Transportation Engineering 10 (2022) 100152

Contents lists available at ScienceDirect

Transportation Engineering
journal homepage: www.sciencedirect.com/journal/transportation-engineering

Evaluation and improvement of the urban transportation networks

resilience in short-term non-recurring traffic congestion: a novel graph
connectivity-based criteria
Mohammad-Ali Gorji a, Meisam Akbarzadeh b, Seyyed-Nader Shetab-Boushehri a, *
a
Department of Industrial and Systems Engineering, Isfahan University of Technology, Isfahan 84156-83111, Iran
b
Department of Transportation Engineering, Isfahan University of Technology, Isfahan 84156-83111, Iran

A R T I C L E I N F O A B S T R A C T

Keywords: Work or study purpose trips usually make up most intercity trips during peak hours. Being at a destination at a
Traffic congestion specific time is one of the characteristics of such trips, and the occurrence of non-recurring short-term traffic
Transportation networks resilience congestion can damage these trips. Due to the necessity of the subject, this study examines the urban trans­
Travel time
portation networks resilience (TNR) in the face of non-recurring traffic accidents with short-term effects. This
Graph connectivity
paper defines a novel criteria to measure the urban TNR by developing the connectivity concept graph theory.
Then, based on the proposed criteria, the resilience of the Isfahan transportation network (one of Iran’s met­
ropolises), has been evaluated, and the critical nodes and arcs of this network have been identified. Also, by
performing sensitivity analysis, the role of the travel time satisfaction threshold is discussed. Finally, a series of
projects have been introduced, which are believed to increase Isfahan’s TNR. The projects have been compared
and evaluated based on the proposed method.

1. Introduction these passengers changes significantly compared to their travel time

under normal circumstances, it will lead to their dissatisfaction. How­
With the growth of cities in recent years, accidents and catastrophes ever, if the urban transportation network provides suitable alternative
have increased. Disasters can be divided into two categories depending routes for the mentioned passengers, it will reduce their dissatisfaction
on the impact they have on the behavior of passengers in the urban [6,7]. In this case, the urban transportation network can be called
transportation network. In the first category, most passengers of the resilient due to its ability to absorb the effects of this type of network
urban transportation network will abandon their trip and choose event.
another destination according to the circumstances [1,2]. This group The resilience of a system is a concept that has been defined con­
includes disasters such as earthquakes, floods, explosions in vital urban cerning its exposure to various disasters [8]. Given the importance of the
facilities, broken dams, and tsunamis. However, the second category of urban transportation network, the ability of the network to absorb the
catastrophes does not damage the infrastructure of the urban network effects of adverse events properly and return it to the level of service
but also blocks some streets and, as a result, reduces the capacity of the before the disaster or higher at a reasonable time is presented as a
transportation network and creates heavy congestion. Although these general definition of urban transportation networks resilience (TNR)
accidents have financial and human losses, their occurrence does not [9]. The urban TNR depends on the type of disasters that occur in it and
change the destination of passengers. Incidents such as riots and civil can be evaluated with specific components and criteria. The focus of this
protests in a part of the city, building collapse on the street, techno­ study is on disasters that do not change the destination of passengers. In
logical accidents such as explosions in a part of the city, chain traffic this regard, this paper answers the following questions:
accidents, and the widespread interruption of traffic lights are in this
category [3,4]. When these disasters occur, the concern of passengers A) What is a suitable criteria that can assess the urban TNR in the
within the network is to reach their desired destination in time [5]. In face of traffic congestion during disasters?
many trips, such as business and educational trips, if the travel time of

* Corresponding author.
E-mail addresses: m.gorji@in.iut.ac.ir (M.-A. Gorji), makbarzadeh@iut.ac.ir (M. Akbarzadeh), shetab@iut.ac.ir, shetab@cc.iut.ac.ir (S.-N. Shetab-Boushehri).

https://doi.org/10.1016/j.treng.2022.100152
Received 18 May 2022; Received in revised form 1 November 2022; Accepted 15 November 2022
Available online 19 November 2022
2666-691X/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

