Documente Academic
Documente Profesional
Documente Cultură
Gian Paolo Cimellaro, A.M.ASCE 1; Chris Renschler 2; Andrei M. Reinhorn, F.ASCE 3; and Lucy Arendt 4
Abstract: In recent years, the concept of resilience has been introduced to the engineering field in particular related to disaster mitigation and
management. However, the built environment is only part of the elements that support community functions. Maintaining community func-
tionality during and after a disaster, defined as resilience, is influenced by multiple components. The paper is proposing a framework for
measuring community resilience at different spatial and temporal scales. Seven dimensions are identified for measuring community resil-
Downloaded from ascelibrary.org by University Of Florida on 12/04/17. Copyright ASCE. For personal use only; all rights reserved.
ience: population and demographics, environmental and ecosystem, organized governmental services, physical infrastructures, lifestyle and
community competence, economic development, and social-cultural capital. They are summarized with the acronym PEOPLES. Each di-
mension is characterized by a corresponding performance metric that is combined with the other dimensions using a multilayered approach.
Therefore, once a hybrid model of the community is defined, the proposed framework can be applied to measure its performance against any
type of extreme event during emergency and in long term postdisaster phases. A resilience index can be determined to reflect all, or part, of the
dimensions influencing the events. Several applications of part of such framework can already be found in literature for different types of
infrastructures, physical and organizational (e.g., gas network, water distribution networks, health care facilities). The proposed framework
can be used as decision support by stakeholders and managers and it can help planners in selecting the optimal restoration strategies that
enhance the community resilience index. DOI: 10.1061/(ASCE)ST.1943-541X.0001514. © 2016 American Society of Civil Engineers.
Author keywords: Resilience; Disaster resilience; Infrastructures; Performance metric; Community resilience; Environment and
ecosystem; Organized governmental services; Physical infrastructure; Structural safety and reliability.
Introduction and Definition of Resilience of systems and of their ability to absorb change and disturbance
and still maintain the same relationships between populations or
Recent disasters around the world have shown clearly that not all state variables. Stability represents the ability of a system to return
the threats can be averted. Modern societies are trying to enhance to an equilibrium state after a temporary disturbance; the more rap-
their resilience against extreme events after realizing that they can- idly it returns to equilibrium and the less it fluctuates, the more
not prevent every risk from being realized, but rather they must stable it would be.” An extended literature review about resilience
manage risks and adapt minimizing the impact on population and has been assembled in the past (Table 1), with each contribution
their support systems. adding new nuances. Primarily resilience has been defined in context
The concept of resilience has several definitions, because of its to the speed of systems to go towards equilibrium (Adger 2000) or
broad utilization in ecology, social science, economy and engineer- capability to cope and bounce back, ability to adapt to new situations
ing fields, with different meanings and implications. As Klein et al. (Comfort 1999), be inherently strong, flexible, and adaptive (Tierney
(2003) stated , the term derives from the Latin word resilio, which and Bruneau 2007), or ability to withstand external impacts and
means to jump back. The term was used first in psychology and recover with least outside interferences (Mileti 1999). After the origi-
psychiatry in 1940s, and it is primarily accredited to Norman nal definition of resilience in ecological systems, the word expanded
Garmezy, Emmy Werner, and Ruth Smith (Garmezy 1974; Werner its meaning to engineering, social, and economical fields.
and Smith 1989). Later, the concept of resilience was established in Engineering resilience is defined as the capability of a system to
the field of ecology by Holling (1973), who stated that the resil- maintain its functionality and to degrade gracefully in the face of
ience of an ecological system is “a measure of the persistence internal and external changes (Allenby and Fink 2005). The main
difference in defining resilience arises between the engineering ap-
1
Visiting Professor, Dept. of Structural and Environmental Engineering, proach, in which resilience occurs by recovering towards a previous
Univ. of California, Berkeley, CA 94703 (corresponding author). E-mail: or an improved stable state (Bruneau et al. 2003), and the ecological
gianpaolo.cimellaro@polito.it
2 approach, in which resilience is achieved by moving towards a dif-
Associate Professor, Dept. of Geography, State Univ. of New York,
Buffalo, 116 Wilkeson Quad, Buffalo, NY 14261. E-mail: rensch@
ferent system state (Handmer and Dovers 1996).
buffalo.edu Social resilience is defined as the ability of groups or societies to
3
Professor Emeritus, Dept. of Civil, Structural and Environmental cope with external stresses and disturbances because of social,
Engineering, State Univ. of New York, Buffalo, 135 Ketter Hall, North political, and environmental change (Adger 2000).
