A. 114, 462–467.
Atasoy, B., Ikeda, T., Song, X., Ben-Akiva, M., 2015. The concept and impact analysis of a flexible
mobility on demand system. Transp. Res. Part C Emerg. Technol. 56, 373–392.
Bansal, P., Kockelman, K.M., Singh, A., 2016. Assessing public opinions of and interest in new
vehicle technologies: An Austin perspective. Transp. Res. Part C Emerg. Technol. 67, 1–14.
Bruun, E.C., 2013. Organizational structures and contract models, in: Better Public Transit Systems :
Analyzing Investments and Performance. Routledge, Abingdon, UK, pp. 237–253.
Bureau of Infrastructure Transport and Regional Economics (BITRE), 2015. Traffic and congestion
cost trends for Australian capital cities. Australian Government, Canberra, Australia.
Bureau of Transport and Regional Economics (BTRE), 2007. Estimating urban traffic and congestion
cost trends for Australian cities, Working paper 71. Canberra, Australia.
Cervero, R., 2007. Transit-Oriented Development’s ridership bonus: A product of self-selection and
public policies. Environ. Plan. A 39, 2068–2085.
Cervero, R., Caldwell, B., Cuellar, J., 2013. Bike-and-ride: Build it and they will come. J. Public
Transp. 16, 83–105.
Chang, S.K., Schonfeld, P.M., 1991a. Optimization models for comparing conventional and
subscription bus feeder services. Transp. Sci. 25, 281–298.
Chang, S.K., Schonfeld, P.M., 1991b. Integration of Fixed- and Flexible-Route Bus Systems. Transp.
Res. Board J. Transp. Res. Board 1308, 51–57.
Chen, T.D., Kockelman, K.M., 2016. Management of a shared autonomous electric vehicle fleet:
Implications of pricing schemes. Transp. Res. Board J. Transp. Res. Board 2572, 37–46.
Chien, S., Schonfeld, P., 1998. Joint optimization of a rail transit line and its feeder bus system. J.
Adv. Transp. 32, 253–284.
Chong, Z.J., Qin, B., Bandyopadhyay, T., Wongpiromsarn, T., Rankin, E.S., Ang Jr., M.H., Frazzoli,
E., Rus, D., Hsu, D., Low, K.H., 2011. Autonomous personal vehicle for the first- and last-mile
transportation services. 2011 IEEE 5th Int. Conf. Cybern. Intell. Syst.
https://doi.org/10.1109/ICCIS.2011.6070337
Correia, G.H.A., van Arem, B., 2016. Solving the User Optimum Privately Owned Automated
Vehicles Assignment Problem (UO-POAVAP): A model to explore the impacts of self-driving
vehicles on urban mobility. Transp. Res. Part B Methodol. 87, 64–88.
Costa, Á., 1996. The organisation of urban public transport systems in Western European
metropolitan areas. Transp. Res. Part A Policy Pract. 30, 349–359. https://doi.org/10.1016/0965-
8564(96)00002-X
Daganzo, C.F., 1978. An approximate analytic model of many-to-many demand responsive
transportation systems. Transp. Res. 12, 325–333. https://doi.org/10.1016/0041-1647(78)90007-
2
DeMaio, P., 2009. Bike-sharing: history, impacts, models of provision, and future. J. Public Transp.
12, 41–56. https://doi.org/http://dx.doi.org/10.5038/2375-0901.12.4.3
Dial, R.B., 1995. Autonomous dial-a-ride transit introductory overview. Transp. Res. Part C Emerg.
15
Technol. 3, 261–275.
Diana, M., Quadrifoglio, L., Pronello, C., 2009. A methodology for comparing distances traveled by
performance-equivalent fixed-route and demand responsive transit services. Transp. Plan.
Djavadian, S., Chow, J.Y.J., 2016. Agent-based day-to-day adjustment process to evaluate dynamic
flexible transport service policies. Transp. B Transp. Dyn. 1–26.
Dong, X., DiScenna, M., Guerra, E., 2017. Transit user perception of driverless buses, in: TRB 96th
Annual Meeting Compendium of Papers. Washington, DC, pp. 1–15.
Eboli, L., Mazzulla, G., 2011. A methodology for evaluating transit service quality based on
subjective and objective measures from the passenger’s point of view. Transp. Policy 18, 172–
181. https://doi.org/10.1016/j.tranpol.2010.07.007
Fagnant, D.J., Kockelman, K.M., 2014. The travel and environmental implications of shared
autonomous vehicles, using agent-based model scenarios. Transp. Res. Part C Emerg. Technol.
40, 1–13.
Fagnant, D.J., Kockelman, K.M., Bansal, P., 2015. Operations of shared autonomous vehicle fleet for
Austin, Texas, market. Transp. Res. Board J. Transp. Res. Board 2536, 98–106.
Fan, H.S.L., 1990. Passenger car equivalents for vehicles on Singapore expressways. Transp. Res.
Part A Gen. 24, 391–396.
Fan, W., Machemehl, R.B., 2002. Characterizing bus transit passenger waiting times. 2nd Mater.
Spec. Conf. Can. Soc. Civ. Eng. .
Frumin, M., Zhao, J., 2012. Analyzing passenger incidence behavior in heterogeneous transit services
using smartcard data and schedule-based assignment. Transp. Res. Rec. J. Transp. Res. Board
2274, 52–60.
Galland, S., Knapen, L., Yasar, A.-U.-H., Gaud, N., Janssens, D., Lamotte, O., Koukam, A., Wets, G.,
2014. Multi-agent simulation of individual mobility behavior in carpooling. Transp. Res. Part C
Hamer, P., 2010. Analysing the effectiveness of park and ride as a generator of public transport mode
shift. Road Transp. Res. A J. Aust. New Zeal. Res. Pract. 19, 51–61.
Housing & Development Board (HDB), 2015. Key statistics: HDB annual report 2014/2015 [WWW
Document]. URL http://www10.hdb.gov.sg/ebook/ar2015/html5/index.html?opf=tablet/Key
Statistics.xml
Jung, J., Jayakrishnan, R., 2016. Simulation framework for modeling large-scale flexible transit
systems. Transp. Res. Rec. 2466, 31–41.
Krueger, R., Rashidi, T.H., Rose, J.M., 2016. Preferences for shared autonomous vehicles. Transp.
Res. Part C Emerg. Technol. 69, 343–355.
Kyriakidis, M., Happee, R., de Winter, J.C.F., 2015. Public opinion on automated driving: Results of
an international questionnaire among 5000 respondents. Transp. Res. Part F Traffic Psychol.
Behav. 32, 127–140.
Lenz, B., Fraedrich, E., 2016. New mobility concepts and autonomous driving: The potential for
change, in: Markus Maurer, J. Christian Gerdes, Barbara Lenz, Hermann Winner (Eds.),
Autonomous Driving: Technical, Legal and Social Aspects. Springer Berlin Heidelberg, Berlin
Heidelberg, pp. 173–191. https://doi.org/10.1007/978-3-662-48847-8
Lesh, M.C., 2013. Innovative concepts in first-last mile connections to public transportation. Third
Int. Conf. Urban Public Transp. Syst. https://doi.org/10.1061/9780784413210.007
Li, X., Quadrifoglio, L., 2010. Feeder transit services: Choosing between fixed and demand
responsive policy. Transp. Res. Part C Emerg. Technol. 18, 770–780.
https://doi.org/10.1016/j.trc.2009.05.015
Li, X., Quadrifoglio, L., 2009. Optimal Zone Design for Feeder Transit Services. Transp. Res. Rec. J.
Transp. Res. Board 2111, 100–108. https://doi.org/10.3141/2111-13
Liang, X., Correia, G., van Arem, B., 2016. Optimizing the service area and trip selection of an
electric automated taxi system used for the last mile of train trips. Transp. Res. Part E Logist.
Transp. Rev. 93, 115–129.
Lu, Z., Du, R., Dunham-Jones, E., Park, H., Crittenden, J., 2017. Data-enabled public preferences
inform integration of autonomous vehicles with transit-oriented development in Atlanta. Cities
63, 118–127.
16
Martínez, L.M., Correia, G.H.A., Viegas, J.M., 2015. An agent-based simulation model to assess the
impacts of introducing a shared-taxi system: an application to Lisbon (Portugal). J. Adv. Transp.
49, 475–495.
Ministry of Transport, 2017. Request for information for pilot deployment of autonomous vehicles as
a form of public transportation in upcoming development areas in Singapore [WWW
Document]. URL https://www.gebiz.gov.sg/ptn/opportunity/opportunityDetails.xhtml (accessed
12.21.17).
Mo, B., Shen, Y., Zhao, J., 2018. Impact of built environment on first- and last-mile travel mode
choice. Transp. Res. Rec.
Payre, W., Cestac, J., Delhomme, P., 2014. Intention to use a fully automated car: Attitudes and a
priori acceptability. Transp. Res. Part F Traffic Psychol. Behav. 27, 252–263.
Qiu, F., Li, W., Haghani, A., 2015. A methodology for choosing between fixed-route and flex-route
policies for transit services. J. Adv. Transp. 49, 496–509. https://doi.org/10.1002/atr.1289
Shaheen, S.A., Finson, R., 2003. Bridging the last mile: A study of the behavioral, institutional, and
economic potential of the Segway human transporter. 82nd Annu. Meet. Transp. Res. Board.
Shaheen, S.A., Guzman, S., Zhang, H., 2010. Bikesharing in Europe, the Americas, and Asia: Past,
present, and future. Transp. Res. Rec. J. Transp. Res. Board 2143, 159–167.
Shin, J., Bhat, C.R., You, D., Garikapati, V.M., Pendyala, R.M., 2015. Consumer preferences and
willingness to pay for advanced vehicle technology options and fuel types. Transp. Res. Part C
Silberg, G., Wallace, R., 2012. Self-driving cars: The next revolution [WWW Document]. URL
https://www.kpmg.com/US/en/IssuesAndInsights/ArticlesPublications/Documents/self-driving-
cars-next-revolution.pdf
Sørensen, C.H., Longva, F., 2011. Increased coordination in public transport: Which mechanisms are
available? Transp. Policy 18, 117–125.
Vakayil, A., Gruel, W., Samaranayake, S., 2017. Integrating shared-vehicle mobility-on-demand
systems with public transit. 96th Annu. Meet. Transp. Res. Board.
van de Velde, D.., 1999. Organisational forms and entrepreneurship in public transport: classifying
organisational forms. Transp. Policy 6, 147–157. https://doi.org/10.1016/S0967-
070X(99)00016-5
Wen, J., Chen, Y.X., Nassir, N., Zhao, J., 2018. Transit-oriented autonomous vehicle operations with
integrated demand-supply interaction, Transportatio Research Part C, under review
Wen, J., Zhao, J., Jaillet, P., 2017. Rebalancing Shared Mobility-on-Demand Systems: a
Reinforcement Learning Approach, in: IEEE ITSC 2017 The Third International Workshop on
Intelligent Public Transports: Toward the Next Generation of Urban Mobility.
Wilson, N.H.M., 1991. Organizational options for public transportation in the US. Transp. Plan.
Wilson, N.H.M., Hendrickson, C., 1980. Performance models of flexibly routed transportation
services . Transp. Res. Part B Methodol. 14, 67–78. https://doi.org/10.1016/0191-
2615(80)90033-8
Wilson, N.H.M., Weissberg, R.W., Hauser, J.R., 1976. Advanced Dial-a-ride algorithms research
project: Final report. Massachusetts Institute of Technology, Cambridge, MA.
Yap, M.D., Correia, G., van Arem, B., 2016. Preferences of travellers for using automated vehicles as
last mile public transport of multimodal train trips. Transp. Res. Part A Policy Pract. 94, 1–16.
Yeung, J.S., Wong, Y.D., Secadiningrat, J.R., 2015. Lane-harmonised passenger car equivalents for
heterogeneous expressway traffic. Transp. Res. Part A Policy Pract. 78, 270–361.
17
Appendix: Supplementary Information
A. Six potential organizational models for the integrated AV-PT system
Model #1: Laissez-faire structure