B) How can the proposed index identify important streets and in­ evaluate transportation systems’ performance in the face of non-
tersections from the perspective of network resilience in the face recurring accidents. Sansó and Soumis [19] have developed a general
of these events? method for evaluating different flow networks (communications, power,
C) How can we identify local alleys and streets that, in the event of a and transportation) in the face of non-recurring events. They considered
disaster, play the role of alternative streets and cause the most the reliability assessment tools, including mathematical description of
significant improvement in the urban TNR? the system (network) understudy, determination of performance index,
mathematical model of failure (failure), and path selection model in the
As a description of contribution, this study examines the urban TNR network. Then, using these tools, they evaluated the reliability of street
in the face of short-term non-recurring events by developing graph networks in the face of this type of accident. Chung [20] proposed a
connectedness concepts in the urban transportation network. The con­ method for quantifying non-return traffic congestion caused by freeway
tributions of this work are derived from: repair activities and evaluated the efficiency of this method using traffic
data from Korea. Some researchers have used queue theory and simu­
1. Developing the connectivity concepts into a transportation network lation tools to study the impact of traffic accidents on the performance of
based on travel time and travellers satisfaction threshold. transportation networks [21,22]. Using a network-based method, Li
2. Providing novel practical criteria to evaluate the urban TNR. et al. [23] simulated and analyzed non-return traffic congestion caused
3. Implementing criteria on a huge real case study and evaluating by transportation network accidents.
Isfahan metropolian’s TNR. Catastrophes such as explosions in parts of the city, chain accidents,
and the widespread cutting off of traffic lights have effects beyond the
In the next section, the literature is reviewed on related topics. In the impact on a street or an intersection of urban transportation networks.
third section, the problem is stated, and its assumptions are presented. In The importance of network passengers’ travel time during these in­
section four, the proposed definitions and criteria are presented. In the cidents has led to the presentation of evaluation criteria such as travel
fifth section, using the proposed criteria, Isfahan’s TNR is evaluated and time and retrieval time of the network [24–26]. Kamga et al. [27]
analyzed. In the sixth section, the methods of improving the resilience of analyzed the performance of the urban network in the event of a traffic
the Isfahan transportation network are introduced and compared. In the accident using travel time modeling. They calculated the difference in
final section, summaries, conclusions, and suggestions for future travel time between the two networks by calculating the travel time of
research are stated. the entire network in two ordinary modes and in the event of an acci­
dent. Then, using this index, they studied the Chicago area network. The
2. Literature review present study also refers to the criterion of passengers’ travel time in the
network when non-recurring accidents occur with short-term effects.
According to the subject of this study, a literature review is presented
on two topics: " Evaluation of urban transportation network perfor­ 2.2. Connectivity concept application in evaluating network resilience
mance in traffic events " and "Application of the graph connectivity
concept in the network resilience evaluation." Among the methods presented to evaluate the resilience networks,
researchers have widely used graph theory due to the simple and
2.1. Evaluation of urban transportation network performance in traffic straightforward representation of networks [28,29]. In this regard,
events graph theory-based criteria focus on the structure of networks regardless
of the current in them [30–32]. Criteria such as the average degree of
In the literature, research related to the issue of evaluating the per­ network nodes [32], betweenness centrality [33], and network diameter
formance of urban thoroughfares in the face of accidental congestion is [31] are widely used examples that have been used to evaluate networks
divided into two groups. The first group includes repetitive and inter­ from a topological point of view [34]. Akbarzadeh et al. [35] have used
mittent events that lead to recurrent and recurring crowds [10]. This the centrality criterion to assess the resilience of the transportation
type of traffic congestion is due to changes in transportation demand or network. Ip and Wang [36] defined network resilience based on the
changes in the capacity of the streets due to the traffic behavior of number of reliable, independent paths between each pair of nodes.
network users. The result of these repetitive hustles and bustles is that Mahdavi M, Bhouri, and Scemama [37] suggested a methodology to
the travel time on different streets of the network at a specific time on count the dynamic resiliency of a public transport network. They rep­
different days is not a fixed value and changes randomly [11,12]. resented the association between system elements by passenger flow,
Numerous studies have been conducted about evaluating the urban network topology, and service performance dynamics. Liu et al. [38]
transportation network performance in the face of recurring traffic proposed an assessment method for urban road network resilience,
congestion [13–15]. including robustness, resilience performance, and recovery indexes.
The second group of investigations is related to network congestion They inferred that the intersection-based attack has the most meaningful
in the face of non-recurring accidents. This type of traffic congestion can impact on resilience, and resilience is associated with the node degree of
be caused by traffic accidents [16], natural disasters such as storms, the attacked intersection.
floods, earthquakes [17], and similar special events. About 65% of urban Connectivity concepts in graph theory are related to the connection
network congestions are non-recurring [10]. In this regard, a group of between all graph nodes [39]. Connectivity concepts have been widely
studies evaluated the performance of transportation networks in the face considered and used by researchers in engineering [40]. Nagarajan et al.
of floods and earthquakes [13]. The occurrence of these incidents has proposed algebraic connectivity as a criterion for evaluating the
long-term effects on the performance of transportation networks. It robustness of networks in various engineering issues. By examining four
causes significant changes in network passengers’ daily behavior, such different problems in engineering sciences, they related the value of
as travel deletion, change in the travel destination, or change in the algebraic connectivity with the objective function of these problems.
travel route. Another group of researchers has evaluated the perfor­ Finally, they presented an innovative method for calculating the amount
mance of transportation networks in the face of accidents whose of algebraic connectivity in networks [41,42]. Nekudari and Ghasemi
occurrence has a short-term impact on the performance of trans­ [43] investigated increasing the degree of algebraic connectivity in
portation networks [18,19]. Events such as traffic accidents do not power networks on reducing chain failures in these networks. These
change the destination of passengers on the network and may only affect researchers compared and evaluated two strategies, including adding an
the route to the passengers’ destination. arc to the network core and adding an arc to the entire network through
Researchers have proposed various methods and approaches to refrigeration simulations.

2
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

have not yet started their journey start their journey only when their
journey is suitable.
5. The occurrence of an event in the streets causes it to be completely
closed and the corresponding arc to be removed from the network.
Also, the event at an intersection has closed the streets leading to it.
The corresponding node of that intersection and the arches entering
that node will be removed from the network.

