Available online at www.sciencedirect.

com

ScienceDirect
Transportation Research Procedia 10 (2015) 714 724

18th Euro Working Group on Transportation, EWGT 2015, 14-16 July 2015, Delft, The
Netherlands

Optimal Critical Infrastructure Retrofitting Model for Evacuation

Planning
Yuval Hadas a, Riccardo Rossi b,*, Massimiliano Gastaldi b, Carlo Pellegrino b, Mariano
Angelo Zanini b, Claudio Modena b
a
Bar-Illan University, Department of Management, Max ve-Anna Webb street, Ramat Gan 5290002, Israel
b
University of Padova, Department of Civil, Environmental and Architectural Engineering, Via Marzolo 9, Padova 35131, Italy

Abstract

In emergency situations, it is necessary to safely evacuate the population in order to save lives. The road network infrastructure is
vulnerable for extreme events, and as a result its ability to supply the required capacity can be seriously hampered. Hence, it is
crucial to identify those critical segments which prohibit safe evacuation, and find an optimal retrofit scheme at the network level
in order to minimize evacuation time. This work introduces an emergency evacuation model that considers infrastructures
vulnerability, event location and magnitude, road network, transportation demand and evacuation areas in order to identify the
critical infrastructures and recommend budget allocation for increasing network capacity for minimizing evacuation time, given
budget alternatives. The infrastructures' analysis was based on the knowledge about mechanics characteristics of a set of bridges,
and about a set of possible seismic scenarios related to the area of interest. By using fragility curves of bridges, the damage state
of them has been assessed. By making a series of hypotheses on how a bridge damage state can influence links functionality,
reduced capacity was assigned to the road network. The result is the estimation of the retrofit cost needed for a specific seismic
scenario, considering the most effective retrofit intervention type, previously identified for each bridge. The infrastructures' analysis
results were used by the evacuation model for optimal budget allocation of retrofits strategies in order to attain a desired evacuation
time frame. The procedure has been applied to an urban network in north Italy.
2015
2015TheTheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B. V.
B.V. This is an open access article under the CC BY-NC-ND license
Selection and peer-review under responsibility of Delft University of Technology.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of Delft University of Technology
Keywords: Evacuation; Emergency; Network rehabilitation; Earthquake; Road network capacity; Bridges; Accessibility

1. Introduction

In emergency situations it is necessary to safely evacuate the population, if such an action will save lives. The road
network infrastructure is vulnerable for extreme events, and as a result its ability to supply the required capacity, when

2352-1465 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of Delft University of Technology
doi:10.1016/j.trpro.2015.09.025
 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724 715

needed most, can be seriously hampered. Hence, it is crucial to identify those critical segments which prohibit safe
evacuation, and find an optimal retrofit scheme at the network level in order to minimize evacuation time. The main
aim of this work is to develop an emergency evacuation model that considers infrastructures vulnerability (in particular
bridges), event location and magnitude, road network, transportation demand and evacuation areas in order to identify
the critical infrastructures and recommend budget allocation for increasing network capacity for minimizing
evacuation time, given budget alternatives. Evacuation analysis requires a multidisciplinary approach integrating
transportation, structural engineering, operations research, and social sciences. We focus our work in particular in the
field of seismic vulnerability assessment of network elements like bridges, considered to be the most vulnerable
elements in transportation networks (Nicholson & Du 1997; Franchin, Lupoi, et al. 2006; Franchin, Pinto, et al. 2006),
and the field of network design. The paper is organized as follows: 1) an overview of the theoretical background
required for this type of analyses is described, 2) a description of the integrated procedure with some details about its
components is provided, 3) a case study, and 4) Concluding remarks and future research directions.

2. Theoretical background

This section summarizes the main concepts related to the structural and transportation key issues used in this work:
in particular, r the first part introduces some concepts on seismic vulnerability assessment of infrastructures whereas
the second describes the main existing literature issues concerning evacuation optimization models and approaches.