In a laissez-faire model, the authority never intervenes in the operations. The private operators
aim to increase market share and maximize their profits. They only operate in profitable areas
or routes and compete against each other for ridership, service coverage, and profitability.
If the service could be potentially profitable, AV operators enter that market as well. The natural
advantage of AVs (e.g., lower operating cost with automation replacing human drivers) may
trigger the traditional PT operator to optimize its workforce planning and crew scheduling
strategies for cost savings to compete. It may also enhance the quality of customer service in
conventional transit since lack of personal maintenance is one of the potential weaknesses of
the driverless system. Another possibility is the conventional transit operator may also reform,
replacing its human driver with driverless technology, making the whole transit system
driverless. This would lead to competition between two driverless systems.
However, the absence of planning and regulation could result in a series of negative
consequences, including separate fare and ticketing system, lack of coordination in network
planning and scheduling. The lack of coordination creates a barrier for passengers to use AV in
conjunction with conventional transit systems owing to the separate fare systems. The
competition in the market can also result in more traffic congestion in profitable areas with lack
of service in unprofitable areas and routes. Without coordination, each AV operator may install
its own vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) system with high capital
cost, which may not even be compatible with that of the other operators. In this case, the
absence of an effective system design fragments the transit market, deteriorates the quality of
service, and reduces revenues. Even with the AV technology, this market may not be sustainable
owing to the reduction of fleet size, service frequency, and the layoff of employees.
Model #2: TNC-regulated structure

This structure is close to the current organizational model with regulations on TNC firms in
many cities like New York City, London, and Singapore, where a vocational driver’s license is
required by the local transport authorities before operating the TNC services. In this model, the
authorities in these cities regulate the private operators to impose service standards upon these
firms, e.g., safety. These TNC firms are not financially supported by the public.
TNC firms like Uber are attempting to develop their own AVs. It is possible that current TNC
firms may become AV operators in the future. In this case, the AV operation can be licensed or
regulated by public authorities, with the service still provided by the private operators and not
publicly subsidized. The AV service operates independently of transit operations. Without the
strings on labor cost for human drivers, AV operators may be more competitive. The
consequences may be similar to model #1, which may lead to a fully driverless system for all
operators to reduce operating cost or a fragmented system. Although there are certain service
standards promulgated by the public authority, the whole system is still unlikely to be efficient.
Model #3: MBTA paratransit structure

This model is another structure in which private operators can be supported by the public
authorities if they offer public services. The model can be understood as replicating the “on-
demand paratransit pilot program” led by the Massachusetts Bay Transportation Authority
18
(MBTA). In this program, the on-demand service provided by Uber and Lyft serve passengers
with disabilities. The fare for each ride is only two US dollars, and the rest of the operating cost
is subsidized by the MBTA. In the program, the MBTA could potentially save a large amount
of cost for operating its own paratransit services. The ticketing and information systems are
still separate. There is also no coordination between Uber/Lyft and MBTA travel modes.
In contrast to model #2, this model forges a new partnership between private operators and PT
agencies. If these TNC firms upgrade their fleet to AVs, the same organizational structure still
applies. The authority subsidizes private AV operations targeting a specific type of service, e.g.,
paratransit for the elderly, people with disabilities, and children. This structure helps to
incentivize the private AV operators to provide more social but (maybe) unprofitable service.
Compared with the current human-driven paratransit service, the on-demand AV service is able
to serve the vulnerable population with greater flexibility and reduced cost.
Model #4: UK deregulation structure

This structure presents an organization option akin to the UK deregulation model (in areas
outside London). In contrast to the TNC-oriented models above, there is a central planning
authority in this structure overseeing the operations at a certain level and the service is more
integrated, although the operators are private and compete against each other. In profitable
areas and routes, the scenario is close to the models above, where the operators set up their
own fare structure, routing and scheduling, and information systems, and compete against each
other. However, the public authority may intervene in the operations in unprofitable areas by
regulating fares and routes or may provide transit service directly. The pros and cons of this
model are briefly summarized by Wilson (1991).
In a model with AVs, the advantages of this model are similar to the virtuous circle discussed
in Model #1 with efficient service, quick response to market changes, and technology
innovation. With basic interventions and regulation, the negative consequences envisioned in
Model #1 may not occur. However, private AV operators are still for-profit, which may squeeze
out the conventional operators from the market. Even in a total driverless transit system, owing
to the unrestricted market entry and exit, the AV service may still be discontinuous in the
unprofitable areas, leading to less demand for PT service in these areas.
Model #5: Scandinavian structure

This model accords to the well-known “Scandinavian” or “London” model (Costa, 1996; van
de Velde, 1999). The public transport and service goals are set and planned by the authority,
which then contracts out the planned transit service to private operators. Competition is
introduced in the model where the operators attempt to reach the service standards set by the
authority to enter the market. For instance, in Stockholm (Sweden), bus service is contracted
to several public or private operators, including Arriva (from Deutsche Bahn), Keolis
(headquartered in Paris), and Nobina (from Sweden). The PT authority ensures the coordination
of the transit system.
In the AV scenario, to ensure coordination, central planning from a public authority is needed.
Both AV and conventional PT operators compete for contracts to enter the market. The
operating cost of AV is lower, but the conventional transit operators may have advantages in
other conditions, e.g., higher security standards and more help from service personnel. The
ticketing and the customer information systems are more likely to be coordinated and integrated,
which may attract more riders in general, owing to its convenience.
19
Model #6: Fully coordinated structure
This model envisions a fully coordinated scenario without competition in or for the market.
Similar to model #5, the fare and information systems are coordinated and integrated. The
authority is responsible for the design of the network route structure and the designation of the
service areas of each operator, aiming at maximizing social welfare. The operators are
responsible for the operation, management, and fleet maintenance. In contrast to model #5,
where the operators are free to compete for all service routes and areas, the operators in this
model run their services in different areas, making comparisons of service performance
difficult.
This model is highly regulated. Different service areas are designated to AV and PT operators
by the public authority, and the operators do not compete against each other for more service
areas. For instance, the planner may design a trunk-feeder or diameter-tangent network
structure. In the network, the trunk (or diameter) routes may be assigned to the conventional
fixed-route fixed-schedule transit operator, while the feeder (or tangent) service connecting to
the arterial routes is assigned to on-demand AVs.
B. First and last mile travel demand in Singapore during workdays

Singaporean residents largely rely on the PT system to travel. According to the statistics from
the LTA, with a total population of 5.47 million, as of 2014, less than 10% of people own
private cars. Moreover, the total number of private cars is continuously decreasing, owing to
the shrinking quota of Certificate of Entitlement. Thus, the availability of private vehicle-based
travel modes is limited. The household interview travel survey (HITS) conducted in 2012 also
suggests that during the a.m. peak hours, 70% of commuters travel to work using PT, including
buses and the MRT, whereas only 16% of them drive cars.
The average MRT travel demand per hour during workdays is shown in Figure A.1, along with
the volumes of the first- and last-mile bus trips9. The percentages of the first- and last-mile bus
trips to the connected MRT trips in each hour of the day are also shown. During morning peak
hours, first-mile bus trips are more than the last-mile bus trips, while the pattern gets reversed
in the evening peak, which is reasonable for any workday. The chart also shows a clear pattern
of increase in the MRT travel demand during morning and evening peak hours. For instance,
during morning peak hours—between 7 and 9 a.m.—there are in total 450,000 passengers
entering the MRT stations. Among them, nearly 134,000 passengers (over 30%) take a bus trip
to reach the MRT station. The findings from the CEPAS database are consistent with the
corresponding modal share from the HITS data. Observing Figure A.2, the sizes of first- and
last-mile trips are quite similar in all MRT stations, i.e., the number of passengers boarding and
alighting from the MRT are almost equal. Both two charts suggest that the travel pattern during
workdays is very regular. Figure A.2 also illustrates the share of first-/last-mile trips in the areas
close to the boundary of Singapore Island is greater than that in the CBD area, which is
probably due to the higher density of MRT stations in the CBD area.
9
For all bus trips connecting with the MRT, we filtered the trip distance of less than 5 km to be the first-/last-mile
(bus) trip.
20
(1) First-/last-mile travel demand during workdays by hour
(2) First-/last-mile travel demand during workdays by MRT station

Figure A. First-/last-mile travel demand on workdays
(Data source: calculations by the authors based on CEPAS smart card data)
C. Bus routes and the repurposing methods
Footprints of bus routes in Tampines

Figure B maps the footprints of bus routes. Figure B.1 shows the spatial coverage of high-
demand bus routes (i.e., from route 291_1 to 18_1 in Figure 3), which in total service over 90%
of the first-mile bus travel demand in the Tampines area. Figure B.2 shows the coverage of
low-demand routes (i.e., from 38_2 to 10_2). The low-demand buses do not significantly
21
enhance the service coverage in this area. The frequency of low-demand buses is quite high
with less than 10-min headway but they carry low first mile demand.
(1) Footprint of high-demand bus routes
(2) Footprint of low-demand bus routes

Figure B. First-mile bus routes and origin-destination patterns
(Data source: calculation by the authors)
Repurposing low-demand bus routes

The details on each of the low-demand bus routes and the corresponding repurposing scenarios
are summarized below:
22
i. Rerouting to reduce the detour. The bus routes presented in Figures C(1) and C(2) both
use the Tampines MRT station as the terminal. The routes deviate largely. After
repurposing, we let these buses go directly to the terminal.
ii. Rerouting to bypass the center. The bus routes presented in Figures C(3), C(4), and C(5)
all pass by the MRT station area. However, the low ridership indicates that, although the
buses are designed to connect to the MRT station, only a few passengers use the service.
Therefore, these routes were redesigned to bypass the station area to mitigate the pressure
on the road traffic around the MRT station.
iii. Keeping the route as is. The bus routes presented from Figure C(6) to C(10) all go to the
MRT station directly. We kept the route as is; however, the bus stops in the Tampines area
were skipped.
iv. Shifting the destination to the adjacent MRT station. The bus routes presented in Figure
C(11) connect Tampines and Simei, which is the MRT station next to Tampines station.
However, the travel demand between the two MRT stations by bus is low; the demand
between the two can be either serviced by the MRT or by the AVs. Hence, we shifted these
two bus routes to end at the Simei MRT station.
i. Rerouting to reduce detour:

ii. Rerouting to bypass the center:
iii. Keeping the route as is:
23
iv. Shifting the destination to the adjacent MRT station:
(11) Bus route 38_2