Now consider the urban transportation network during peak travel

hours when passengers are traveling. According to the above assump­
Fig. 1. Example of vertex and edge connectivity: The above network is of type tions, with an accident in one of the streets or intersections of the
1-vertex connected (Deleting vertex F disconnects the network) and 2-edge network, all network passengers are immediately informed of its
connected (Deleting edges BF and DF disconnects the network). Also, AG is occurrence and location. If the damaged street or intersection is in their
an 2-path connected origin-destination. path, regardless of the size of the accident, they change direction, which
may increase their travel time. If the passenger’s travel time increases
The connectivity concept has had applications in communication beyond a specific limit (satisfaction threshold), he will be dissatisfied. It
and telecommunication engineering [44]. Shi and Servatius [45] is clear that the more suitable alternative routes for these passengers are
analyzed the reliability of degradable networks. They presented a rela­ available in the network, the less dissatisfaction is created for passen­
tion to calculate the reliability of the entire network by expressing the gers. It can be said that the network is more resilient in the face of such
concept of network degradability and the reliability of network arcs accidents.
based on the degree of arc connectivity. Le et al. [46] examined the After defining the problem, then by presenting the concepts of node
metro system information network using the concept of graph connec­ and arc connectivity in graph theory and development in the trans­
tivity. Chiesa et al. investigated information routing in Internet portation network, the proposed criteria to evaluate the TNR are
communication networks when damage occurs in several studies. Using defined.
graph connectivity concepts, they proposed various methods such as
random routing [47–49]. Ma et al. [50] addressed the issue of designing 4. Network node and arc connectivity
a street network to reduce travel time. By stating that the smaller the
diameter of the network reduces passengers’ travel time within the In graph theory, graph G is connected if and only if there is at least
network, they theoretically proved that the upper and lower limits of the one path between the two desired nodes. Also, a graph is a k-vertex
network are inversely related to the algebraic connectivity of the graph connected if and only if it has more than k vertices and remains con­
network. They proposed a model to increase network connectivity. nected whenever fewer than k vertices are removed. Finally, a connected
By summarizing the studies, it can be said that the use of graph graph is k-edge-connected if it remains connected whenever fewer than
theory in network resilience evaluation has shown its efficiency. How­ k edges are removed [52].
ever, graph connectivity theory application in the urban transportation According to the definition, an origin-destination is called a k-path
network according to the demand of origin-destination has not been connected if that destination can be reached by k separate paths (paths
reviewed yet. Continuing the previous research, we develop the concept that do not have a common arc) from that origin. Also, a network is
of arc and node connectivity to evaluate the TNR during short-term non- called k-path connected if all its origins-destinations are at least k-path
recurring traffic congestion from a travel time perspective. Review connected [53]. Fig. 1 shows an example of the mentioned concepts.
studies also point to the need for models to assess the resilience of large-
scale transportation networks with actual data [51]. The present study
4.1. Development of connectivity concepts in urban transportation
uses real data and based on the proposed model, evaluates the Isfahan’s
network
TNR. To the best of our knowledge no study has been developed the
k-link connectivity concepts into a transportation network based on
A street network N(V, A) can be represented using a directed graph
travel time and travel satisfaction threshold.
in which V is the set of network nodes (vertices), and A is the set of arcs
(edges). At a specific time, the state of the network can be represented by
3. Problem definition and assumptions
the state vector c = {c11 , …, cmn } where the element cij represents the
state of the arc (i, j) in the network at the time. cij = 0 means that the arc
In this section, the problem under study is defined. This study aimed
(i, j) is closed and cij = 1 means that the arc is open. c = {1, 1, …, 1}
to determine the resilience of an urban transportation network in the
means the condition that no accident occurred in the network and all
face of non-recurring accidents with short-term effects. First, it is
network arcs are open. This state is called the normal state for the
necessary to express the assumptions used to define the problem. The
network and is denoted by c0 .
assumptions considered in the problem definition are as follows:
Definition 1: Suitable travel sequence: The passenger travel
sequence in position c shown as (rtes (c)), is a trip whose starting point
1. Events occurred on the streets or intersections of the city’s trans­
(where the passenger is at point) is the node in the network (node e), its
portation network.
destination point (the main destination of the passenger) is node (s), and
2. When an accident occurs, all network passengers are immediately
informed of its occurrence and location. This awareness is usually starting state is (c). Also, c0 and c show the state of the network before
done through city signs and information systems. and after the accident, respectively. According to Eq. (1), if the ratio of
3. For network passengers, it is only essential to reach their destination the travel time of a travel sequence after the accident (t es (c)) to the travel
at the right time. Therefore, after being informed of the event, if the time of that travel sequence under normal conditions (t es (c0 )) is less than
damaged street (intersection) is on its way, regardless of the size of network passenger’s satisfaction threshold (θ), then it called an suitable
the accident, they want to change the route to reach the destination travel sequence. In Eq. (1), the unit of tes (c) and t es (c0 ) equals time
at the right time. (minute), and (θ) is unitless.
4. An event only affects the situation of passengers traveling in the { es
t (c)
}
network at the time of its occurrence. In other words, people who es 0
≤θ (1)
t (c )

3
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

In this regard, node connectivity degree and arc connectivity degree

of ND for transportation networks are defined as follows:
js
The ND (j, s) is k-node connected (VCθ = k), if and only if, it is
necessary to remove at least k members from the set of network nodes
for delete all suitable paths of this ND. Also, the ND (j, s) is k-arc con­
js
nected (ACθ = k), if and only if, it is necessary to remove at least k
members from the network arc set to remove all suitable paths of this
ND.

4.2. Proposed criteria of urban TNR

The previous section defined the concepts of the degree of node

connectivity and arc connectivity of a destination-destination. In this
section, the TNR criteria are presented using the node connectivity and
arc connectivity.
Definition 4: Network arc importance (IMP(m,n) ): The importance
Fig. 2. Node-destination (j, s) is 2-path connected. of arc (m, n) ∈ A from the Transportation network is equal to the vol­
ume of ND trips in the network, which will be unsuitable by removing
Definition 2: Suitable path sequence: If tρes (c) is the travel time of the arc. Mathematically:
∑ ∑
rt (c) through path ρ, then path ρ is a suitable path sequence if Eq. (2) is
es
IMP(m,n) = xjs − xjs (3)
satisfied. (j,s)∈ND[N(V,A)] (j,s)∈ND[N(V,A′ )]

{ es }
tρ (c) where in Eq. (3) xjs is the suitable trip volume between the ND (j, s),
≤ θ (2)
tes (c0 ) ND[N(V, A)] is the set of NDs belonging to the network N(V,A), and A =
′

js
According to Eq. (2), if the ratio of the travel time of a path sequence A − (m, n). The unit of x and IMP(m,n) trip volume (passenger).
after the accident (tρes (c)) to the travel time of that path sequence under In other words, there are 1-arc connected trips from the NDs in the
normal conditions (tρes (c0 )) is less than network passenger’s satisfaction network for which the arc (m, n) plays a key role. These trips will be
threshold (θ), then it called an suitable path sequence. unsuitable after removing the arc (m, n). The sum of these unsuitable
Definition 3: k-Path connected Node-Destination (ND): The ND trips is equal to the importance of arc (m, n), which is called IMP(m,n) .
(j, s) is called k-path connected, if and only if the destination s can be Definition 5: Network node importance (IMPV ): The importance
accessed from node j with k suitable separate paths [26]. For example, in of the v ∈ V node from the Transportation network is equal to the vol­
Fig. 2, there are four paths between the ND (j, s), one of which is longer ume of ND trips that will be unsuitable by removing the arcs entering
and unsuitable than the others. As a result, there are three suitable paths that node. Mathematically:
between the ND (j, s), but there are two independent paths, so the ND
(j, s) is a 2-Path connected ND.

Fig. 3. Traffic flows on links of Isfahan during peak hours.