Campus, Buffalo, NY 14260-4300. E-mail: reinhorn@buffalo.edu Economic resilience is defined as the inherent ability and adap-
4
Associate Professor of Management, Associate Dean, College of Pro- tive response that enables individual business firms and entire
fessional Studies, Director, Austin E. Cofrin School of Business, Univ. of regions to avoid maximum potential losses (Rose and Liao 2005).
Wisconsin-Green Bay, 2420 Nicolet Dr., Green Bay, WI 54311. E-mail: It has primarily been studied in context to seismic response and
arendtl@uwgb.edu
Note. This manuscript was submitted on September 13, 2015; approved
recovery (Tierney 1997), community behavior (Chang and Shinozuka
on January 12, 2016; published online on April 15, 2016. Discussion period 2004), and disaster hazard analysis (Rose 2004).
open until September 15, 2016; separate discussions must be submitted for Research advancements have proven that resilience should be
individual papers. This paper is part of the Journal of Structural Engineer- addressed at the large-scale level and not just locally. Bruneau et al.
ing, © ASCE, ISSN 0733-9445. (2003) identified four types of resilience that should be adequately
measured: technical, organizational, social, and economical Bruneau (2003) by proposing a metric of system functionality
(TOSE). Technical and economical resilience are primarily related (Q) that is evaluated comparing the extreme events scenario with
to physical systems, whereas organizational and social resilience the pre-event conditions, and they applied the method to a case
are related to society and nonphysical systems. study of the water system in Memphis, TN. Miles and Chang
• Technical resilience describes the capability of a system to func- (2006) presented a comprehensive functionality restoration model
tion and perform adequately. that establishes the relationships between a community’s house-
• Organizational resilience describes the ability of the organization hold, business, and lifeline networks. The same year Cagnan et al.
to manage the system. For example, measures of organizational (2006) developed a discrete event simulation for modeling the post-
resilience could include how well emergency units function, how earthquake restoration of an electric power system. The resilience
quickly spare parts are replaced, or how quickly repair crews are concept as input to decision support methodologies has been
able to reach the affected components of a system. applied to hospitals (Cimellaro et al. 2010b; Cimellaro and Pique
• Social resilience concerns how well society copes with the loss 2016), lifeline structures (Ouyang and Duenas-Osorio 2011;
of services because of a disaster. For example, social resilience Cimellaro et al. 2014b, c), and cities (Chang et al. 2014) using
can become the most critical dimension of global resilience be- different optimization methods based on economic (Chang and
cause of severe blackouts during a disaster. Shinozuka 2004), downtime (Cagnan et al. 2006), or multicriteria
• Economic resilience describes the capability to reduce both in- analysis (Javanbarg et al. 2012).
direct and direct economic losses (Rose and Liao 2005). Several methods for the quantification of infrastructures’
Following the initial resilience framework by Bruneau et al. resilience have been proposed that can be grouped in probabilistic
(2003), other frameworks have been developed, expanding and methods (Miller-Hooks et al. 2012; Queiroz et al. 2013), graph
identifying different metrics to quantify resilience. For example, theory methods (Berche et al. 2009; Dorbritz 2011), fuzzy
Chang and Shinozuka (2004) refined the method proposed by logic methods (Heaslip et al. 2010), and analytical methods
This paper suggests a novel framework to evaluate resilience of resilience index R, therefore it should be maintained fixed when
a community and to assess the performance of critical infrastruc- comparing different scenarios.
tures and their interdependencies although taking into account the
influence of the human behavior, societal, organizational, and eco-
nomic issues. The framework is based on seven major groups of Seven Dimensions of Resilience
characteristics, defined here as dimensions, which can measure
resilience at different scales. These are as follows: (1) population The proposed framework expands the initial research at the Multi-
and demographics, (2) environmental and ecosystem, (3) organized disciplinary Center of Earthquake Engineering Research (MCEER)
governmental services, (4) physical infrastructure, (5) lifestyle and and links with the previously identified resilience characteristics
community competence, (6) economic development, and (7) social- (technical, organizational, societal, and economic) and with the
cultural capital; they are identified with the acronym PEOPLES. resilience attributes (r4 : robustness, redundancy, resourcefulness,
The framework can be used for resilience-based design (RBD) and rapidity) (Bruneau et al. 2003; Bruneau and Reinhorn 2007;
at different spatial (e.g., local, regional) and temporal (e.g., emer- Cimellaro et al. 2010b).
gency response, recovery, and reconstruction phase) scales. It can The new framework, identified by the acronym PEOPLES, in-
also be used by decision makers for disaster and postdisaster man- corporates the initial MCEER’s definitions of service functionality
agement, minimizing all the possible consequences following an of community components (assets, services, demographics) and
extreme event, both natural and artificial, allowing the perturbed parameters influencing resilience, all assembled into a layered ap-
system to return to the initial conditions as quickly as possible. proach. The seven dimension groups of the PEOPLES framework
(Renschler et al. 2010, 2011), listed next, are further explained in
this section.