Figure C Repurposing low-demand bus routes
D. Simulation results for alternative scenarios

The section demonstrates two alternative scenarios with various repurposing proportions of
shared AV, based on Figure 3. For each scenario, 12 sub-scenarios were simulated with
different fleet sizes. With each AV fleet size, we run the simulation for 100 iterations. Every
passenger is assumed to be willing to share the ride. We focus on the impact on the road
resources, i.e. PCU-km. The results of the two scenarios are illustrated in Figure D.1 and D.2,
respectively.
Figure D. AV fleet size vs total PCU-kilometers traveled in alternative scenarios
Figure D.1 shows the PCU-km assuming the AVs serving 45% of the first-mile demand
originally served by bus. The five busiest bus routes are kept in the system while the other 22
routes are repurposed. The total saving of PCU-km by repurposing the bus routes are about
2,400 km. However, to serve the first-mile travel demand with the same level of service, at
least 96 shared AVs are required, generating nearly 4,300 PCU-km. The total PCU-km
increases. Similarly, Figure D.2 shows the results where all first-mile bus demands are replaced
by the shared AVs. By simply removing all bus routes in this area, about 7,600 PCU-km are
released. However, 208 shared AVs are needed, generating over 9,100 PCU-km. In both
24
scenarios, much more road resources are occupied, making the road traffic worse and harming
the performance of the overall mobility system. We are able to improve the system performance
based on an intuitive design of repurposing only the low-demand buses that serve 10% of the
demand. However, the comparison to the alternative scenarios shows that we cannot take it for
granted that an integrated AV-PT system will automatically improve the system performance.
25
Global Journal of Engineering and Technology Advances, 2020, 02(02), 011–022
Global Journal of Engineering and Technology Advances

Cross Ref DOI: 10.30574/gjeta
Journal homepage: http://www.gjeta.com
(R E V I E W A R T I C L E )
A review and analysis of railway transportation system in the economic community

of West African States: Towards the development of sustainable regional goal
Bouraima Mouhamed Bayane, Qiu Yanjun * and Yusupov Bekhzad
School of Civil Engineering, Southwest Jiaotong University, Chengdu 610031, China.
Publication history: Received on 24 January 2020; revised on 06 February 2020; accepted on 08 February 2020
Article DOI: https://doi.org/10.30574/gjeta.2020.2.2.0004
Abstract
A comprehensive evaluation of the current state of the railway transport system in eleven countries of the Economic
Community of West African States (ECOWAS) is presented. Four themes of the railway transportation system
comprise of infrastructure, rolling stock, operational performance, telecommunication and signaling are examined in
the article. The transport policy regarding the railway transport system is analyzed along with the regulation and
institutional framework. The findings of this study show that the railway transportation system is characterized by
deteriorated infrastructure, obsolete technologies, fragmented and old databases, scarcely accessible investments,
low-quality operations regarding safety and performance, and restricted connectivity and interoperability. Based on
the socio-economic context in the region and review of recent railway projects, a series of strategies are proposed to
meet future regional visions. These strategies are envisaged to contribute to the development of an efficient and
interoperable railway transportation system in the Economic Community of the West African States which in turn will
improve the interconnectivity and enhance the economic growth and trade in West Africa.
Keywords: Railway transport system; Economic Community of West African States (ECOWAS); Sustainable railway;
Regional goal; Strategy.
1. Introduction
Railway transportation has recently shown a global revival through the extension of its network that becomes the
vision of the African Union to meet the second aspiration of the Agenda 2063 [1]. This revitalization of the railway is
due to the significant socio-economic and environmental spatial relations generated by the transport. While road
contributes to over three-quarters of global greenhouse gas emissions of the transport sector [2], the railway is
considered to have considerable potential advantages than its counterpart regarding traffic safety, energy
consumption, lower costs, and environmental protection [3, 4].
While significant return on investment has been noticed worldwide during the last thirty years, there is low freight
railway market development and a decline in passenger transport in West Africa [3]. Road network has the exclusive
dominant transportation role in West Africa [5], however, its network stays poor quality with insufficient density and
extent. As a result, an exclusive partnership for railways is a prerequisite for transportation operation services.
According to Seba Minsili, Kisito, Gilbert, Jean, Gadam and Christian [6], huge investment is needed for the
construction and maintenance of railway, which gave the feeling of its expensiveness to challenge road. Yet rail is
much cheaper through feasible economies of scale [3]. Notwithstanding the limited financial resources, the West
African countries have begun to put the railway transport system into a new phase of modernization at international
standard for the next decades [7], because of its crucial role in the industrial revolution in Europe and the remarkable
economic developments in North America, India, Russia, China, Japan, Hong Kong and Australia [8-10].

Corresponding author
E-mail address:
Copyright © 2020 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0.
Bouraima et al. / Global Journal of Engineering and Technology Advances, 2020, 02(02), 011–022
The current timing seems to be suitable, as rapid economic growth than the standard has been perceptible in the
region directing to favorable forthcoming [11]. The opportune possibility to develop railway is comprehensible when
putting together this growth along with unused resources and arisen middle class [11]. In conformity with this, the
extension of the railway network become the eyesight of the African Union to meet the second aspiration of its Agenda
by 2063 [1] and solve the mobility difficulties of its increasing population [12].
There is consequently an important discontinuity between the existing condition of rail infrastructure, the recurrent
perception of regional integration development and temporary projects in execution [13]. While sustainable
development goals cannot be achieved without financing rail infrastructure, there should have certain caution in the
implementation of the project, particularly in countries with limited financial resources [14].
The objective of this paper is to point out specific areas for capacity development in order to strengthen access to
more reliable, safe and secure railway transportation within the countries of ECOWAS. That is to say, the goals of each
component of the railway were to diagnose the current state of the railway transportation system in ECOWAS to
pinpoint ways regarding interconnectivity targets. The study eventually selected the lacunas in the regional
perspectives with the current potentiality for developing strategies for the development of the regional railway
transportation system.
2. Methodology
The methodology of the study includes the use of major sources from a practical survey, analysis of evidence from the
experimental study as well as analysis of data from different published sources. The list of selected countries consists
of four English speaking countries (Ghana, Liberia, Nigeria, and Sierra Leone) and seven French-speaking countries
(Benin Republic, Burkina Faso, Guinea, Ivory Coast, Mali, Senegal, and Togo).
3. State of the existing railway transportation system

In this section, we evolve an examination of various elements of the railway transport system of the region, with a
special focus on the infrastructure, rolling stock, operational performance, and telecommunication and signaling.
3.1. Infrastructure
The railway network of the ECOWAS as a whole is around 10,188 km of which only half out of 12 of the national
networks are for sub-regional purposes. Freight is principally being transported on the railway infrastructure (Figure
1). This is due to the fact that the region is characterized by innumerable natural and mineral resources. The majority
of railways are connected to mining zones and ports because of the dominance of the regional importation and
exportation traffic by sea.
Figure 1 Track density in the selected countries of ECOWAS

Figure 1 Track density in the selected countries of ECOWAS
12
Restricted size and quality of the railway network have characterized the region which results in the decline in traffic
volumes that line should bear, influencing severely efficiency and productivity. Only a smaller number of lines
traverse borders, for instance, Sitarail between Burkina Faso and Ivory Coast, and Transrail between Mali and Senegal.
The deficient length of rail routes indicates that there is a transfer of freight to road transport in order to get the final
destinations. For instance, in the Benin Republic, the railway provides inland railheads in Parakou city from which
goods can be transported to Niger by roads, which raises the costs compared to road-only transport.
A variety of track gauges ranging from narrow (600mm) to standard (1435mm) contribute to the hindrance of a
balance traffic flow on the railway networks within the countries of the ECOWAS [15]. This explains the highest
transportation costs internationally noticed in West Africa in comparison to other developing countries of other
regions [16]. While the meter gauge is mostly used in all French-speaking countries, the Cape gauge can be found in
English speaking countries [5]. Except for Guinea and Liberia where there is the existence of the standard gauge, it is
also found in the newly constructed railway lines of the Port Harcourt-Onne, Itakpe-Ajaokuta-Warri and Kaduna-
Abuja in Nigeria [17]. The operated track gauges are elucidated in Figure 2. It can be seen that no significant efforts
have been done until now to improve the limited connectivity and interoperation of railway infrastructure in the
ECOWAS zone.
The railway tracks in terms of density and quality are significantly poorer and still operated at low standards in the
region in comparison to the international ones. Their track consists of light rails that are not adequate with modern
requirements. The railway networks are more than a century old, which characterized the obsolescence state of their
track materials. All the original railway track of the selected countries in the region are ballasted tracked, however,
they are not adequately or frequently maintained. The operational performance of ECOWAS railways has been
seriously influenced due to the combined effect of aging infrastructure, tamping and deficient maintenance. First of all,
the maximum axle load that the railway structure can bear is roughly 17 tonnes in countries such as the Benin
Republic [18], Burkina Faso [19], Ivory Coast [19], Mali [19], Ghana [20] and Senegal [19]; while in Nigeria, it is 20
tonnes [21]. In contrast to European countries, a standard axle load greater than 25 tonnes is seen as standard [22].
The maximum speed achieved in these networks is thus influenced by the poor condition of the track. By 2010, an
average speed reached by the freight transport is 36 km/h in Benin Republic [18], 31.8 km/h in Burkina Faso and
Ivory Coast [23, 24], 20 to 50km/h in Mali and Senegal [23], 45km/h in Ghana [20], and 65 km/h in Nigeria [25].
There is no available data about the axle load and speed in Guinea, Liberia, and Sierra Leone.
All the networks are almost single track with the exception of the first 30.6 km of the Western line (Takoradi-Manso)
in Ghana, and 70 km double track of Dakar-Thies line in Senegal. None of the existing networks of the selected
countries are electrified which explained the use of diesel locomotives [26].
Figure 2 Track gauges operated in the selected countries of ECOWAS
3.2. Rolling stock