4
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 4. Twelve traffic zones of Isfahan city.

∑ ∑
IMPv = xjs − xjs (4)
(j,s)∈ND[N(V,A)] (j,s)∈ND[N(V,A′′)]

where in Eq. (4): A′′ = A − ∀(i, v).

Similar to the previous explanation, there are 1-node connected trips
from the NDs in the network for which the v ∈ V node plays a key role.
These trips will be more unsuitable after removing the arcs entering
node v. The sum of these unsuitable trips is equal to the importance of
node v, which is called IMPv .
Definition 6: TNR from the arcs perspective (ResA ), defined as the
flow of k-arc connected network NDs which (k ≥ 2), divided by the total
flow of network NDs. Mathematically: Fig. 5. The steps of the proposed method to calculate the importance of
∑ ∑ ∑ network arcs.
js js js
(j,s)∈B x (j,s)∈B x − (j,s)∈B′′ x
′
ResA = ∑ js
= ∑ js
(5)
(j,s)∈B x (j,s)∈B x
Table 1
where in Eq. (5): Classification of Isfahan network arcs based on arc importance.
Category number 1 2 3 4 5
B = {(l, d) ∈ ND[N(V, A)]}
Category color Blue Green Yellow Orange Purple
{ / }
B′ = (l, d) ∈ ND[N(V, A)] ACθld ≥ 2 The importance of Very Low Middle High Critical
category arcs low
{ / } Number of arcs 4150 116 21 24 12
B′′ = (l, d) ∈ ND[N(V, A)] ACθld = 1 Impact of each arc ≤ 1500 1500 to 3000 to 4500 to ≥ 6000
category (trips) 3000 4500 6000
Definition 7: TNR from the nodes perspective (ResV ), defined as
the flow of k-node connected network NDs which (k ≥ 2), divided by the
total flow of network NDs. Mathematically: network NDs, the lower the TNR. In contrast, the higher flow of network
∑ js
∑ js
∑ js trips on the k-arc (node) connected network NDs which (k ≥ 2), the
(j,s)∈C x (j,s)∈C x − (j,s)∈C′′ x
higher the network resilience. Also, the characteristic of Eqs. (5) and (6)
′
Resv = ∑ js
= ∑ js
(6)
(j,s)∈C x (j,s)∈C x is that the resilience quantities from these formulas do not change when
the network components (intermediate nodes or arcs) change.
where in Eq. (6): In this study we discuss resilience based on the concept of 1-arc
C = {(l, d) ∈ ND[N(V, A)]} connected NDs and 1-node connection NDs (k =1), which means there
is an incident in the network. Calculating the resilience value for k =2
{ / }
′
C = (l, d) ∈ ND[N(V, A)] VCθld ≥ 2 and other quantities is also possible. But it is rare for two incidents to
occur in a network at the same time leading to street documents.
{ / } However, it is helpful to examine resilience for different values of k for
C′′ = (l, d) ∈ ND[N(V, A)] VCθld = 1
situations where crises occur, such as war attacks.
In explaining Eqs. (5) and (6), the transportation network’s weakness
is the NDs that become unsuitable by removing one street (or intersec­
tion). The higher flow of network trips on 1-arc (node) connected

5
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 6. Isfahan network arcs coloring based on arc importance.

Fig. 7. Comparison of traffic zones based on the number of critical and high importance arcs.

5. Case study: Isfahan city transportation network Isfahan. The city of Isfahan has nearly two million people, and there are
more than 750,000 vehicles. All daily trips add up to 2.6 million, of
In this section, Isfahan transportation network is introduced, and its which 37% are work trips, and 24% are education trips. This city’s peak
resilience in the occurrence of short-term non-recurring traffic accidents hours of travel time are 7:00 am to 9:00 am. Most peak hours trips are
is evaluated and analyzed using the proposed criteria. business and educational trips, and these trips need to reach the desti­
nation on time. Trips within the Isfahan transportation network during
peak hours are shown in Fig. 3.
5.1. Introduction of Isfahan city transportation network
Isfahan Municipality has divided the city into twelve traffic zones,
which is shown in Fig. 4.
Isfahan is a historical city in the center of Iran, which is the third-
largest city in terms of size and population. The loading and concen­
tration of the country’s primary industries in Isfahan province and the 5.2. Determining arcs importance
rapid population growth emphasize the need for comprehensive plan­
ning in socio-economic and environmental dimensions in the city of The method of determining the importance of network arcs is as

6
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 8. Isfahan network arcs categories based on arc importance.

follows: arcs better. The classification intervals and their numerical results are
shown in Table 1.
1. The travel time and flow/volume are calculated for all NDs of the The categorized arcs are placed on the map of Isfahan in Fig. 6.
network. Then, a comparison was made between traffic zones to identify their
2. To calculate each arc’s importance, it is then removed from the relative importance in terms of critical and high important arcs in them.
network, and the travel time is recalculated for all NDs of the new Fig. 7 shows the results of this comparison.
network. Then, the travel time of all NDs of the network before and Also, the arcs related to each category can be seen in Fig. 8. Ac­
after removing the desired arc is compared. According to Eq. (1), the cording to Fig. 8, it can be concluded that the western streets of Isfahan
NDs that increase travel time more than the passenger satisfaction transportation networks have a more significant role than the eastern
threshold (θ) are determined. Here, the satisfaction threshold (θ) is streets in the resilience of the transportation network. This difference
assumed to be 1.20 for the city of Isfahan. could be due to educational facilities such as universities, industrial
3. The total travel volume obtained in this way is calculated and estates, and steel industries in the city’s west, northwest, and southwest.
introduced as the importance of the desired arc. Also, by looking at category 5 in Fig. 8, it can be concluded that the
northern entrance of Isfahan has many critical arcs of Isfahan.
The steps of the proposed method to calculate the importance of
network arcs can be seen in Fig. 5.
5.3. Determining nodes importance
The importance of 4323 arcs in the Isfahan city network was calcu­
lated with this method. Then, network arcs were divided into five cat­
The method of determining the importance of network nodes is very
egories based on their importance to identify the importance of these
similar to what was done with arcs. First the travel time and volume are

7
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Then, a comparison was made between traffic zones based on high

importance and critical nodes. Fig. 11 shows the result.
Traffic zone rankings based on critical nodes and arcs can be used to
prioritize resource allocation to traffic zones to prevent streets and in­
tersections from being blocked.
Also, the nodes related to each category can be seen in Fig. 12. By
analyzing and summarizing the results of nodes’ importance in terms of
the number of affected NDs and the number of affected trips, five in­
tersections in the south of the city network and three intersections in the
north of the city are critical intersections. These intersections are con­
nected to bridges, resorts, and intercity bus terminals.