2.1. Seismic vulnerability assessment

The efficiency and reliability of a transportation system have a significant influence on the economy of a territory;
indeed, the system must be able to guarantee accessibility and allow the safe and smooth movement of people and
goods. With reference to emergency situations, the first requirement is to investigate the effects on infrastructures by
extraordinary events (specifically, earthquakes), and to identify the connections between these physical and
mechanical impacts and the functional characteristics both of single components and of the road network as a whole.
It is important that the transportation system remain operative or that its function be repaired or restored as soon as
possible (Nicholson & Du 1997; Franchin, Lupoi, et al. 2006). In particular, past experience has shown too often that
earthquake damage to road network components (e.g., bridges, tunnels, retaining walls, etc.) can severely interrupt
traffic flow, thus negatively impacting the economic activity of a region as well as on post-earthquake emergency
response, evacuation and recovery activities (Franchin, Pinto, et al. 2006; Lupoi & Franchin 2006; Schotanus et al.
2004; Franchin, Lupoi, et al. 2006). Past works have focused on seismic performance assessment of individual
components of the road network (Banerjee & Shinozuka 2007), and neglected to pay attention to system performance
assessment and to the optimal economic allocation in the network e in order to improve/retrofit the components
(Pellegrino & Modena 2010; Zanardo & Pellegrino 2004; Gastaldi et al. 2013; Carturan et al. 2013; Banerjee &
Shinozuka 2007), which is crucial for fast evacuation of the population, if needed. Hence, it is required to assess the
seismic vulnerability of road network elements; in this work we refer in particular to bridges. Bridges have been
proven to be the most vulnerable elements in transportation networks during earthquakes (Auza et al. 2010; Banerjee
& Shinozuka 2007; Pellegrino & Modena 2010; Zanardo & Pellegrino 2004), therefore their seismic vulnerability
assessment is necessary for a proper planning of the emergency response and to define priority on retrofit
interventions. Fragility curves allow assessing bridge seismic vulnerabilities (Carturan & Zanini 2014; Zanini &
Pellegrino 2013; Lee et al. 2007; Padgett & DesRoches 2008; Moschonas & Kappos 2009; Shinozuka et al. 2003),
taking into account uncertainties of the variables and using probabilistic distributions to describe the properties of the
materials composing the structure. These curves can be developed empirically as well as analytically. Empirical curves
are usually developed based on the damage reports from past earthquakes. Whereas actual bridge damage and ground
motion data are not available, analytical curves can be derived to assess the performance of bridges (Nielson &
DesRoches 2007). As described in Risk-UE 2004 (Mouroux & Brun 2006) reports, empirical fragility curves are
usually based on bridge damage data from past earthquakes; analytical fragility curves are instead developed through
seismic analysis of the structure.
716 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724

2.2. Evacuation optimization

Evacuation planning can be related to the network design problem, the facility location problem, and network flow
models. The network design problem (Magnanti & Wong 1984) is a set of problems designed to construct networks
with different objective functions in mind, given the flow can be served by a network constrained by capacity. Facility
location problems (Nagy & Salhi 2007) , aims at locating a set of facilities, both serving and being served, in a network,
in order to achieve an objective function with a set of constraints (Avella & Boccia 2009) . Most evacuation planning
models are dealing with predetermined networks, such as the model by Sherali et al. (Sherali et al. 1991) , a location-
allocation model that minimizes evacuation time but disregards costs, or a model by Xie et al. (Xie et al. 2010) that
increases network capacity for evacuation by lane reversal and optimizing crossing. Such models are also aiming at
routing design such as hierarchical multi-objective evacuation planning (Fang et al. 2011) , evacuation in urban areas
(Bretschneider & Kimms 2011) , or building evacuation routes (Pursals & Garzn 2009; Choi et al. 1988). Other
models were developed to locate relief facilities in a known network (Balcik & Beamon 2008) , shelters location (Li
et al. 2012) , or facility location under demand uncertainty (Murali et al. 2012; Ng & Waller 2010). Network flow
models, such as the maximum-flow and minimum-cost problems (Hillier & Lieberman 2005) are well known
problems that find the total flow from origin to destination (the former), or the minimal cost for flow from origin to
destination, given costs associated with arcs and nodes (the later). These models assume costs per unit, rather than
construction costs associated with network design problems and facility location problems. Hadas and Laor (Hadas &
Laor 2013) were the first to present a model for the design of an optimal network in terms of minimizing both
evacuation time and network constructions costs. The model requires: 1) a set of construction alternatives, each
assigned with cost and capacity, 2) demand in the form of population concentrations as well as construction costs,
and, 3) supply in the form of evacuation sites, defined with capacity and construction costs. Based on budget
constraints, the model minimizes the evacuation time and budget allocation. The proposed model introduces a revised
optimization model that differs from the original model as followed: 1) it considers retrofit alternatives, 2) partial
evacuation is permitted, and 3) demand locations are known.