In West Africa, the current condition of rolling stock has resulted in a decrease in the operational performance of the
railway. In the Benin Republic, the stability and availability of rolling stock at less than 50% characterized the current
scarce and poor condition [27]. When 50.18% of wagons were considered to be functioning, 35% of passenger
13
coaches were fit for use [28]. As for locomotives, all of them were more than 30 years old with 10.52% having a
reduction in their capacity` and the other percentage needing urgent repairs [27]. A field investigation by the
Japan International Cooperation Agency JICA [23] in Ivory Coast, Burkina Faso, Mali, and Senegal showed that the
rolling stock is equipped with facilities that are in poor condition and far from what is required for efficient
maintenance.
The fleet of locomotives in Ghana is aging with 42.8 percent between 11 and 13 years old, 16.3 percent between 20
and 24 years old, 36 percent between 28 and 31 years old and 4.9 percent over 40 years old [20]. There is no
standardization in the current locomotive fleet, of which only 63.9 percent can be used. Fifty percent of wagons are
available and fit for the transport of mineral resources, while in the case of passenger coaches, most of them were
purchased in the late 1980s, with 44 percent that has been rehabilitated in the last 15 years [20]. In Nigeria, a look at
the current capacity of infrastructure showed a severe dearth as it is not extremely sufficient. The locomotives, rolling
stock and other facilities have become obsolete and extremely unbalanced with the increase of population [17]. For
now, the corporation had just 50% of locomotives, 40% of freight wagons, and 47% of passenger vehicles operating
for a total population of 177 million. This poor level of locomotives and rolling stock explains the poor-quality services
provided by the Nigerian Railway Corporation, which led to the loss of patronage of both passengers and freight to
road transport. As for Togo, many problems occur in the development of its railway since the rolling stock has
deteriorated and insufficient. Additionally, the maintenance facilities of the depot/workshop are old type and
deteriorated [29]. No specific data has been found regarding the current rolling stock in Guinea, Liberia and Sierra
Leone.
3.3. Operational performance

Low density has characterized the traffic in ECOWAS. The freight services are the main dominant traffic and represent
90 percent of the movement in the economic community. This situation can be explained by the fact that the costs
through passenger operations with poor track conditions and lack of subsidies are difficult to recuperate. The lower
volume of traffic in the region is due to the emerging effect of low axle load and restricted speed on most of the
deteriorated existing lines.
Figure 3 Rail traffic density in the selected countries of ECOWAS
In fact, many sections have been dilapidated or closed because of poor maintenance, civil wars, or natural disasters. In
the Benin Republic, the lines of Cotonou-Ouidah, Pahou-Sègbohouè, Cotonou-Pobè, viz Porto-Novo and Abomey-
Zagnanado via Bohicon are out of service [27]. Meanwhile, passenger traffic has been stopped since 2007. In Burkina
Faso, the 104-kilometer section between Ouagadougou and Kaya are out of service since 2003 [19, 30]. As for Ghana,
only Accra to Tema, Accra to Kotoku, Awaso to Dunkwa and south to Takoradi lines are still operating out of the total
railway network [20]. In Mali, the Bamako-koulikoro section, which is 53 km long, is no longer used because of the
state of dilapidated infrastructure [19]. The trains are currently out of service in the sections between Thies-Saint-
Louis, Guinguineo-Kaolack, and Diourbel-Touba in Senegal [31]. In the case of Togo, some sections of the railway have
been stopped since 1987; the Lomé-Kpalimé line (117km) has not been used since 1996; Lomé-Blitta (281km) and
Lomé-Aného lines (45 km) have not been in service since February 2012 [29]. The passenger transport has been
stopped for more than fifteen years in Togo [32]. In Liberia, most sections of the rail network were disconnected
during the civil wars. By 2010, the Bong mine railway is the only operational line, while the Lamco Railway was put
back into service in 2011 after partially rebuilt. In Sierra Leone, the Makeni branch and the Kenema- Pendembu
14
section were successively closed in 1968 and 1971. Additional shutdowns prior to Bo happened in 1973, and the
railway eventually closed totally in 1974.
A tolerable and reasonable operational railway is weighted through the measurement of traffic volume, which shows
the utilization of infrastructure. The competitiveness between railway transports with its counterpart transport
modes depends on the large volume of traffic carried, which will generate huge funds for the maintenance and
upgrade of its infrastructure. A study on the railway traffic in the selected countries in ECOWAS (Figure 3) showed
that all countries have a density below 1 million traffic units (passenger-km plus tonne-km) per route-km, except
Guinea where the railway lines are considerably used by the private mining companies. The data was not available in
the case of Liberia and Sierra Leone.
Labour productivity is considerably lower in most countries in ECOWAS because of the low volumes reached (Figure
4). Under these situations, little traffic units are covered by fixed costs, which are expensive in contrast to other
transportation modes. This leads in reciprocity to a violent cycle where lower demand results on higher unit costs,
which impacts the unit costs and return. While an average employee in Burkina Faso produces approximately 481,
000 traffic units per year, Nigeria produces the lowest traffic units at 37, 200. No data was found for Togo, Ivory Coast,
Guinea, Liberia, and Sierra Leone.
The safety on railways tracks in selected countries of the ECOWAS has been a subject of considerable concern.
According to Bullock [4], there is the availability of little data regarding safety records in addition to uncertain
standards and compliance. The safety of the railways is the most revealed feature during most of the workshops in
ECOWAS because of the main issues such as derailments and thievery of infrastructure encountered in the region.
Figure 4 Labour productivity in the selected countries of ECOWAS
Based on the statistics of the international union of railways (UIC), the safety records in the selected countries of
ECOWAS are considerably disagreeable in comparison to other countries [14]. Representative safety records of Ivory
Coast and Nigeria in 2002 and 2003 show that they have a higher rate of accident in comparison to developed
countries such as Great Britain (see Table 1). With 0.402 accidents for every million traffic units, the accident rate of
Nigeria is 402 times greater than Great Britain (0.001 accidents per million traffic units), and almost twice than Ivory
Coast.
Table 1 Safety records of sample ECOWAS countries compared to Great Britain [14]
Country Traffic Units (million) Accidents Accidents per Million Traffic Units
Ivory Coast 436 91 0.209
Nigeria 268 108 0.402
Great Britain 68, 912 71 0.001
15
3.4. Telecommunication and signaling

Manual systems mostly depend on telecommunication and signaling networks through manual train’s orders or
mechanical signals [4]. The use of manual systems is proved to be sufficient because of the lower traffic volume on
existing networks. However, safety problems occur through human error.
In the Benin Republic, most of the existing telecommunication network dates back to 1960/70s and is completely
outdated and cannot be rehabilitated, which results in bad communication within yards and stations. The
management of the trains dependent on external services and does not allow a rigorous application of all the safety
rules due to the use of grouped GSM for station-to-station calls in certain areas [18, 33]. In Burkina Faso and Ivory
Coast, the railway system is non-electrified with no interlocking system between turnouts and traffic lights. The safety
of the railway is uncertain due to the manual operation of turnouts, and obsolete signaling and telecommunications
resulting in speed restriction over long sections [23, 34]. The mechanical and color light are the main train control
signals in Ghana. Communication is done through telephone radio and microwave systems [20]. However, they did not
provide adequate communication capabilities for the train. All radios systems have been defective due to problems of
frequency allocation and interference with third-party frequencies. In addition, the equipment for the signaling
department is out of date and necessitates spare parts which have been very difficult to procure or very expensive in
terms of foreign exchange [35]. As for Nigeria, the functioned mechanical semaphore signals and electric signals are
installed on the telecommunications lines. However, the signaling system is not being used at all and radio is often
used for communication, despite the old state of the facilities [36]. In Mali, the telecommunications system is obsolete,
and the starting signals are probably operated with wires and may not be perceptible at night. The coverage of private
operators of GSM telephony is not sufficient and does not allow to ensure regular traffic in safety [37]. The manual
quartering system is employed as a means of controlling train movements within the railway system. The
communications are ensured by the exchange of telephone dispatches from the station to the station [38]. The
circulation of the trains is ensured by exchanges of dispatches via the fixed and GSM telephony of the public operator
SONATEL-ORANGE in Senegal [19]. The BAL light signaling, which is installed on the heavy traffic section between
Dakar and Thies is no longer in operation [23]. There was not any system for train operation in Togo. Drivers, station
staff and security guards used dedicated mobile phones (green line) at level crossings for safety instead [29]. In
countries such as Guinea, Liberia, and Sierra Leone, the manual systems are mostly through manual trains’ orders or
mechanical signals [4].
4. Transport policy
Governments of the region are still grappling with the old policies decided by the colonial powers rather than having a
break [39]. In the Benin Republic, the texts of the OCBN have seen to be obsolete and do not allow effective and
modern management. The national transport sector plan has favour road transport mode over its rail counterpart.
This undoubtedly explained why more funds are given to road transport instead of the railway because of its
sustainability and immediate operation [40]. In Burkina Faso, Mali, Guinea, Liberia, and Sierra Leone, the main lines
drawn in the national policy are more related to the road sector over rail transportation. This is felt through the
number of pages devoted to road transport. In Ivory Coast, the authorities have resolutely oriented their efforts
towards the construction of the standard gauge track for all new constructions. However, the question of the renewal
of railway equipment which is in a state of advanced degradation is ignored. In Ghana, the irrelevant policies and
strategies generate inactivity in the execution of sub-regional and regional agreements which impede the sustainable
development of the railway transport sector. Additionally, the public-private partnership (PPP) was not encouraged
by the political leaders in the development and operation of the transport infrastructure. These explain the fact that
the liberalization and privatization in rail sector are still in its earlier stage [35]. In Nigeria, the failures to rescue the
railway system were more related to faulty policy design and government lack of self-determination rather than
scarcity of funds [21]. While Odeleye [41] accused the government of its complete holding, Aderamo [42] and Oye
Abioye [17] identified the decreased capacity and profitability of the railway to be related to discontinuity and
incoherence in policy implementation. In 2009, an institutional study in Senegal showed a relatively large number of
diverse actors with diffuse and poor skills. This situation has weakened the supervision of the state and reduces the
effectiveness of the management and control mechanisms of the public rail transport service. In Togo, with the lines to
increase considerably the railway traffic by 2020, the government has planned in its railway development policy, the
rehabilitation and construction of the Lomé-Cinkassé line and the rehabilitation of the existing network by 2014 [29].
However, the execution of these projects is not yet materialized. Above all, there is a weak coordination of policies and
regulations applicable to cross border transport, as many West African countries have not fully enacting consensus at
promoting cross border movement by rail from which each country could take advantage. Additionally, the
accessibility of master plans at the national level and the lack of digitalization of all documents for basic access were
accentuated.
16
5. Regulation and institutional framework