5.4. Determining the Isfahan’s TNR

TNR criteria from the arcs and nodes perspective are given in Eqs. (5)
and (6), respectively. As a reminder, these equations are listed in
Table 3:
The number of nodes in the street network of Isfahan city is 1481 and
the number of destinations is 321. As a result, with the calculations
performed for the Isfahan transportation network, it was determined
that the number of NDs within the network when the network is in
normal condition and no event has occurred equal to:
∑ ∑
ljs = ljs = 321 × 1481 = 475401
(j,s)∈B (j,s)∈C

The number of trips within the network when the network is in

normal condition is equal to:
∑ ∑
xjs = xjs = 61965
(j,s)∈B (j,s)∈C

It is necessary to explain here that the total number of daily trips in

Isfahan is 2.6 million as mentioned.“61965” is the number of trips
available in the network at the time of the accident (in the peak hour).
The number of 1-arc connected NDs of the network is equal to:
∑
ljs = 216215
(j,s)∈B′′
Fig. 9. The steps of the proposed method to calculate the importance of
network nodes. The number of 1-arc connected NDs of the network trips is equal to:
∑
xjs = 35615
(j,s)∈B′′
Table 2
Classification of Isfahan network nodes based on node importance. The number of 1-node connected NDs of the network is equal to:
Category number 1 2 3 4 5 ∑
ljs = 232475
Category color Blue Green Yellow Red Purple (j,s)∈C′′
The importance of Very Low Middle High Critical
category nodes low The number of 1-node connected NDs of the network trips is equal to:
Number of nodes 1109 120 119 122 11 ∑
Impact of each ≤ 2500 2500 to 5000 to 7500 to ≥ 10000 xjs = 34130
node category 5000 7500 10000 (j,s)∈C′′
(trips)
Based on the above values and equations in Table 3, the resilience of
the Isfahan transportation network from the arcs and nodes perspective
calculated for all NDs of the network. To calculate the importance of is calculated as follows:
each node, it then removes that node and its incoming arcs from the ∑ ∑
js js
network. The travel time is recalculated for all NDs of the new network. ResA =
(j,s)∈B x −
∑ (j,s)∈B′′ x
=
61965 − 35615
= 0.42
Then, the travel time of all network NDs before and after the removal of xjs 61965
(j,s)∈B
the node is compared. According to Eq. (1), the NDs that increase travel ∑ ∑
time more than the passenger satisfaction threshold (θ = 1.2) are
js js
(j,s)∈C x − (j,s)∈C′′ x 61965 − 34130
Resv = ∑ = = 0.45
determined. Finially, the total travel volume obtained in this way is (j,s)∈C xjs 61965
calculated and introduced as the importance of the desired node.
The steps of the proposed method to calculate the importance of The above results show that 42% of Isfahan passengers pass through
network arcs can be seen in Fig. 9. NDs that are not the 1-arc connected during the peak hours travel time,.
The importance of 1481 nodes in the transportation network of As a result, Isfahan’s TNR can be considered 42% from the arcs
Isfahan was calculated, and for better identification, network nodes perspective (ResA ). Also, 45% of Isfahan passengers pass through NDs
were divided into five categories based on their importance. The nu­ that are not the 1-node connected, and Isfahan’s TNR is 45% from the
merical results are shown in Table 2. nodes perspective (Resv ).
The categorized nodes are placed on the map of Isfahan in Fig. 10.

8
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 10. Isfahan network nodes coloring based on node importance.

Fig. 11. Comparison of traffic zones based on the number of critical and high importance nodes.

5.5. The effect of travel satisfaction threshold on the TNR TNR from the arc and node perspective. The slope of this uptrend
gradually decreases. As one reason for this, we can refer to short trips in
Another important issue addressed in this section is the effect of that the node is at the beginning of a street, and the destination is at the
passengers’ satisfaction threshold (θ) on the resilience level of the end of the street. The time of these trips is significantly increased after an
transportation network. Passengers’ satisfaction threshold for each accident on this street. These 1-arc connected NDs will remain 1-arc
person varies depending on the type of travel and his conditions. In the connected, with changes in the satisfaction threshold.
previous section, Isfahan’s TNR was calculated with the assumption =
1.2. In this section, the Isfahan’s TNR is calculated for different values of 6. Isfahan’s TNR improvement
(θ) and shown in Fig. 13.
As expected, as the travel satisfaction threshold increases, so do the Isfahan managers have two main approaches to improve the TNR.

9
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 12. Isfahan network nodes categories based on node importance.

6.1. Improvement by strengthening important streets and intersections

Table 3
TNR criteria.
Improving the traffic situation on some of the critical streets (in­
∑
TNR from the arcs perspective (ResA ) (j,s)∈B′ xjs tersections) of the network can be done by correcting the design,
ResA = ∑ =
(j,s)∈B x
js
increasing the presence of law enforcement officers, and improving the
∑ ∑
(j,s)∈B x −
∑
js
(j,s)∈B′′ x
js
status of traffic signs on them. As mentioned before, the critical streets
∑
(j,s)∈B x
js
(intersections) are those where the connection of many NDs of the
TNR from the nodes perspective (j,s)∈C′ x
js
network is unsuitable when an accident occurs in them, and they are
Resv = ∑ =
(j,s)∈C x
js
blocked. Therefore, by improving the traffic situation of these streets
(ResV )
∑ js
∑ js
(j,s)∈C x − (j,s)∈C′′ x (nodes), the number of 1- arcs (nodes) connected NDs in the network
∑
(j,s)∈C x
js
will be reduced as much as possible. The ENH index is defined as Eq. (7):
∑ ( js )
γ − 1 js
The first approach is to reduce traffic events in some important streets ENH(m,n) = x (7)
′ γjs
(intersections) of the network by strengthening important streets and
(j,s)∈ND[N(V,A)− N(V,A )]

intersections. The second one is the implementation of new projects to

where ENH(m,n) in Eq. (7) is the index of passengers’ trip improvement
improve some local alleys to create alternative routes for traffic
congestion at the time of the accident. This section examines these two due to the improvement of the traffic situation on the (m, n) street. xjs is
approaches. the number of ND (j, s) passengers, γ js is the number of arcs that the ND

10
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 13. Effect of passenger satisfaction threshold on the Isfahan’s TNR.