3. Integrated procedure

The framework of the integrated procedure is composed by three basic components: 1) bridges information system
(BIS), 2) seismic information system (SIS), 3) transportation and land-use information system (TLIS). Data archived
in the information systems are used in two sequential processes: 1) stochastic damage state assessment, concerns the
assessment of how bridge damage state (and consequently link damage state) affects link functionality as a
consequence of earthquake events, and 2) transportation and land-use system analysis, refers to the tasks aimed to the
assessment of transportation and land-use system performance indexes in the case of evacuation.

3.1. Bridges Information System

System components potentially subjected to risk, in road network risk assessment, are bridges, tunnels, slopes,
retaining walls and roads. In this analysis only bridges were taken in account since they have been considered as the
critical elements of the network. Each exposed component (bridges) is surveyed, its fragility parameters are evaluated,
and stored into an information system. For a proper analysis of the potential bridges criticalities due to an earthquake
occurrence it is necessary to know bridge physical and geometrical characteristics (Pellegrino & Zanini 2014), which
are essential input data for the fragility characterization: typically, bridge information relates to span length, span
width, number of spans, materials, foundation soil, foundation type, skew angle, year of built, design code, etc. One
element of the BIS refers to bridge fragility (a measure of bridge seismic vulnerability): the Risk-UE classification
(Mouroux & Brun 2006), based on four possible damage states, can be used to estimate bridge fragility on the basis
of bridge geometrical and physical characteristics. Another significant element in the Bridges Information System is
the collection of the possible retrofit intervention costs (Padgett et al. 2010a). They are suitable indexes for assigning
the most effective retrofit intervention (related to characteristics of each bridge) among possible alternatives.
 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724 717

3.2. Seismic Information System

The seismic information system contains data regarding the seismogenetic sources and their parameters to build
seismic hazard map; examples of this information are: geo-localized area of seismogenetic source, focal mechanism,
seismic source depth, annual occurrence ratio. Earthquakes Scenario (SEs) overcome the problem of representing the
actual distribution of shaking over a spatially distributed system (Carturan & Zanini 2014). A SE simulation allows
to evaluate the effects of a seismic event (characterized by magnitude value Mw and epicentral geographic
coordinates) in the neighborhood of the affected site. The procedure for the evaluation of the seismic action spatial
distribution due to a potential SE is based on the application of three models representing some of the most recent
formulations in these fields: the representation of the seismogenic sources (DISS Working Group 2010), a Ground
Motion Prediction Equations (GMPE) that represents seismic magnitude variation as a function of epicentral distance
(Bindi et al. 2011) and the Vs,30 map describing soil characterization for seismic purposes (USGS n.d.). These models
are the main components of the Seismic Information System.

3.3. Stochastic damage state assessment

The level of vulnerability of an infrastructure reflects its attitude in the face of physical damage (physical
vulnerability) and/or loss of functionality (functional vulnerability) occurring as the effects of abnormal
external/internal events depending also on its geometrical and structural characteristics and on the type and intensity
of the event. In the first instance (physical vulnerability), it will be necessary to get at a definition of sensitivity
functions for the single infrastructure, or for classes of infrastructures having similar typological and structural
characteristics. With reference to a generic critical infrastructure element, in simplified terms, it is:

PRi Sti u PSi (1)

Where PRi is the Physical Response of infrastructure element #i (consequence of the event); Sti represents the
Stimulus (in the case of an earthquake, this may depend on the extent of the quake and the distance of the infrastructure
site from the epicentre) to which the element is subjected; PSi is the Physical Sensitivity of infrastructure element #i,
which measures the likelihood of its being affected by the event (physical vulnerability). The sensitivity functions can
be achieved, adopting two approaches: 1) assessment from pieces of qualitative and/or quantitative evidence, and 2)
development of computation models. In our case, bridge damage states are defined according to bridge fragility curves
(a fragility curve is a representation of the probability of exceeding a certain level of damage state) with the Montecarlo
random number generation. A random number is generated for an intensity measure value: this number can define
five bridge damage states according to the position between the four curves.