1. Population and demographics;
Mathematical Definition of Resilience 2. Environment and ecosystem;
The definition of resilience used in this paper is the one described 3. Organized government services;
also by Cimellaro et al. (2010a) [in which a resilience index, R, of a 4. Physical infrastructure;
system is defined as the normalized area underneath the function- 5. Lifestyle and community competence;
ality-performance function, QðtÞ, as shown in Fig. 1], although 6. Economic development; and
analytically it is defined as 7. Social-cultural capital.
The specific dimensions represent groups of interwoven soci-
Z t þT
OE LC etal, technical, economic and organizational issues. Although other
Rð~rÞ ¼ QTOT ð~r; tÞ=T LC dt ð1Þ definitions of multiple dimensions were described by Rockefeller
tOE
Foundation (Huq et al. 2007), United Nations (2013), the Institute
for Social and Environmental Transition (ISET) (Tyler and Moench
2012), Arup (Silva and Morera 2014), the previously discussed
dimension groups were selected based on similar characteristics.
A resilience index can be established for each of the above
dimensions; however, the whole community resilience would be
influenced by all, or only some dominant dimensions, as it is de-
scribed in the following. The Appendix shows the extended list of
components and subcomponents of the PEOPLES Framework; a
detailed description of each dimension is given in the next para-
graphs. The description of the dimensions follows the order of
the PEOPLES acronym, so it is not based on a specific hierarchy.
Additional details can also be found in Renschler et al. (2010,
2011).
the population and demographic dimension could be the social vul- lated with the aboveground net primary productivity (ANPP) index
nerability index (SVI) (Barry et al. 2011). The domains that form (Pettorelli 2005; Olofsson et al. 2007) that is based on filed
the basis of the SVI are (1) socioeconomic status, (2) household measurements of the biomass accumulation and, therefore, can
composition and disability, (3) minority status and language, and be considered as an indicator of the ecosystem resilience. Several
(4) housing and transportation. The data can be collected from applications can be found in the literature in which the NDVI
the European Census of Population and Housing at the census values obtained from Landsat images have been used to observe
tract level. Each of the domains can be described as per following the restoration of the vegetation after a fire (Diaz-Delgado et al.
variables: 2002) and using time-series analysis (Simoniello et al. 2008).
1. Socioeconomic status (comprising income, poverty, employ- The NDVI index [in Eq. (3)] can be used to quantify the envi-
ment, and education variables); ronmental and ecosystem dimension by comparing the NDVI val-
2. Household composition and disability (comprising age, single ues before and after the event, to determine the variations of
parenting, and disability variables); ecosystem productivity through the space and the time caused by
3. Minority status and language (comprising race, ethnicity, and natural disasters (such as fire, flood, hurricanes, tsunami). Instead,
language proficiency variables); and in other types of disasters (such as blizzards, terrorist attacks), the
4. Housing and transportation (comprising housing structure, variation of this index could be negligible because the vegetation
crowding, and vehicle access variables). density might not be altered although other indicators could be
Each of these variables, except per capita income, could be more relevant.
ranked from highest to lowest across all census tracts, to construct
the SVI. Per capita income should be ranked from lowest to highest Organized Governmental Services
because, unlike the other variables, a higher value indicates less
vulnerability. A percentile rank is then calculated for each census The dimension of organized governmental services includes legal
tract over each of these variables. A percentile rank is defined as the and security services (e.g., police, emergency departments (EDs),
proportion of scores in a distribution that a specific score is greater fire departments, the military) and also, for example, the public
than or equal to. Percentile ranks are calculated by using the health, the hygiene departments, and the cultural heritage depart-
formula: ments. Each of the previously discussed organized government
services play a key role in sustaining societies before and after
Percentile Rank ¼ ðRank − 1Þ=ðN − 1Þ ð2Þ an extreme event.