The setting of a fundamental basis for railway development is depending on a strong regulatory and institutional
framework that preserves its investment. The formulation and implementation of more appropriate policies and
regulations depending on the degree of closeness government participants have in the structure of the railway
industry. Regulation and legal system were set in some countries to observe safety concerns, corruption, and market
performance. The Act 1955 which established the Nigerian Railway Corporation is an example. But, it needed to be
revoked in order to allow sufficient competent and robust system, market participation, and competition
development. In Ghana, the Railway Act 2008, which was ratified by the President’s Office on January 6, 2009, has
established the Ghana Railway Development Authority as a statutory body in charge of the development of the railway
and promotion of railway services, management and operation of railway assets as well as construction and
improvement of urban railways. In other countries like the Benin Republic where railways are public enterprises, no
regulatory body exists.
In spite of the theoretical role played by the Ministry of Transport (MOT) on supervising railways, the responsibility of
examining has been assigned by some governments to concessionaire (as in the case of Senegal, Mali, Ivory Coast,
Burkina Faso, and Togo). The representatives from the government in the concessionaire board are so often from the
Ministry of Finance rather than from MOT as they mostly agonize from appropriate financial aid. The participants of
the government involved in railways have so often lacked sufficient knowledge while well informed of the sector have
limited power to make any decisions. Additionally, the participants were familiar with the heritage of the public
enterprises instead of modern railway management, which leads to their resistance to dynamic change. There is no
specific data on the regulation and institutional framework regarding Liberia, Sierra Leone, and Guinea.
6. The socio-economic context of ECOWAS

The West Africa sub-region has been widely recognized for having commercial potential. The region has a market of
more than 300 million consumers that projected to be half a billion within the next 35 years, contributing directly to
the dynamics of the global economy [43]. In addition, its geographic area is characterized by immeasurable potentials
minerals and agricultural crops [44-46].
The member countries of the Economic Community of West African States (ECOWAS) have proved to have unequal
distribution of natural resources. Prior to the attainment of their independence in the region, numerous countries
were endowed with one or other forms of mineral resources. While Ghana and Nigeria are well known for gold and
crude oil respectively, Ivory Coast, Liberia, and Sierra Leone are known for being rich in the diamond. Iron ore and
bauxite are also contributing considerably in countries like Liberia, Guinea, and Sierra Leone. Other ECOWAS
members including Senegal, Mali, and Togo are rich in phosphate. With the exception of Nigeria which produces crude
oil since the 1960s, it's finding in West Africa is of latter development, with Ghana, Ivory Coast, Liberia, and Sierra
Leone lately appearing as oil-rich economies.
Most of the countries are known to be largely dependent on rich agricultural potential for the income of their
population before the discovery of mineral resources in the region. West African countries are major producers of
cocoa, rubber, cotton, and timber. For instance, while Ghana; Nigeria and Ivory Coast have great potentials for coffee
and cocoa plantations, Liberia is known for its large rubber plantation. Also, while Mali is noted for cotton, livestock,
millet and rice, Senegal and Burkina Faso have great potential for groundnut, cotton and sorghum production. In
Burkina Faso, peanuts and she nut total 80 percent while in the Benin Republic, the same percentage is attributed to
oil palm products. Most of the countries along the coast of West Africa, including Senegal, Ivory Coast, Guinea, Sierra
Leone, Liberia, Ghana, Togo, Benin, and Nigeria are known to have fishing as a major activity.
7. Recent railway projects

Many projects have been recently planned in different countries with China, to build new lines or upgrade existing
deteriorated infrastructures [16]. However, the sustainability of these projects showed some considerable
apprehension in terms of the economic and technical point of view.
From an economic standpoint, worries have been intensified about the huge loans resulting in the dependence of the
developing countries with foreign governments. A reduction of Sri-Lanka debt through the supervision of its deep
seaports by China and the high cost-effectiveness of the Mombasa-Nairobi line project in Kenya are such examples of
this concern [47, 48].
17
With regard to the technical perspective, the development of the railway transportation system has only considered
the national level, with little attention to connectivity and interoperability within the region [11]. Divergence in track
structure (rail, ballast, sleeper, and fasteners), track gauge (narrow, metric, and standard), and inappropriate rolling
stocks are probably to be restricted the regional railway transportation [48]. Ethiopia is an example of a country
where lines run on separate signaling systems because of being constructed by different companies [49]. In Kenya,
identical apprehension is raised by stakeholders at the Nairobi workshop regarding the connection between new rail
lines and larger development regional planning, where signaling and communications, and maximum speed and axle
loads to be applied in independent lines are not standardized.
Furthermore, recent financing programs on the railway sector are appeared not to be oriented to the current gaps in
potentiality. Foreign governments and international organizations are the main decision-makers on the technical part
of new projects where regional plans with comprehensive viewpoints are ignored. This explained the fact that
financing railway projects are incompatible with regional requirements and development.
A comprehensive inconsistency exists between present actions and future targets. Since, the development of an
integrated region where the populations enjoys free mobility, access and engage in businesses while living in a secure
and peaceful environment is the goal of ECOWAS, and view the fact there is concurrently a lack of strategy to guide
regional development towards a common target, head of governments and states are involved in buying ready-made
technology that will be too exorbitant in the projection of nearby traffic. By doing so, their route towards development
is expected to only pursue the same footsteps of advanced countries over the past and stay obsolete everlasting. There
is no consideration regarding the precise condition of the region and its exigency for acceptable solutions that can
upsurge antecedent curvy developments and produce appropriate solutions. If no solution has been found for these
problems, there will have a considerable possibility for hardy accessible financial resources to be misused, and for
new infrastructure systems to less perform. Such a result is expected to reduce the development of railways in the
ECOWAS community and cause problems related to efficiency and environment in the long-term. Accordingly, it is
crucial to make a steady and reasonable of hardly accessible investments in order to achieve regional future goals.
8. Regional strategy development for railway transport

The perspectives of a regional-large railway network are earlier mentioned to ease interoperability as well as to cope
with the mobility needs of population growth. The international union of railways has published a specified vision of
the African railways by paying particular attention to the revitalization of its networks, accentuating the necessity of
considering regional and continental transport networks [14].
This study recommends some strategies to be used in the development of an efficient and interoperable railway
transportation system in the Economic Community of the West African States which in turn will improve the
interconnectivity and enhance the economic growth and trade in West Africa.
8.1. Strategies for comprehensive railway transport system database

A comprehensive data on the state of the infrastructure, rolling stock and operational performance of each country’s
railway is decisive for the growth of more practical strategy and structured networks. There was an absence of many
indexes in the existing railway infrastructure and trains operation in many countries such as Liberia, Sierra Leone, and
Guinea. Countries where there are available indicators; they dated back to more than a decade ago producing
imprecision in these countries for promoting economic growth. Additionally, new projects financed by international
organizations or foreign countries are lack of accessible technical data.
Although data collection is a temporary prerogative for the establishment of any strategy, its procedure should be
methodical in pinpointing critical parameters that can show if potential development coincides with future
perception. Accordingly, an inventory of indicators to be computed as well as the criterion to be implemented and
norm assemblage operations to secure trustworthiness in the results must be undertaken within the ECOWAS.
Database and technique must be set up for recording the transport services and contentment of passengers and users
respectively. Information systems and databases can be initiated for the yearly checking of the services, volume, and
costs of railway passengers on each trajectory.
By implementing this strategy, the level of each country will be easily identified so that countries of the region will
appropriately act. Additionally, the lacuna between present-day performance and each benchmark can be specified
through indexes, so that prerogative sectors for research and development (R&D) will be emphasized.
18
8.2. Strategies for capacity building and training centers

The comparatively poor awareness of the railway sector has been one of the major issues presently hindering the
railway development and influencing all the decision-makers and managers at different positions of the railway
sector. Besides that, the revamping of supervisory expertise and productivity of operations is being compelled by a
powerful patrimony and lifetime. The railway sector is identified by mainly an absence of constancy in railway
schemes for the last three decades. This absence implies limited technical arrangements and ceasing of existing
railway institutes. Therefore, establishing a regional well-known curriculum in railway technology; producing
knowledge exchange and skill-building for future railway management positions and policy decisions makers;
assisting technically government authorities in the planning, preparation and analysis of project; and developing
training centers that produce resident labor force to employ the technologies and evolve new ones are the key drivers
for promoting regional capacity building and training centers. If these strategies are implemented, they can improve
the proficiency of the ECOWAS countries, and allow them not to be dependent on advanced countries for making
decisions for the time ahead. Also, the railway system in the region will enlarge the attraction of private companies or
investors in the sector and ameliorate lucidity in taking decisions. Finally, the regional institutional framework will
thus be enhanced allowing a strong base of any additional action from either public or private enterprises.
8.3. Strategies for railway infrastructure development and transport policy

The governments must proceed with durable strategic programs for the rehabilitation of existing networks. Necessary
measures are extremely recommended to ensure the survival of all railways networks within ECOWAS by focusing on
the main axes that allow clearing a good part of their accounts payable in order to be able to resume a correct supply
of spare parts for locomotives, wagons and railway assets. Sections of the railway network where traffic has suffered
the most should have their productivity overhaul through the revamping and restoration of the track, which will, in
turn, improve the interoperability with suitable technologies and cost-effective lines. The interconnection and
interoperability of railway networks require, above all, the setting of an administrative, legal, technical and financial
framework capable of covering the divergent interests of users, private companies, local authorities, and nations at the
regional level. Consequently, several constraints will have to be lifted in order to expect the achievement of these
major objectives.
The construction and maintenance standards of railways should be harmonized. For instance, the adoption of a 60 kg
rail to support at least axle loads of 22 tons is recommended. Meanwhile, the consideration of future high
transportation demand must be taken into account through careful consideration of gauge (standard gauge), track
density and specified geometries.
The equipment fleet along with the maintenance standards should be harmonized by governments within the region.
It is obvious that a good availability and an excellent rate of reliability of the rolling stock park can compensate for the
financial charges. The acquisition of railway infrastructure assets, as well as the rolling stock, should be globalized by
the governments. In this respect, railway networks of the ECOWAS members must seek to use the same type of rails
and locomotives to make their orders representative and attractive to the international rail industry in terms of
economy of scale.
An establishment of a good environment for a workable and more reasonable railway system conforms to the
development of transport policy which did not only favor road transport should be suggested. Additionally, equal
funding of road and rail transport infrastructures as well as intermodal transportation systems should be taken into
account in the global vision of ECOWAS countries. At the national level, governments should not only enact a policy
and management strategy for the separation of the operations and infrastructure for clear transparency but also
should encourage the intervention of the private sector within the framework of a public-private partnership in order
to have a good environment for investments.
9. Conclusion
A review and analysis of the railway transportation system in the Economic Community of West African States,
including four English speaking countries and seven French-speaking countries, is conducted in this paper. The review
of the current situation of the regional railway transportation system certifies the global vision of its assets inherited
from colonial powers with few restorations since then and single-track lines with light rails, low axles loads and
restricted speeds.
19
Notwithstanding the huge investments for the construction of new railways lines as well as the rehabilitation of
existing ones and the restriction of financial resources, most of the West African countries have decided to modernize
their railway transportation system to meet international standards. However, their action plans appeared not to be in
line with the global vision of regional-large integrated railways systems that can be able to challenge other
transportation modes. As a consequence, the interoperability and interconnectivity problems still persist as it was in
the case of an inherited existing infrastructure system. There is thus a necessity of additional study to propose some
practical strategies as bellows.
 Develop a comprehensive information system and database on the state of railway transportation in each
country of ECOWAS in order to provide a list of indexes, metrics, and norms to be measured, used and
collected respectively.
 Establish a regional well-known curriculum in railway technology and produce skill-building capacity for
future railway management positions and policy decision-makers
 Develop training centers for the local workforce in order to promote regional capacity building.
 Develop a stable regional master plan for the development of railway infrastructure through a durable
strategic programs for the rehabilitation of existing networks, the harmonization of construction and
maintenance regulations of railway, the harmonization of equipment fleet along with the maintenance
standards, and the setting of administrative, legal, technical and financial framework able to cover the
divergent interests of all the stakeholders.
 Provide a policy and management strategy through the separation of the operations and infrastructure of the
railway transportation system for clear transparency especially in terms of accounting.
 Encourage the intervention of the private sector within the framework of a public-private partnership in
order to have a good environment for investments.
Compliance with ethical standards
Acknowledgments
The authors especially thank all the contributions of all survey participants. This research was funded by the National
Science Foundation (No. 51778541).
Disclosure of conflict of interest