Using the above method, the ENH index was calculated for 4323 arcs in
the Isfahan city network and 30 arcs with the highest ENH index were
selected. Selected arcs for improvement are shown in Fig. 14.

6.2. Improvement through the implementation of proposed projects

As mentioned, the second approach for Isfahan’s TNR improvement

is to identify local roads such as alleys and unique routes in the network
and equip and improve their condition to be ready for evacuating ve­
hicles and reduce traffic congestion at the time of the accident. In order
to use this approach, with the help of specialists of the Isfahan trans­
portation department, fifteen alleys and side roads that have the
essential capabilities to be used as an alternative route at the accident
time were identified and selected as network resilience improvement
projects. These alleys and sidewalks can be seen in Fig. 15.
In order to investigate the effect of each project on Isfahan’s TNR,
each time adding it to the Isfahan transportation network. Then the
amount of reduction in 1-arc connected NDs and 1-node connection NDs
was calculated. In this way, it can be understood how much the imple­
mentation of each of these projects will increase Isfahan’s TNR. The
effect of each project in reducing the 1-arc connected NDs trips and 1-
node connected NDs trips is shown in Table 4.
Analysis of the results in Table 4 shows that projects 15, 5, and 1
have the most significant effect on increasing Isfahan’s TNR,
respectively.
Fig. 14. Selected arcs for Isfahan’s TNR improvement.
7. Conclusions and suggestions for future studies
(j, s) becomes 1-arc connected as a result of removing each of them from
the network. For instance, consider an ND that contains 15 streets. Three In recent years, events leading to traffic in urban transport have
streets of this ND have this feature that the path becomes unconnected increased. This issue has highlighted the necessity to assess the urban
by removing each of them. In this case, gamma is equal to three. After transportation networks resilience (TNR). Non-recurring traffic acci­
improving and reinforcing one of these three arcs, ND will remain 1-link dents with short-term effects are accidents whose occurrence does not
connected, but the probability of its disconnection will decrease ac­ eliminate passengers’ travel and their destination. They may only affect
( js ) the movement of some passengers and increase their travel time.
cording to the expression γ γ−js 1 .
Excessive travel time for travelers who have trips such as business and
The γjs value can be calculated using an iterative algorithm that educational trips, and it is necessary to be at the destination at a specific
removes one node from the network each time and counts the number of time, will cause dissatisfaction. Since business and educational trips are
node-destinations added.Finally A = A − (m, n). a significant part of the peak hour trips, this article evaluated urban TNR
′

About Isfahan, it is assumed that the available resources have the in short-term non-recurring traffic congestions.
maximum ability to improve the traffic of 30 streets. We calculate the First, the concept of connectivity in graph theory was developed
ENH index from the following Equation for all the streets of the trans­ based on travel time and passenger satisfaction threshold. When an
portation network and rank the streets based on the values of this index. accident occurs on one street (intersection), at a specific satisfaction
Then we introduce the first thirty streets to improve the traffic situation. threshold, the number of unsuitable network destination nodes and the

11
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

Fig. 15. Location of Isfahan’s TNR improvement projects.

Table 4
effect of each project on Isfahan’s TNR.

number of unsuitable trips of network passengers can be calculated resilience of the entire city transportation network was evaluated. The
using the proposed correlation criteria. Accordingly, the importance of satisfaction threshold of Isfahan transportation network users was
streets and intersections in the transportation network can be deter­ considered at the peak of 1.2, which means that the network passengers
mined. Also, the network’s resilience was defined based on the number at the time of the accident considered a maximum of 20% increase in
of 1-arc(node) connected NDs. their travel time acceptable. Results showed that 45% of the NDs are 1-
Then, based on the proposed criteria, the resilience of the Isfahan arc connected, and 58% of network passengers pass through these NDs.
transportation network was evaluated. First, the nodes and arcs of the Also, 50% of the NDs in the city of Isfahan are 1- node-connected, and
transportation network were ranked and categorized according to their 55% of network travelers pass through these NDs. Then by performing
importance. Based on the number of high importance and critical arcs sensitivity analysis, the effect of changing the passengers’ satisfaction
and nodes, the twelve traffic areas of the city were compared. Then the threshold with travel on the urban TNR was studied. Finally, solutions to