3.4. Transportation and land-use system analysis

With reference to evacuation modelling the following data are requested: 1) post-earthquake supply model
(damaged network), and 2) post-earthquake demand model (O/D matrices in the case of evacuation both in night and
day time). With regard to post-earthquake condition of the system, two main problems arise: 1) to estimate the travel
demand during evacuation as a function of the modification of land-use/transportation system (as a consequence of
the earthquake). In this assessment, the evaluation of the variation in production and attraction indexes of Origin and
Destination zones is of the primary interest. 2) to evaluate the supply system (road network) functional deterioration
as a function of the damage state of the infrastructure: the physical response (consequence) of the critical element of
the network (arc or node) must be related, by means of suitable functional forms, to a capacity (but in similar way to
other parameters as, for instance, allowed speed, vehicle weight, etc.) defined as the maximum number of vehicles
served per unit of time by this same element; a variation in capacity is therefore definable as the functional response
of the infrastructure element to a change in its physical characteristics. The functionality of an element is likely to
change as the consequence of a certain event (by which the physical response is induced), and this represents the
718 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724

functional vulnerability of the element. The transition from stimulus (event) to Functional Response (translated in
terms of capacity) occurs according to a simplified schema of this type:

Ci FRi VuFui u PRi VuFui u  Sti u PSi  (2)

Where Ci is the capacity function of element #i, or rather its functional response; VuFui is the functional sensitivity
of infrastructure element #i, measuring the likelihood that it will undergo the effects of the stimulus and thus
representing its functional vulnerability. In this situation, therefore, the physical response of the infrastructure assumes
the role of input, as the functional conditions of the single element are evaluated according to a suitably defined
capacity function.

3.5. Retrofit strategy

A retrofit strategy is a set of possible retrofits coupled with estimated cost and estimated capacity. For each bridge
belonging to the transportation network to be analyzed, an evaluation of retrofit intervention types has to be identify.
The most common retrofit intervention types are the insertion of shear keys, elastomeric bearings, seat extenders,
restrainer cables or piers jacketing. Then, on the basis of the vulnerability outcomes derived from a SE simulation for
each bridge, the respective damage states occurrence probabilities can be used to evaluate the retrofit cost, using the
algorithm described in (Padgett & DesRoches 2008). The result given by the procedure is the estimation of the retrofit
cost needed for a specific SE, considering the most effective retrofit intervention type, previously identified for each
bridge. Each retrofit intervention is characterized by a set of corrective coefficients to be applied for deriving
retrofitted bridge fragility curves and again to evaluate the new Bridge Damage Index in the case of the SE occurrence
on the same retrofitted bridge. Bridge Damage Indexes are then related to Link Damage Indexes by joining a set of
possible reduced link capacity values to the BDI outcomes of the Montecarlo method.

3.6. Evacuation analysis

Assume a network, as illustrated in Figure 1, with: a) set of origin nodes designed to serve as populated districts
each with capacity, b) set of destination nodes designed to serve as evacuation areas such as assembly areas, shelters,
safe zones, each with capacity as well, c) set of given transportation infrastructure, each with capacity, and d) a set of
links (representing critical infrastructures such as bridges and tunnels which can be damaged), each with a set of
retrofit alternatives (construction cost and resulted capacity). We are looking for a recommendation for a set of retrofits
that will minimize costs, given a set of evacuation time alternatives. Evacuation time represents the time needed in
steady state conditions to clear all demand locations. Travel times are not considered, as the aim of the model is to
provide a planning tool for capacity management, not travel time optimization. The capacity can be defined in terms
of vehicles or pedestrians. Figure 1 illustrates a network with nodes 1 and 2 the population concentrations, and nodes
12, 13, and 14 as candidate evacuation sites. Numbers over the links represents capacity. The bold links are the critical
links, for which each has retrofit alternatives. Each alternative comprises capacity and retrofit costs (for example, for
link 3-4, without retrofit the estimated capacity after an event will be 2, while it is possible to increase capacity to 4
or 8, with costs of 10 and 30 respectively).

Let be a graph, with N nodes and A arcs, O N origin set (demand), D N destination candidate set

(supply), such as O D . Also let ^ i, j ` A the arc set, with i, j N . Let Ac A , a subset of all critical
arcs, and K a set of alternatives. Furthermore, we assume that all models parameters and variables are deterministic.

Parameters: Uai , j = capacity of arc i, j A , resulted from an extraordinary event. Dni = demand at node i O
 
. Cai , j , k = retrofit cost of alternative k K for arc i, j Ac . Rai , j , k = additional capacity of alternative
 
 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724 719

k K for arc  i, j  Ac . Cn = construction cost of node i D . Un = capacity of node i D . M = a very

i i
large value. T = evacuation time.

Decision variables: f i , j = flow along arc  i, j  . bi = demand allocated to destination node i D . NEi = non-
evacuees from node i O . Yai , j , k = an indicator if alternative k was selected for arc (i, j). Yni = an indicator if
node i was selected as an evacuation site.

Fig. 1. A simplified evacuation network.