The authors declare no conflict of interest.
References
[1] A Union. (2015). Agenda 2063: the Africa we want, African Union Commission.
[2] EÓ Broin and C Guivarch. (2017). Transport infrastructure costs in low-carbon pathways, Transportation
Research Part D: Transport and Environment, 55, 389-403.
[3] AfDB. (2015). Rail Infrastructure in Africa: Financing Policy Options, African Development Bank, Abidjan, Cote
d’Ivoire.
[4] R Bullock. (2009). Off Track: Sub-Saharan African Railways, Africa Infrastructure Country Diagnostic
Background Paper 17.
[5] K Gwilliam. (2011). Africa's transport infrastructure: Mainstreaming maintenance and management, The World
Bank.
[6] L Seba Minsili, MJ Kisito, T Gilbert, N Jean, KA Gadam and DP Christian. (2017). Requirements to the
modernization of African railway networks: The standard gauge versus the metric gauge. International Journal
of Civil Engineering and Technology 8, 925-941.
[7] M Lowe. (2014). Rail revival in Africa? The impact of privatization, Joint RES-SPR Conference on
Macroeconomic Challenges FacingLow-Income Countries: New Perspectives.
[8] A Yeboah. (2017). Taking off from where we left off, Daily Graphic.
20
[9] V Sarma and S Sinha. (2016). Indian Railways: On the fast track to growth. McKinsey & Company.
[10] BM Bayane and Q Yanjun. (2017). Transport infrastructure development in China. Journal of Sustainable
Development of Transport and Logistics, 2(1), 29-39.
[11] VN Olievschi. (2013). Rail transport: Framework for improving railway sector performance in Sub-Saharan
Africa.
[12] A Union. (2009). Declaration on the development of transport and energy infrastructure in Africa. Declaration.
[13] M Blumenfeld, W Wemakor, L Azzouz and C Roberts. (2019). Developing a New Technical Strategy for Rail
Infrastructure in Low-Income Countries in Sub-Saharan Africa and South Asia, Sustainability, 11(16), 4319.
[14] UIC. (2013). Railway Revitalization in Africa-Destination, International Union of Railways, Paris, France.
[15] M Ncube and C Leyeka Lufumpa. (2017). Infrastructure in Africa: Lessons for future development. Policy Press.
[16] A Habiyaremye. (2016). Is Sino-African trade exacerbating resource dependence in Africa? Structural Change
and Economic Dynamics, 37, 1-12.
[17] O Abioye, K Shubber and J Koenigsberger. (2016). Evaluating the role and impact of railway transport in the
Nigerian economy, options and choices: Case of Nigerian Railway Corporation, AshEse Journal of Economics,
2(4), 103-13.
[18] J Adiko. (2009). La relance du ferroviaire béninois, Université Libre de Bruxelles.
[19] UEOMOA. (2011). Union Economique et Monétaire Ouest Africaine. Politique sectorielles: transport
ferroviaires. Etude pour l’elaboration d’une strategie de developpement du transport ferroviaires dans l’espace
UEMOA. Rapport final (Volume 1), 181.
[20] GRDA. (2013). Ghana Railway Master Plan-Annex-2: Environmental and Social Management Framework, Final
report, 134.
[21] AF Akwara, JE Udaw and GE Ezirim. (2014). Adapting Colonial Legacy to Modernism: A Focus on Rail Transport
Development in Nigeria. Mediterranean Journal of Social Sciences, 5(6), 465.
[22] E Union. (2004). Commission Decision 2004/446/EC of 29 April, European Union, Brussels, Belgium.
[23] JICA. (2010). The study on regional freight corridor development based on port and rail system in Sub-Saharan
Africa. Chapitre 5, present situation of interregional freight transport infrastructure in West Africa, Final report,
in: L.T.E.C. Japanese International Cooperation Agency. PADECO Co, Inc (Ed.) Fourth Tokyo International
Conference on African Development, 116.
[24] KA Brou. (2015). L'EXPLOITATION COLONIALE DANS LA MISE EN PLACE DU RÉSEAU ROUTIER ET
FERROVIAIRE DE LA COLONIE DE CÔTE D'IVOIRE DE 1893 A 1960.
[25] A Abubakar. (2006). Nigerian Railways: Past, present and future perspectives. Rail News, 4(6), 4.
[26] R. Carruthers, RR Krishnamani and S Murray. (2009). Improving connectivity: investing in transport
infrastructure in Sub-Saharan Africa, AICD Background Paper 7, 6.
[27] MB Bouraima and Y Qiu. (2018). Toward Innovative Solutions for Revitalizing Benin Republic Railway
Transportation System. International Conference on Transportation and Development: Traffic and Freight
Operations and Rail and Public Transit, American Society of Civil Engineers Reston, VA, 238-246.
[28] INSAE. (2016). Annuaire Statistique, 49.
[29] JICA. (2013). The Project for The Study on Togo Logistics Corridor Development in The Republic of Togo.
Chapter 3 Current Transport System in Togo. Final report, in: l. Japan international cooperation agency (jica)
central consultant inc. Yachiyo engineering co. (ed.) Ministry of public works and transports, the republic of
togo.
[30] P Cosaert. (1990). Les voies de communication au Burkina Faso, Les cahiers d'outre-mer, 43(169), 53-75.
[31] FMA Samb. (2010). Transport international de marchandises par voie ferroviaires (cas de Transrail SA),
transport et supply chain, centre trainmar/Dakar (ULCO) DUNKERQUE.
[32] BAD. (2014). Le secteur des transports au Togo, Département des Transports, du Développement Urbain et des
TIC.
21
[33] G Olivier. (2011). Contribution à l'amélioration de la situation financière d'une entreprise d'état: cas de l'OCBN
(Organisation Commune Bénin-Niger des chemins de fer et des transports), Université d'Abomey Calavi au
Bénin.
[34] R Bullock. (2009). Africa Infrastructure Country Diagnostic: Railways in Sub-Saharan Africa, Unpublished
manuscript. World Bank, Washington, DC.
[35] RD Dinye. (2012). Prospects and issues of railway infrastructure and development in Ghana. Journal of
Construction Project Management and Innovation, 2(1), 259-284.
[36] JICA. (2014). The Second Data Collection Survey on Mass Rapid Transit in the Federal Republic of Nigeria,
Federal Republic of Nigeria.
[37] BAD. (2014). Mali, Note sur le secteur des transports, Département des Transports, du Développement Urbain
et des TIC.
[38] S Consult. (2003). Evaluation environnementale et sociale du secteur transport ferroviaires international sur
l’axe Dakar-Bamako et sa mise en concession. E807, Volume 1, Ministere de l’infrastructure de l’equipement et
des transports, rapport final, deuxieme projet sectorielle des transports, 198.
[39] J Lombard and O Ninot. (2010). Connecter et intégrer: les territoires et les mutations des transports en Afrique,
Bulletin de l'Association de géographes français, (1), 69-86.
[40] MB Bouraima and CMD Dominique. (2018). Road Infrastructure and Economic Growth in Benin Republic.
Scholars Bulletin (Civil Engineering), 4(7), 680-688.
[41] JA Odeleye. (2000). Public-private participation to rescue railway development in Nigeria. Japan Railway and
Transport Review Journal, 23(42-49), 104-111.
[42] A Aderamo. (2011). Transport infrastructure and the Nigerian environment: a review, Int Rev Business Soc Sci,
1, 49-66.
[43] JP Madeira and NA Monteiro. (2016). Cape Verde in West Africa region: towards a new paradigm for regional
integration.
[44] M Jalloh. (2013). Natural resources endowment and economic growth: The West African Experience. Journal of
Natural Resources and Development, 3, 66-84.
[45] L Weng, AK Boedhihartono, PH Dirks, J Dixon, MI Lubis, JA Sayer. (2013). Mineral industries, growth corridors
and agricultural development in Africa. Global Food Security, 2(3), 195-202.
[46] V Markwitz, KA Hein and J Miller. (2016). Compilation of West African mineral deposits: Spatial distribution
and mineral endowment. Precambrian Research, 274, 61-81.
[47] P Mourdoukoutas. (2018). China Is Doing The Same Things To Sri Lanka That Great Britain Did To China After
The Opium Wars.
[48] M Rahmatullah. (2006). Transport issues and integration in South Asia, Promoting Economic Cooperation in
South Asia, 174.
[49] M Delelegn. (2018). The hottest projects in Ethiopia and what is being done to open for investments, In
Proceedings of the East Africa Rail Conference, Nairobi, Kenya.
How to cite this article

Bouraima MB, Qiu Y and Yusupov B. (2020). A review and analysis of railway transportation system in the economic
community of West African States: Towards the development of sustainable regional goal. Global Journal of
Engineering and Technology Advances, 2(2), 11-22.
22
ISSN 2278-3091
Volume
Jihane LARIOUI et al., International Journal of Advanced 9 No.2,
Trends MarchScience
in Computer -Apriland
2020
Engineering, 9(2), March - April 2020, 2471 – 2478
International Journal of Advanced Trends in Computer Science and Engineering
Available Online at http://www.warse.org/IJATCSE/static/pdf/file/ijatcse236922020.pdf
https://doi.org/10.30534/ijatcse/2020/236922020
Towards a Semantic Layer Design for an Advanced

Intelligent Multimodal Transportation System
Jihane LARIOUI 1, Abdeltif EL BYED 2
Laboratory of computer science and modelling of decision support systems
HASSAN 2 University Faculty of Science Ain Chock
Casablanca, Morocco
1
jihane.larioui-etu@etu.univh2c.ma
² abdeltif.elbyed@univh2c.ma
develop an intelligent solution that will facilitate