12
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

increase the resilience of Isfahan’s roads were presented. The first so­ [17] I.B. Potts, D.W. Harwood, J.M. Hutton, C.A. Fees, K.M. Bauer, L.M. Lucas, R.
J. Frazier, Design guide for addressing nonrecurrent congestion, Transportation
lution was to improve the condition of the street to prevent accidents.
Research Board, 2014.
The second solution was to improve some alleys and side roads to create [18] K. Berdica, Z. Andjic, A.J. Nicholson, Simulating Road Traffic Interruptions–Does it
alternative routes and reduce traffic congestion at the time of the acci­ Matter What Model We Use? Emerald Group Publishing Limited, 2003.
dent. These two solutions have been studied, and their results for the [19] B. Sansó, F. Soumis, Communication and transportation network reliability using
routing models, IEEE Trans. Reliab. 40 (1) (1991) 29–38.
Isfahan roads network were shown. [20] Y. Chung, Assessment of non-recurrent traffic congestion caused by freeway work
Several ideas can be suggested for future research. The first idea is to zones and its statistical analysis with unobserved heterogeneity, Transport Policy
develop the concept of graph connectivity in assessing the urban TNR in 18 (4) (2011) 587–594.
[21] J.M. Morales, Analytical procedures for estimating freeway traffic congestion, ITE
the face of recurring traffic accidents. The release of some intended as­ J. 57 (1) (1987) 45–49.
sumptions can also be considered. For example, all passengers’ aware­ [22] Wicks, D., & Lieberman, E. B. (1980). Development and Testing of INTRAS, a
ness of the accident is one of the assumptions that does not always Microscopic Freeway Simulation Model, Volume 1: Program Design, Parameter
Calibration and Freeway Dynamics Component Development. Retrieved from.
happen in reality. One-way streets and prohibition of turning move­ [23] S. Li, J. Shi, X. Zhang, H. Zhu, G. Meng, Simulation and analysis of non-recurrent
ments undoubtedly have some effects on TNR that can be addressed in traffic congestion triggered by crashes in road networks using a grid-based
future studies. Finally, the development of the graph connectivity approach, Adv. Mech. Eng. 11 (10) (2019), 1687814019884166.
[24] A. Karaer, M.B. Ulak, E.E. Ozguven, T. Sando, Reducing the Non-Recurrent
concept in evaluating the resilience of networks such as air trans­ Freeway Congestion with Detour Operations: Case Study in Florida, Transport.
portation networks in which the issue of travel time is of great impor­ Eng. 2 (2020), 100026.
tance will be a suggestion for future studies. [25] A. Soltani-Sobh, K. Heaslip, J. El Khoury, Estimation of road network reliability on
resiliency: An uncertain based model, Int. J. Disast. Risk Reduct. 14 (2015)
536–544.
Declaration of Competing Interest [26] Z. Zhu, A. Mardan, S. Zhu, H. Yang, Capturing the interaction between travel time
reliability and route choice behavior based on the generalized Bayesian traffic
The authors declare that they have no known competing financial model, Transport. Res. Part B Method. 143 (2021) 48–64.
[27] C.N. Kamga, K.C. Mouskos, R.E. Paaswell, A methodology to estimate travel time
interests or personal relationships that could have appeared to influence using dynamic traffic assignment (DTA) under incident conditions, Transp. Res.
the work reported in this paper. Part C Emerg. Technol. 19 (6) (2011) 1215–1224.
[28] M.Z. Serdar, M. Koç, S.G. Al-Ghamdi, Urban Transportation Networks Resilience:
Indicators, Disturbances, and Assessment Methods, Sustain. Cities Soc. (2021),
Data availability 103452, https://doi.org/10.1016/j.scs.2021.103452.
[29] M.Z. Serdar, M. Koç, S.G. Al-Ghamdi, Urban Transportation Networks Resilience:
Data will be made available on request. Indicators, Disturbances, and Assessment Methods, Sustain. Cities Soc. 76 (2022),
103452.
[30] B. Berche, C. Von Ferber, T. Holovatch, Y. Holovatch, Resilience of public transport
networks against attacks, Eur. Phys. J. B. 71 (1) (2009) 125–137.
References [31] L.A. Schintler, R. Kulkarni, S. Gorman, R. Stough, Using raster-based GIS and graph
theory to analyze complex networks, Netw. Spat. Econ. 7 (4) (2007) 301–313.
[32] A.C. Testa, M.N. Furtado, A. Alipour, Resilience of coastal transportation networks
[1] M. Orui, Y. Suzuki, M. Maeda, S. Yasumura, Suicide rates in evacuation areas after
faced with extreme climatic events, Transp. Res. Rec. 2532 (1) (2015) 29–36.
the Fukushima Daiichi nuclear disaster, Crisis (2018).
[33] N.Y. Aydin, H.S. Duzgun, F. Wenzel, H.R. Heinimann, Integration of stress testing
[2] T. Takabatake, T. Shibayama, M. Esteban, H. Ishii, G. Hamano, Simulated tsunami
with graph theory to assess the resilience of urban road networks under seismic
evacuation behavior of local residents and visitors in Kamakura, Japan, Int. J.
hazards, Adv. Clim. Changes, Global Warming, Biol. Probl. Nat. Hazards, 3rd
Disast. Risk Reduct. 23 (2017) 1–14.
WSEAS Int. Conf. Clim. Changes, Global Warming, Biol. Probl. (CGB ’10) 3rd
[3] N. Sugiyama, T. Nagatani, Multiple-vehicle collision in traffic flow by a sudden
WSEAS Int. Conf. Nat. Hazards (NAHA ’10) 91 (1) (2018) 37–68.
slowdown, Physica A 392 (8) (2013) 1848–1857.
[34] Y. Zhou, J. Wang, H. Yang, Resilience of transportation systems: concepts and
[4] Zanini, M. A., Faleschini, F., Morbin, R., Gastaldi, M., Rossi, R., Pellegrino, C., &
comprehensive review, IEEE Trans. Intell. Transp. Syst. 20 (12) (2019) 4262–4276.
Modena, C. (2016). Interaction between building collapse and road serviceability
[35] M. Akbarzadeh, S. Memarmontazerin, S. Derrible, S.F.S. Reihani, The role of travel
during seismic emergency operations in urban centers. Paper presented at the
demand and network centrality on the connectivity and resilience of an urban
Applied Mechanics and Materials.
street system, Transportation 46 (4) (2019) 1127–1141.
[5] D.A. Hensher, Integrating accident and travel delay externalities in an urban speed
[36] W. Ip, D. Wang, Resilience evaluation approach of transportation networks, Paper
reduction context, Transp. Rev. 26 (4) (2006) 521–534.
presented at the, in: 2009 International Joint Conference on Computational
[6] H. Li, F. He, X. Lin, Y. Wang, M. Li, Travel time reliability measure based on
Sciences and Optimization, 2009.
predictability using the Lempel–Ziv algorithm, Transp. Res. Part C Emerg. Technol.
[37] M Mahdavi, M.H. S, N. Bhouri, G Scemama, Dynamic Resilience of Public
101 (2019) 161–180.
Transport Network: A Case Study for Fleet-Failure in Bus Transport Operation of
[7] L. Zhang, C. Xiong, A novel agent-based modelling framework for travel time
New Delhi, Transport. Res. Proced. 47 (2020) 672–679, https://doi.org/10.1016/j.
reliability analysis, Transportmetr. B Transp. Dyn. 5 (1) (2017) 78–95.
trpro.2020.03.146.
[8] J.C. Chu, S.-C. Chen, Optimization of transportation-infrastructure-system
[38] Z. Liu, H. Chen, E. Liu, W. Hu, Exploring the resilience assessment framework of
protection considering weighted connectivity reliability, J. Infrastruct. Syst. 22 (1)
urban road network for sustainable cities, Physica A 586 (2022), 126465, https://
(2016), 04015008.
doi.org/10.1016/j.physa.2021.126465.
[9] D. Freckleton, K. Heaslip, W. Louisell, J. Collura, Evaluation of resiliency of
[39] Y.-L. Wang, M.-S. Li, M.-H. Yung, Graph-connectivity-based strong quantum
transportation networks after disasters, Transp. Res. Rec. 2284 (1) (2012)
nonlocality with genuine entanglement, Phys. Rev. A 104 (1) (2021), 012424.
109–116.
[40] M. Fiedler, Algebraic connectivity of graphs, Czechoslovak Math. J. 23 (2) (1973)
[10] J.C. Falcocchio, H.S. Levinson, Road traffic congestion: a concise guide (Vol. 7),
298–305.
Springer, 2015.
[41] Nagarajan, H. (2018). On Maximizing Weighted Algebraic Connectivity for
[11] Iida, Y., & Wakabayashi, H. (1989). An approximation method of terminal
Synthesizing Robust Networks. arXiv preprint arXiv:1805.07825.
reliability of road network using partial minimal path and cut sets. Paper presented
[42] Nagarajan, H., Rathinam, S., & Darbha, S. (2015). On maximizing algebraic
at the Transport policy, management & technology towards 2001: selected
connectivity of networks for various engineering applications. Paper presented at
proceedings of the fifth world conference on transport research.
the 2015 European Control Conference (ECC).
[12] R. Smeed, G. Jeffcoate, The variability of car journey times on a particular route,
[43] Nekudari, Z., & Ghasemi, A. (2021). The effect of increasing the algebraic
Traffic Eng. Contr. 13 (6) (1971) 238–243.
connectivity on cascading failures in power grid networks. Paper presented at the
[13] C. Ma, J. Zhou, X.D. Xu, J. Xu, Evolution regularity mining and gating control
2021 26th International Computer Conference, Computer Society of Iran (CSICC).
method of urban recurrent traffic congestion: a literature review, J. Adv. Transport.
[44] Ma, C., Zhang, J., Zhao, Y., Li, X., Yuan, J., Jin, Z., Gu, W. (2014). Pre-configured
(2020) 2020.
polyhedron (p-Poly) with optimal protection efficiency for dual-link failure in
[14] S.-N. Riahi Samani, A. Shetab-Boushehri, R Mahmoudi, Reliable Urban
optical mesh networks. Paper presented at the 2014 6th International Workshop on
Transportation Network Design Problem Considering Recurrent Traffic
Reliable Networks Design and Modeling (RNDM).
Congestions Adv. Smart Coat. Thin Films Future Ind. Biomed. Eng. Appl. 55 (1)
[45] W. Shi, B. Servatius, Analyzing the reliability of degradable networks, J. Comput.
(2021) 69–89.
Sci. Tech. 8 (1) (1993) 77–86.
[15] N. Tempelmeier, U. Feuerhake, O. Wage, E. Demidova, Mining Topological
[46] G. Le, S. Ferdousi, A. Marotta, S. Xu, Y. Hirota, Y. Awaji, B. Mukherjee, Survivable
Dependencies of Recurrent Congestion in Road Networks, ISPRS Int. J. Geoinf. 10
virtual network mapping with content connectivity against multiple link failures in
(4) (2021) 248.
optical metro networks, J. Optic. Commun. Network. 12 (11) (2020) 301–311.
[16] H. Teng, Y. Qi, Detection-delay-based freeway incident detection algorithms,
[47] Chiesa, M., Gurtov, A., Mądry, A., Mitrović, S., Nikolaevkiy, I., Panda, A., Shenker,
Transp. Res. Part C Emerg. Technol. 11 (3-4) (2003) 265–287.
S. (2014). Exploring the limits of static failover routing. arXiv preprint arXiv:
1409.0034.