Objective functions
Minimize Ca
( i . j )Ac kK
i , j ,k Yai , j ,k  Cni Yni  NEi M
iD iO
(3)

Subject To
0 d bi d Uni Yni i D (4)

b Dn  NE
iD
i
iO
i i (5)

NEi d Dni i O (6)

fi , j d Uai , j T   i, j  A \ Ac (7)

fi , j d Uai , j   Rau , j ,k Yai , j ,k  T   i, j  Ac (8)
k
n n

j 1
f i , j  f j ,i
j 1
Dni  NEi i O (9)

n n

j 1
f j ,i  f i , j
j 1
bi i D (10)

n n

f
j 1
j ,i  fi , j
j 1
0 i O D (11)

fi , j t 0, fi , j '   i, j  A (12)

Ya i , j ,k d 1   i, j  Ac (13)
k
720 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724

Yai , j ,k ^0,1`   i, j  Ac , k K (14)

Yni ^0,1` i D (15)

T !0 (16)

Objective (3) represents the constructions and retrofit costs, as well as a (very large penalty), represented by M)
for non-evacuees. The use of the third term in the objective function stems from the fact that it is possible not to be
able to evacuate all the population based on the characteristics of the problem. As full evacuation is necessary, it is
essential that the model will reflect that, hence whenever full evacuation is not possible, the objective function will
represents a non-feasible solution. Furthermore, due to the large value of M, full evacuation will be achieved if budget
permits. Constraint (4) restricts supply to facility capacity. Constraint (5) retains that total demand meets total supply.
Constraint (6) restricts non-evacuees. Constraints (7) and (8) define non-critical and critical arcs capacity over time,
respectively. Constraints, (9), (10) and (11) define conservation of flow. Constraint (12) defines integral flow.
Constraint (13) enforces the selection of at most one retrofit alternative. Constraints (14) and (15) define binary
variables, and constraint (16) enforces positive evacuation time. The results of the model are: for each evacuation time
and allocated budget pair, 1) a set of retrofit recommendations, 2) location of selected evacuation sites, along with
assigned capacity, 3) number of non-evacuees within the evacuation time, and 4) assigned flow over links.

4. Case study

A test area has been identified in order to verify the effectiveness of the proposed procedure. The attention has been
focused on an urban area, the Municipality of Conegliano, a town of 40,000 inhabitants located in the northern part
of the province of Treviso, North-Eastern Italy; this area was chosen for its significant seismic hazard. In this test area
there are 51 bridges of various typologies: single span, multi span, concrete, steel, and masonry bridges, straight or
skewed.

4.1. Bridges Information System characteristic

In this analysis only bridges were taken in account since they have been considered as the critical elements of the
network. Nevertheless, the process can be extended to include information about other elements at risk. Data used to
build the Bridge information System were retrieved from a preliminary in situ survey, performed to evaluate and
collect the main physical and geometrical characteristics of bridges, from the Bridge Database of the Department of
Civil Engineering of the University of Padova and from maps. The information used to build fragility curves are span
length, span width, number of spans, materials, foundation soils, foundation type, skew angle, year of building and
design code. In this case study curves were built using the procedure described in Risk-UE Project (Mouroux & Brun
2006). Seismic Information System characteristic

The Seismic Information System has been constructed based on the historical seismicity and geological
substructures of the area surrounding the Conegliano municipality (Meroni et al. 2008; Anselmi et al. 2011; Poli et al.
2008). On the basis of the historical seismicity and the literature review concerning the identification of the existing
seismogenic sources (DISS Working Group 2010) a finite number of scenario earthquakes has been considered for
the execution of the simulations, each one characterized by a specific magnitude level, related to the seismogenic
zones characteristics, and a geographical definition of possible epicenters. Table 1 shows the eight earthquake
scenarios considered in this study. Soil characterization has been carried out by the use of the Vs,30 map of the
Treviso Province (USGS n.d.) and finally, the most recent GMPE formulation (Bindi et al. 2011) has been considered
for the detection of the seismic action spatial distribution in relation to the epicentral distance, quake intensity and soil
characterization.

Table 1. Main features of the eight analyzed earthquake scenarios

 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724 721

Geographical Coordinates Magnitude value Mw

Epicentral distance
Longitude Latitude Mw = 5.8 Mw = 6.0 Mw = 6.3 Mw = 6.6
12.2780 45.8826 2 Km SE #1 SE #2 SE #3 SE #4
12.1786 45.8443 10 Km SE #5 SE #6 SE #7 SE #8

4.2. Land-use and Transportation Information System characteristic

A land-use and Transportation Information System was developed (Rossi et al. 2008) for the Municipality,
characterized by: 1) the supply subsystem (road network) with links' capacities, 2) the demand subsystem, represented
by 59 origin zones and 12 destination zones (evacuation areas), and 3) data concerning designed evacuation areas
(position, capacity and construction costs), as shown in Table 2.