ABSTRACT decision-making and route planning. Moreover, a semantic
approach can be effective in overcoming, sharing knowledge
Over time, intelligent transport systems (ITS) have shown and communication between the different agents of our
their relevance and have become essential for good urban system.
mobility management within smart cities. They facilitate In multimodal transport, users use at least two different
traffic management and make flow more fluid and types of transport to reach their destination. Therefore,
autonomous. Indeed, the data manipulated in the ITS is large multimodality refers to the optimal use of different modes of
and diversified which makes the information less exploitable transport. Hence, the need for a system providing real-time
and leads us to a considerable loss of time before obtaining the
information on departures, routes and traffic conditions
desired information. In order to manage these different
before and during the travel. In such a situation, the passenger
sources of information, their semantic conflicts and ensure
must manage a variety of information: different modes of
interoperability between all these sources involved in the
multi-modal transport network, we decided to integrate all transport that can be combined (bus, train, metro, etc.), travel
this information in a way that it is unified, flexible and time, cost, connection points, different operators, and so do
understandable by all the agents of our multimodal we. In order to manage these different sources of information,
information system. Thus, the objective of this semantic their semantic conflicts and ensure interoperability between
approach is to localize the different resources of the WEB, to all these sources involved in the multi-modal transport
standardize the description models so that they are usable by network, we decided to integrate all this information in a way
the different agents of the system. In this article, we propose a that it is unified, flexible and understandable by all the agents
new approach for the development of an architecture based on of our multimodal information system.
Multi-Agent systems (SMA) coupled with semantic Web and Usually, most research contributions consider a
this in order to help decision-making and improve the multi-agent system to improve route planning in a
management in the context of urban mobility. The system to multimodal network, aside from that, they have not
develop covers all modes of transport and capable of considered using the Semantic Web to allow flexible querying
managing planned trips as well as answering questions information.
related to safety during travel. The global purpose of the advanced multimodal
information system is to propose the optimal route that
Key words: Multi-agent System (SMA), Intelligent transport
corresponds to the different criteria of preferences chosen by
systems (ITS), Ontology, Semantic Web, Multi-modal
the passenger according to their needs in terms of: travel time,
Transportation
number of mode changes, cost and safety. This system is
1. INTRODUCTION based on a multi-agent system architecture coupled with
semantic layer approach for a better management of the
Nowadays, more attention is given to urban mobility due to multimodal transport network. The purpose of the Semantic
the increase in population density, traffic congestion, lack of layer is to characterize existing information in order to
information and all the problems encountered by the transport facilitate the automation of services, to discover, link and
network. The complexity of traveling in a multimodal infer similar knowledge. Its architecture is based on a
network increases over time due to the lack of accurate hierarchy that aims to represent knowledge on the web by
knowledge and necessary information. Hence the need to satisfying the criteria of standardization, interoperability and
flexibility.
2471
Jihane LARIOUI et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(2), March - April 2020, 2471 – 2478
With this in mind, the system to develop is based on an (Analytical Hierarchical Process), which consists of defining
ontology of public transport and should be capable of weights to calculate the impedance. The related works
reasoning. In this article, we detail the semantic layer of our mentioned above give an idea of a domain-based transport
system that acts as middleware between the physical layer and ontology to manage the planning of trips between two sites.
the information layer. In addition, the semantic layer is used Related work only offers systems to meet travel needs.
to efficiently integrate data and information into our system, However, they do not take into account user preferences.
also reduce uncertainty in the decision-making process and In all these related works, no formal semantics describing all
resolve semantic conflicts resulting from cooperation between aspects relating to intelligent road transport systems have
different sources. been proposed. In this article, we present a transport system
The structure of this paper is as follows: Section 2 discusses ontology for a sensitive semantic transport system which
some related works and the role of the ontology in a covers all modes of transport and which is capable of
multi-agent system. The approach used for ontology managing travel planning as well as answering questions
development and design are stated in section 3. Section 4 related to transport cost, travel time, route, number of modes
describes the Implementation part and discusses the results. changes and safety, etc.
Finally, we conclude the document with some remarks and
2.2 The Role of the ontology in a multi-agent system
perspectives.
The word ontology first appeared in Aristotle's philosophical
2. BACKGROUND essays, where he described the nature and organization of
being. In artificial intelligence, an ontology is used to
2.1 Related Works represent a domain of knowledge.
In the literature, agent-based information systems are The definition of ontology as a semantic primitive is simply a
commonly referred to as Advanced Traveler Information hierarchical description of the important concepts in a
Systems (ATIS). The multi-agent system has been applied to domain, coupled with a description of each of these concepts
various problems related to the transport system. In [17]. The ontologies consist of several concepts that include
particular, planning personalized itineraries. Meeting the class, subclass, class hierarchy, instance, location, value,
needs and preferences of users has recently become a major default value, facet, type, cardinality, inheritance, variable
objective of ITS. and relation [18].
Wang, Ding and Jiang [4] have developed an ontology based The main functionalities of ontology are to share a generic
public transport query system to provide an efficient and understanding of the structure of information between human
timely information service. They opted for an improved and / or software agents, to allow the reuse of domain
public transport request algorithm. This algorithm takes into knowledge, to make domain hypotheses more precise, to
account both user needs and traffic conditions. They were create a semantics of interoperability between different
interested in solving the problem of the same station name in information resources, to provide a clear and exact analysis of
different places to find the shortest distance between two terms [7].
places using a radius search algorithm. Houda et al. [6] The objective of this work is to implement these advantages to
developed a domain based transportation system for set up a semantic layer to model the different sources of data,
passenger planning system. The system supports a to standardize their description models so that they are
multi-model journey pattern, which helps the user to find the comprehensible and usable by the various agents of the
most convenient path from one position to another with system.
several possible options. Choice of the user depends on Indeed, in a multi-agent system, agents are led to recognize
interesting journey patterns, shopping journey pattern, etc. and understand the different concepts describing the
The system is limited to the roadways and railways only. application of the domain. These concepts are explicitly
Otherwise, Zgaya et al. [14] have implemented a negotiation defined by ontologies and are associated to the agent with
protocol for the transport sector. This protocol is based on a their relationships and constraints [8]. Interoperability
flexible ontology to facilitate communication between agents. between agents is achieved by unifying these concepts using
The work of Niaraki and Kim [16] is focused on determining common ontologies that allow agents to cooperate while
an impedance model for the road geographic information retaining their autonomy. In Multi-agent systems, agents are
system and the intelligent transport system. Impedance is a linked to various information resources. These information
model that aims to calculate the amount of cost or resistance, must be effectively integrated into the system so that semantic
of the link between a node of origin and that of destination. conflicts are reduced. The most effective way to remedy this
For this, they defined an ontology of the road segment based problem is to set up a global ontology that will be used by all
on user preference criteria and environmental context agents. Agents sharing the same ontology can exchange
criteria, and from this ontology, they created a hierarchical knowledge and communicate fluently. In fact, their
structure divided in two: a part linked to the criteria of users knowledge representations are compatible with respect to the
(such as information on tourist attractions and preferences) concepts considered relevant and with respect to the names
and the other to contextual criteria (such as weather and given to these concepts [4]. In this way, the decision-making
safety). This multi-criteria study is carried out using the AHP process will be established in better conditions.
2472
3. SEMANTIC LAYER FOR ITS communicate and coordinate between the different agents and
a way to improve decision-making process. Previous work for
3.1. Global Architecture the development of an advanced and intelligent multimodal
Faced with the complexity of the data manipulated in the ITS information system has been proposed to develop a new
systems which are large and diversified [21] , we have architecture based on multi-agent systems. This architecture
proposed a novel approach, which combines between consists of a six layers which are IHM layer, Selection Layer,
Semantic Web technologies and multi-agents paradigm to Decision Making Layer, Information Layer, Semantic Layer
design an advanced multi-agent information system for and physical layer [18] and [21].(Fig.1)
multimodal transportation. It would be an effective way to
Figure 1: Multi-Agent Information System Architecture

The IHM layer is composed of the PTA “Personal (travel time, number of mode changes, cost and
Travel Agent” representing the Human Machine Interface. safety).
• The Selection Layer is composed of The DSA • The Information Layer is composed of the IA
“Directory Selecting Agent”, which defines the search “Information Agents”, which are responsible for searching,
domain by specifying the information agents that will work collecting, integrating and manipulating the information
together to plan the route and propose the different from the different sources of information.
combinations of possible information agents. • The Semantic layer, which uses semantics web
• The Decision-making layer is composed of three agents: technology to improve the flexibility between the different
o The SA “Sorting Agent”, which examines the agents of the system, in different terms; the semantic layer act
different routes proposed by the DSA and decides like a middleware between the physical layer and the
how to treat them according to the preferences of the information layer. In addition, the semantic layer is used to
users. efficiently integrate data and information in our system, also
o The DMA "Decision Making Agent" which is reduce the uncertainty in the decision-making process and to
based on the method TOPSIS “Technique for order resolve the semantic conflicts generated from the cooperation
of preference by similarity to ideal solution” as a between the different sources.
MCDM methodology to facilitate decision-making • The physical layer, which encapsulates multiple
and choose the itinerary that will satisfy the user's information resources, for example: the databases of the
preferences. different transport operators (Tramway, metro, bus…etc), the
transport network architecture, road ITS applications,
o The CA “Calculating Agent”, which takes up the Sensors, surveillance camera…etc. These systems produce
routes proposed by the DSA in order to calculate for information on urban networks and are essential when the
each route its necessary parameters. These passenger is moving because they provide multimodal
parameters are calculated based on user preferences information and facilitate the use of networks.
2473
3.2. Semantic Layer • The second phase: after having defined the local
Going back to the semantic layer, the global role of this layer ontologies, now, we can pull out the global ontology
consists of four main functionalities: The first concerns the from different defined concepts. This phase includes
connectivity to information resources in order to have access two stages: first, the analysis of local ontologies,
to the different data sources. This first step then allows the then the selection of all concepts and the resolution
extraction of the most relevant concepts and the elimination of semantic conflicts in order to define the global
of the miss understanding between these information [22]. ontology and its particular concepts.
Then comes the most important step, which allows generating • The third phase: it consists of defining the mapping
standard and unified models for these data sources and finally between global and local ontologies. The global
to have a flexible Data Model understandable by all the agents ontology is constructed from local ontologies using
of the system. (Fig.2) OWL Annotations.
Figure 2: Semantic Layer Process Description
The purpose of this semantic layer process is to formulate

all data and concepts contributing to the semantic definition
of the multimodal transport system and to standardize the
description models, so that they are comprehensible and
usable by all agents. Figure 3: Process of building Ontologies trough 3 steps
3.3. Ontology Design 4. IMPLEMENTATION AND EXPERIMENTAL