13
M.-A. Gorji et al. Transportation Engineering 10 (2022) 100152

[48] Chiesa, M., Gurtov, A., Madry, A., Mitrovic, S., Nikolaevskiy, I., Shapira, M., & [51] R. Twumasi-Boakye, J. Sobanjo, Civil infrastructure resilience: state-of-the-art on
Shenker, S. (2016). On the resiliency of randomized routing against multiple edge transportation network systems, Transportmetrica A Transp. Sci. 15 (2) (2019)
failures. Paper presented at the 43rd International Colloquium on Automata, 455–484.
Languages, and Programming (ICALP 2016). [52] Jamakovic, A., & Uhlig, S. (2007). On the relationship between the algebraic
[49] M. Chiesa, I. Nikolaevskiy, S. Mitrović, A. Panda, A. Gurtov, A. Maidry, S. Shenker, connectivity and graph’s robustness to node and link failures. Paper presented at
The quest for resilient (static) forwarding tables, in: Paper presented at the IEEE the 2007 Next Generation Internet Networks.
INFOCOM 2016-The 35th Annual IEEE International Conference on Computer [53] S. Pierre, A. Elgibaoui, A tabu-search approach for designing computer-network
Communications, 2016. topologies with unreliable components, IEEE Trans. Reliab. 46 (3) (1997) 350–359.
[50] Ma, T., Hou, Y., Zhao, X., & Song, D. (2013). A Global Constrained Optimization
Method for Designing Road Networks with Small Diameters. Paper presented at the
IJCAI.

14