4.3. Stochastic damage state assessment and transportation land-use system analysis

The fragility assessment was performed, first deriving seismic actions for each of them, and then using Montecarlo
method (10,000 iterations) for the definition of the respective BDIs. The random number generated by the Montecarlo
method determines the Bridge Damage Index (BDI). Figure 2 represents the BDIs for some significant SE between
those analyzed.

Results obtained from the bridge damage indexes (BDIs) were coupled with the TLIS to derive the respective Link
Damage States (LDIs) useful for the characterization of the damaged network due to the potential failures of bridges.
A correlation has been made to relate the BDI to the link damage state (by way a so called Link Damage Index, LDI);
according to (Shinozuka et al. 2006) the following relation has been used:

Fig. 2. Bridge Damage Indexes for 4 of the 8 SE simulated

LDI BDI 2
bridge link (17)

The link damage indexes were subsequently used to determine the functionality reduction, caused by bridge
damages, to be applied to the damaged network links: in particular 100% functionality was considered in case of none
or slight damages, 50% for moderate damage condition, 25% for extensive damage state and 0% in case of structural
collapse damage state. Once evaluated the damaged networks for each of the considered earthquake scenarios,
evacuation analysis was performed using the evacuation analysis. Information about bridge damages, spatial
distribution and capacity of harvesting areas, boundary conditions in terms of predetermined maximum timing for
carrying out the evacuation and needed costs for the seismic retrofit of bridges belonging to network links involved
in the evacuation were the main input data for the simulations.
722 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724

After the definition of the damaged networks, the first step of the evacuation analysis is the identification, capacity
estimation and setting up costs evaluation of the evacuation areas in which citizens will be directed during the
immediate post-quake evacuation operations. Coneglianos Municipality evacuation areas are described in Table 2.

Table 2. Evacuation areas identified in the Conegliano Municipality

Evacuation areas Area (sqm) Capacity (units) Set-up costs (euro)
EA 65 7,500 2,500 15,000
EA 71 4,500 1,500 9,000
EA 94 4,000 1,333 8,000
EA 98 9,500 3,167 19,000
EA 102 4,400 1,467 8,800
EA 107 3,000 1,000 6,000
EA 108 7,700 2,567 15,400
EA 113 3,500 1,167 7,000
EA 114 1,100 367 2,200
EA 189 8,000 2,667 16,000
EA 190 9,000 3,000 18,000
EA 191 5,000 1,667 10,000

4.4. Retrofit strategy

The costs for seismic retrofit of bridges used for evacuation operations were then estimated. The procedure
proposed by Padgett et al. (Padgett et al. 2010b) was applied to 8 of the 51 bridges belonging to the analyzed network;
they are the most important ones (all multi-span bridges) and also actually needs for retrofitting. This decision was
taken since the majority of the other bridges are single-span structures, in many case buried shallow ones,
characterized by lower seismic fragilities. Table 3 represents the residual functional capacity assigned to the 8 bridges
in the as- built state and in the case of retrofitted configuration.

Table 3. Bridge Residual capacities in as-built and retrofitted configurations

Bridge As-built Bridge Residual traffic flow capacity (%)
ID SE #1 SE #2 SE #3 SE #4 SE #5 SE #6 SE #7 SE #8
10-230-P 0 0 0 0 25 25 25 0
10-232-P 0 0 0 0 25 25 0 0
10-240-P 25 25 25 0 100 50 50 25
10-241-P 25 25 25 0 25 25 25 0
Bridge Retrofitted Bridge Residual traffic flow capacity (%)
ID SE #1 SE #2 SE #3 SE #4 SE #5 SE #6 SE #7 SE #8
10-230-P 25 25 25 25 100 50 50 25
10-232-P 25 25 25 25 50 50 50 25
10-240-P 100 100 50 50 100 100 100 100
10-241-P 100 100 50 50 100 100 50 50

4.5. Results

The evacuation optimization model was executed for each SE, for a range of 10 180 minutes evacuation time
alternatives. Execution time was measured in minutes, due to the relativity small sized problem. Each execution
provided the retrofit budget required to attain the desired evacuation time, the selected evacuation areas, the selected
retrofit scheme, the flow distribution and non-evacuees. For evacuation times lower than 60 minutes, retrofits must be
performed in order to provide safe evacuation. On the other hand, for longer evacuation times, it was found that no
retrofits were required. Such results are reasonable, as the faster the required evacuation, the higher required network
capacity. Figure 3 provides a GIS representation of SE #4, for 30 minutes evacuation time.
 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724 723