In this section, we present the ontology design methodology, RESULTS
which is based on a hybrid architecture proposing a bottom up In this section, we focus on the implementation part. First,
process to create global ontology from local ontologies and we detail the first phase, which concerns the construction of
define the relationships between them [19]. local ontologies and we present only the cases of two
The proposed solution avoids misunderstandings that can information resources, those of the bus and the tramway.
arise during the exchange and communications between The global ontology of the multimodal transport network
agents and aims to make them able to understand each other take up the different concepts and relationships of the local
when using this information processed by these ontologies. ontologies that we have created: Tramway Ontology and
We use Protégé 5.1.0 [13] to build our ontology as it is a free
Bus Ontology. The figures below respectively show the
open source editor and framework and it supports the latest
local ontologies built based on the data from Bus and
OWL 2 Web Ontology Language and RDF specifications
Tramway network. The ontologies created use the
from the World Wide Web Consortium.
modeling of the transport network for each mode by
As Fig.3 shows, the development of the ontology includes
three phases: illustrating major concepts such as lines, stations, operator,
• The first phase: consists of building local ontologies. vehicle ...etc. (Fig.4, Fig.5, Fig.6 and Fig.7). By analyzing
Foremost, it is essential to analyze the different these concepts, we notice that each ontology defines its
information resources, each independently. After concepts differently, for example, the concept Station
that, we define the concepts considered relevant for defined in the Tramway ontology is equivalent to Bus Stop
each ontology, their relationships and their usage in Bus ontology. In the other hand, we have concepts that
constraints. have the same definition, for example, Timetable or
Operator, also the notion of Vehicles and Lines.
2474
Jihane LARIOUI et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(2), March - April 2020, 2471 –
2478
Figure 4: Fragment of Bus Network local Ontology Figure 5: Fragment of Tramway Network local Ontology
Fig. 6. Bus Network local Ontology
Fig. 7. Tramway Network local Ontology
Now, we focus on the second phase that consists in such as the concept of Transport mode or even
pulling out the global ontology from different concepts Connection Links (Fig.8 and Fig.9).
defined in the local ontologies (Tramway network local
Moreover, in order to deal with the notion of Safety in
ontology / Bus Network local ontology).
the transport field, we defined areas in cities as a concept
After analyzing the local ontologies, we have determined and we decided to give them notes according to the
the global ontology using OWL Annotation. This Global degree of safety in this zone. In addition, the concept of
Ontology called Transportation Network takes up the Travel also joins other concepts of the Global ontology
concepts of the local ontologies and adds new concepts to and has as attributes: cost, duration, Start Point and
them in order to meet the need for the multimodal system End Point.
2475
Jihane LARIOUI et al., International Journal of Advanced Trends in Computer Science and Engineering, 9(2), March - April 2020, 2471 –
2478
Figure 8: Fragment of Transportation Network Global Ontology
Figure 9: Transportation Network Global Ontology
Finally, we present the third phase that consists in defining

the mapping between global and local ontologies using OWL
annotation proprieties. This latter allows classes, properties, Table 1: The five, pre-defined annotations in OWL.
individuals also the ontology itself, to be annotated with five,
Property Typing Purpose
pre-defined properties. These properties are illustrated via
the Protégé OWL Classes view in Fig.10 and presented in owl:versionInfo String OWL version
Table1. information
rdfs:label String Meaningful,
human-readable names
rdfs:comment String Additional information
rdfs:seeAlso URI Identify related

resources
rdfs:isDefinedBy URI Reference an ontology
Figure 10: OWL annotation properties illustrated via Protégé
2476
8. F. Giunchiglia and B. Dutta. Dera: A faceted knowledge

5. CONCLUSION organization framework. Technical report, University
of Trento, 2011.
In this article, we focus on the semantic layer of our 9. F. Giunchiglia, B. Dutta, and V. Maltese. From
agent-based multi-modal information system. The purpose of knowledge organization to knowledge representation.
this layer is to manage the different sources of data and to Technical report, Università di Trento, 2013.
build a flexible data model that can be understood by all the 10. S. Das and F. Giunchiglia. Geoetypes: Harmonizing
agents in the system. In order to deal with this problem, an diversity in geospatial data. In OTM Confederated
ontology for the transport system was proposed. This ontology International Conferences” On the Move to Meaningful
has been developed by understanding the common terms used Internet Systems”, pages 643–653. Springer, 2016.
in the transport system and by looking at the different 11. B. Dutta, D. Nandini, and G. Kishore Shahi. Mod:
preference criteria chosen by the user such as cost, duration, metadata for ontology description and publication. In
number of connections and safety. To build this ontology, we International Conference on Dublin Core and Metadata
opted for a hybrid architecture, which consists in developing a Applications, pages 1–9, 2015.
global ontology from local ontologies. For the present case, 12. C. Masolo, S Borgo, A. Gangemi, N Guarino, A.
we are taking into account only two information resources: Oltramari, and L Schneider. Dolce: a descriptive
Tramway and Bus Data Bases. In a future work, we propose to ontology for linguistic and cognitive engineering.
extend our ontology in order group all the sources so that WonderWeb Project, Deliverable D17 v2, 1:75– 105,
these data are comprehensible and accessible for the agents of 2003.
the system. Moreover, we will detail, the negotiation and 13. M. A Musen. The protégé project: a look back and a
communication part between the agents themselves and look forward. AI matters, 1(4):4–12, 2015.
between the different layers of the system. The ontology can https://doi.org/10.1145/2757001.2757003
be used with a data integration technique to build a 14. H. ZGAYA, and S. HAMMADI, Multi-Agent
multimodal transport system with an agent based semantic Information System Using Mobile Agent Negotiation
approach. Based on a Flexible Transport Ontology, In
Proceedings of the 2007 International Conference on
REFERENCES Autonomous Agents and Multi-agent Systems
(AAMAS’2007). May 14-18, Honolulu, Hawaii, 2007.
1. C. Comtois and B. Slack. The geography of transport
15. K. Basak , R. Seshadri, L. Azevedo, C. M., 2018.
systems. Routledge, 2009
SimMobility – Integrated Simulation Platform.
2. M. Becker and S. F. Smith. An ontology for
Research project. Massachusetts Institute of Technology,
multi-modal transportation planning and scheduling.
Cambridge, MA, USA [Accessed: 31-Aug-2018].
Technical Report CMUR I-TR-98–15, 1997.
16. A.S. Niaraki, K. Kim, Ontology based personalized
3. S. Timpf, Ontologies of wayfinding: a traveler’s
route planning system using a multi-criteria decision
perspective. Networks and spatial economics,
making approach, Expert Systems with Applications,
2(1):9–33, 2002.
36, pp. 2250– 2259. (2007)
https://doi.org/10.1023/A:1014563113112
https://doi.org/10.1016/j.eswa.2007.12.053
4. Junli Wang, Zhijun Ding, and Changjun Jiang. An
17. J. LARIOUI, A. EL BYED, A Multi-Agent Information
ontology-based public transport query system. In
System Architecture For Multimodal
Semantics, Knowledge and Grid, 2005. SKG’05. First
Transportation, Embedded Systems and Artificial
International Conference on, pages 62–62. IEEE,2005
Intelligence Proceedings of ESAI 2019, Fez, Morocco,
5. A. Sadeghi Niaraki and K. Kim. Ontology based
Pages 795-803
personalized route planning system using a
18. P.Mulholland, 1999, Introduction to ontologies, Version
multi-criteria decision making approach. Expert
2, Internal Report, code: RichODL-OU-3/1999,
Systems with Applications, 36(2):2250– 2259, 2009.
Enriching ODL by knowledge sharing for
collaborative computer- based modeling and
6. H. Mnasser, M. Khemaja, K. Oliveira, and M. Abed. A
simulation.
public transportation ontology to support user travel
19. F. Lhafiane, A. Elbyed, M. Bouchoum, Reverse
planning. In Research Challenges in Information
Logistics Information Management Using
Science (RCIS), 2010 Fourth International Conference
Ontological Approach, World Academy of Science,
on, pages 127–136. IEEE, 2010.
Engineering and Technology International Journal of
7. K. M. De Oliveira, F. Bacha, H. Mnasser, and M. Abed.
Information and Communication Engineering Vol:9,
Transportation ontology definition and application
No:2, 2015.
for the content personalization of user interfaces.
20. J. LARIOUI, A. EL BYED, An Advanced Intelligent
Expert Systems with Applications, 40(8):3145–3159,
Support System for Multimodal Transportation
2013.
Network Based on Multi-Agent Architecture,
2477
Advanced Intelligent Systems for Applied Computing

Sciences, Volume 4 - Pages 98-106,2020.
https://doi.org/10.1007/978-3-030-36674-2_10
21. A. Ghasempour et al., Using Traffic Control Scheme In
Intelligent Transportation System, International
Journal of Advanced Trends in Computer Science and
Engineering, 8(1.4), 2019, 165- 172
https://doi.org/10.30534/ijatcse/2019/2581.42019
22. Nur Raidah Rahim et al., Integrated Ontology
Development for Clinical Decision Support System in
the Case Study of Methadone Maintenance Therapy,
International Journal of Advanced Trends in Computer
Science and Engineering, 8(1.6), 2019, 272 – 282
https://doi.org/10.30534/ijatcse/2019/4181.62019
2478

E1A119089 - Wa Ode Ainun Jariah

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

E1A119089 - Wa Ode Ainun Jariah

Uploaded by

Copyright:

Available Formats

NUMBER

Dolya C., Botsman A.,

1. To establish the scientific

To establish the scientific approaches proposed by contemporaries on the

Kaveh Bakhsh Kelarestaghi and Kevin

Intelligent Transportation System

This study attempts to examine the impacts

The methodology of risk assessment (Figure 1) comprises of (1)

The risk-based approach of this study delivers the

Reza Sayyadi & Anjali

An integrated approach based on

The results of our study reveal that

The proposed methodology

Extending the current study to integrate large size

Yu Shen, Hongmou Zhang

Integrating Shared Autonomous

Agent-based simulation of passengers,

Finally, the case study presented in the paper is

A review and analysis of

These strategies are envisaged to

The methodology of the study includes the

A review and analysis of the railway

Towards a Semantic Layer Design for an

The global purpose of the advanced

The structure of this paper is as follows:

This ontology has been developed by

, Danman WU1 , Qiuwei WU2 ,

The penetration of EVs in modern

With the increasing

Vol. 19, no. 1, 2018

There are following tasks to achieve the

In scientific literature is a lack of

HOF routes and processes evaluation

Vol. 18 No. 3, September

Nurul Qamar and Aan Aswari

This paper is not intended to reject

Presence of Healing Innovative

The evolution of ideal conditions initiated by

VMS Importance in ITS Network

Methodology of Risk Assessment

FIGURE 1 Methodology of Risk Assessment (adapted from NIST SP 800-30 [20]).

© 2018 SAE International. All Rights Reserved.

Threat Source and

Subsequently, by entering the default usernames and pass-

Threat Events and System

© 2018 SAE International. All Rights Reserved.

© 2018 SAE International. All Rights Reserved.

TABLE 1 Summary of known hacking events in the U.S.

Cyber-Attack: “Sun Hacker”

© 2018 SAE International. All Rights Reserved.

© 2018 SAE International. All Rights Reserved.

Adverse Impact and Countermeasures

© 2018 SAE International. All Rights Reserved.

a malicious node in the TMC severity level form Property

Operational Impact: Traffic

Operational Impact: System

operators' financial losses

Efficiency Impact: Increase in Reliability Impact: Road users’

energy consumption distrust of a system

© 2018 SAE International. All Rights Reserved.

against cyber-physical attacks. Secure and resilient transportation infrastructure enables

It needs to be understood that with the progression of technology, the operator’s

© 2018 SAE International. All Rights Reserved.