5. Conclusion and future works

In this paper an integrated procedure for infrastructures (in particular bridges) retrofitting based on a multi-
disciplinary, at the network level approach has been proposed. The main objective of the work is the design and
implementation of an integrated procedure for the identification of optimal evacuation strategies on a road network to
minimize the impact produced by extraordinary events, in particular earthquakes. The procedure allows assessing
consequences of a seismic event in terms of evacuation time and uses these assessments to identify a priority retrofit
order over infrastructures. Further analyses are necessary in order to generalize the results: 1) to improve the
simulation of the seismic scenario; 2) to improve the procedure for obtaining fragility curves, via non-linear finite
element methods; 3) to embed other vulnerable elements in the analysis (buildings, slopes, tunnels, etc.); 4) to consider
the demand variability as a consequence of the seismic event in the post emergency situation; and 6) to develop a
stochastic, multi-objective optimization model.

Fig. 3. SE#4 morning, optimization output for an evacuation time of 30 minutes.

References

Ahuja, R.K., Magnanti, T.L. & Orlin, J.B., 1993. 1HWZRUNIORZVWKHRU\DOJRULWKPVDQGDSSOLFDWLRQV, Englewood Cliffs, N.J.: Prentice Hall.
Anselmi, M. et al., 2011. Seismicity and velocity structures along the south-Alpine thrust front of the Venetian Alps (NE-Italy). Tectonophysics.
Auza, P., Jayakrishnan, R. & Shinozuka, M., 2010. Using Mesoscopic Traffic Simulation in a Seismic Risk Analysis Framework Applied to a
Downtown Los Angeles Network. Transportation Research Board .
Avella, P. & Boccia, M., 2009. A cutting plane algorithm for the capacitated facility location problem. Computational Optimization and
Applications, 43(1), pp.3965.
Balcik, B. & Beamon, B.M., 2008. Facility location in humanitarian relief. International Journal of Logistics: Research & Applications, 11(2),
pp.101121.
Banerjee, S. & Shinozuka, M., 2007. Nonlinear static procedure for seismic vulnerability assessment of bridges. Computer-Aided Civil and
Infrastructure Engineering, 22, pp.293305.
Bindi, D. et al., 2011. Ground motion prediction equations derived from the Italian strong motion database. Bulletin of Earthquake .
Bretschneider, S. & Kimms, A., 2011. A basic mathematical model for evacuation problems in urban areas. Transportation Research Part A: Policy
and Practice, 45(6), pp.523539.
Carturan, F., Pellegrino, C. & Rossi, R., 2013. An integrated procedure for management of bridge networks in seismic areas. Bulletin of Earthquake
.
Carturan, F. & Zanini, M., 2014. A unified framework for earthquake risk assessment of transportation networks and gross regional product. Bulletin
of earthquake .
724 Yuval Hadas et al. / Transportation Research Procedia 10 (2015) 714 724

Choi, W., Hamacher, H.W. & Tufekci, S., 1988. Modeling of building evacuation problems by network flows with side constraints. European
Journal of Operational Research, 35(1), pp.98110.
DISS Working Group, 2010. Database of Individual Seismogenic Sources (DISS), Version 3.1.1: A compilation of potential sources for earthquakes
larger than M 5.5 in Italy and surrounding areas.
Fang, Z. et al., 2011. Hierarchical multi-objective evacuation routing in stadium using ant colony optimization approach. Journal of Transport
Geography, 19(3), pp.443451.
Franchin, P., Lupoi, A. & Pinto, P., 2006. On the role of road networks in reducing human losses after earthquakes. Journal of earthquake
engineering.
Franchin, P., Pinto, P. & Schotanus, M., 2006. Seismic loss estimation by efficient simulation. of earthquake engineering.
Gastaldi, M., Rossi, R. & Vescovi, R., 2013. Analysing the capacity of a transportation network. A general theoretical approach. European
Transport - Trasporti Europei, pp.116.
Hadas, Y. & Laor, A., 2013. Network design model with evacuation constraints. Transportation Research Part A: Policy and Practice, 47(0), pp.1
9.
Hillier, F.S. & Lieberman, G.J., 2005. Introduction to operations research 8th ed., Boston: McGraw-Hill Higher Education.
Lee, S., Kim, T. & Kang, S., 2007. Development of fragility curves for bridges in Korea. KSCE Journal of Civil Engineering.
Li, A.C.Y. et al., 2012. Shelter location and transportation planning under hurricane conditions. Transportation Research Part E: Logistics and
Transportation Review, 48(4), pp.715729.
Lupoi, G. & Franchin, P., 2006. Seismic fragility analysis of structural systems. Journal of Engineering .
Magnanti, T.L. & Wong, R.T., 1984. Network Design and Transportation Planning: Models and Algorithms. Transportation Science, 18(1), p.1.
Meroni, F., Pessina, V. & Bernardini, A., 2008. Damage risk and scenarios in the Veneto-Friuli area (NE Italy). Boll. Geof. Teor. Appl.
Moschonas, I. & Kappos, A., 2009. Seismic fragility curves for greek bridges: methodology and case studies. Bulletin of Earthquake .
Mouroux, P. & Brun, B. Le, 2006. RISK-UE project: an advanced approach to earthquake risk scenarios with application to different european
towns. Assessing and Managing Earthquake Risk.
Murali, P., Ordez, F. & Dessouky, M.M., 2012. Facility location under demand uncertainty: Response to a large-scale bio-terror attack. Socio-
Economic Planning Sciences, 46(1), pp.7887.
Nagy, G. & Salhi, S., 2007. Location-routing: Issues, models and methods. European Journal of Operational Research, 177(2), pp.649672.
Ng, M. & Waller, S.T., 2010. Reliable evacuation planning via demand inflation and supply deflation. Transportation Research Part E: Logistics
and Transportation Review, 46(6), pp.10861094.
Nicholson, A. & Du, Z., 1997. Degradable transportation systems: an integrated equilibrium model. Transportation Research Part B:
Methodological.
Nielson, B. & DesRoches, R., 2007. Analytical seismic fragility curves for typical bridges in the central and southeastern United States. Earthquake
Spectra.
Padgett, J., Dennemann, K. & Ghosh, J., 2010a. Risk-based seismic life-cycle costbenefit (LCC-B) analysis for bridge retrofit assessment.
Structural Safety.
Padgett, J., Dennemann, K. & Ghosh, J., 2010b. Risk-based seismic life-cycle costbenefit (LCC-B) analysis for bridge retrofit assessment.
Structural Safety.
Padgett, J. & DesRoches, R., 2008. Methodology for the development of analytical fragility curves for retrofitted bridges. Earthquake Engineering
& .
Pellegrino, C. & Modena, C., 2010. Analytical model for FRP confinement of concrete columns with and without internal steel reinforcement.
Journal of Composites for Construction.
Pellegrino, C. & Zanini, M., 2014. Contribution of in situ and laboratory investigations for assessing seismic vulnerability of existing bridges.
Structure and .
Poli, M. et al., 2008. Seismogenic sources responsible for destructive earthquakes in north-eastern Italy. Boll. Geof. Teor. Appl.
Pursals, S.C. & Garzn, F.G., 2009. Optimal building evacuation time considering evacuation routes. European Journal of Operational Research,
192(2), pp.692699.
Rossi, R. et al., 2008. Development of an information system supporting transport planning. In Proceedings of the 10th International Conference
on Applications of Advanced Technologies in Transportation.
Schotanus, M.I.J. et al., 2004. Seismic fragility analysis of 3D structures. Structural Safety, 26, pp.421441.
Sherali, H.D., Carter, T.B. & Hobeika, A.G., 1991. A location-allocation model and algorithm for evacuation planning under hurricane/flood
conditions. Transportation Research Part B, 25(6), pp.439452.
Shinozuka, M. et al., 2006. Cost-effectiveness of seismic bridge retrofit. international conference on bridge .
Shinozuka, M., Murachi, Y. & Dong, X., 2003. Effect of seismic retrofit of bridges on transportation networks. Earthquake Engineering .
USGS, Seismic Hazard Analysis Tools.
Xie, C., Lin, D.-Y. & Travis Waller, S., 2010. A dynamic evacuation network optimization problem with lane reversal and crossing elimination
strategies. Transportation Research Part E, 46(3), pp.295316.
Zanardo, G. & Pellegrino, C., 2004. Performance evaluation of short span reinforced concrete arch bridges. Journal of Bridge .
Zanini, M. & Pellegrino, C., 2013. Seismic vulnerability of bridges in transport networks subjected to environmental deterioration. Bulletin of
Earthquake .