Key indicators for this dimension include the number of avail-
Where N = total number of data points, and all sequences of ties are able response units and their capacity, if they are opportunely nor-
assigned the smallest of the corresponding ranks. In addition, a malized with respect to the number of residents involved. This
tract-level percentile rank is calculated for each of the four domains dimension can provide a measure on how much the various organ-
based on an across-the-board sum of the percentile ranks of the ized government services participate in emergency preparedness
variables comprising that domain. Finally, an overall percentile planning (e.g., survey), developing a memorandum of understand-
rank for each tract could be calculated as the sum of the domain ing (MOU), and other mutual aid agreements (Tierney 2009).
percentile rankings. This process of percentile ranking (for all var- Other examples of performance metrics for this dimension can
iables, for each domain, and for an overall SVI) is then repeated for be the patient waiting time (WT), that is the time a patient waits
the individual communities. before receiving assistance (Cimellaro et al. 2011), in the case when
Other social vulnerability indexes (SVI) (Cutter 1996; Cutter the organized service is the ED of a hospital. This specific indicator
and Mitchell 2000) can be used as well, such as the one proposed measures the ability of the ED to provide service to all patients after
by Cutter that integrates exposure to hazards with the social con- a disaster. Analytically, the functionality-performance metric (Qo )
ditions that make people vulnerable to them. is given by
WT0
Environmental and Ecosystem Qo ¼ ð4Þ
WT
In the PEOPLES Framework, the environmental and ecosystem
dimension measure the capability of the ecological system to go where WT0 = waiting time in normal operating conditions; and
back to its pre-event condition defined as its basic functionality. WT = waiting time during the emergency.
This dimension measures the capability of an ecosystem to The deficiencies associated with this resilience dimension have
deal with disturbance, but also the amount of disturbance an eco- been observed during the 2010 Haiti Earthquake, in which the lack
system can absorb without considerably varying its processes and of organized government services and orderly control together with
structures. a perception that the government could not deal with the disaster
and Quarantelli 1985; Drabek and McEntire 2002). The emergency There are also other examples for housing units in which a pos-
management system following a disaster involves different groups sible functionality-performance metric might be the proportion of
such as the emergency response teams, the volunteers, the mass housing stock not rated as substandard or hazardous and vacancy
media, and the economic network. These groups, instead of trans- rates for rental housing (Tierney 2009). Examples of functionality-
ferring information in a hierarchic way in the top-down direction, performance metrics for the communication networks might be
use an upward flow of information that is the most preferred direc- (1) the acceptable linkages between official and unofficial informa-
tion of communication used during disasters. In fact, the experience tion sources, (2) the number of ties between the mass media and the
in the field has shown that decentralized networks with flatter emergency management entities, and (3) the sufficiency of mea-
organizations and less hierarchical structure are quicker in respond- sures for communicating the public’s need and information after
the disaster (Tierney 2009).
ing to disasters because they promote a free flow of information
(Simoniello and Quarantelli 1985).
Lifestyle and Community Competence
Physical Infrastructure The lifestyle and community competence dimension deals with
flexibility, creativity, and problem solving skills of a community
The physical infrastructure dimension includes facilities (e.g., hous- through political partnerships (Norris et al. 2008). Principal ele-
ing, commercial and industrial facilities, and cultural facilities) ments of this dimension include collective actions and decision
and lifelines (food supply, utilities, transportation, communication making, collective efficacy and empowerment, and quality of
networks) within a built environment (Cimellaro et al. 2014b). life. This dimension captures both the raw abilities of a community
Although facilities are traditional essential life support for its pop- (e.g., skills to find multifaceted solutions to complex problems
ulation, lifelines are essential utilities which serve communities through the engagement in political networks) and the perceptions
across all jurisdictions such as: (1) energy utilities [e.g., power of a community (e.g., perception to have the ability to do a positive
and natural gas networks (Bruneau et al. 2003; Cimellaro et al. change through a common effort that relies on peoples’ aptitude to
2015b)]; (2) transportation systems (Arcidiacono et al. 2012) resourcefully envision a new future and then move in that direction)
(e.g., highways, railroads, airports, seaports); (3) water, storm- (Brown and Kulig 1996). In fact, the societies that believe that they
water, and sewerage pipelines; (4) communication systems; and can restore, renew, and rebuild themselves are expected to be more
(5) health care facilities (e.g., hospitals) (Cimellaro et al. 2011). determined when facing a disaster or in general, any type of
Functionality of physical infrastructures has an important impact changes.
on the restoration process following a disaster; therefore, the organ- Quality of life surveys can be used as indicators of this percep-
ized government services work actively to restore their functional- tion because they reveal whether people inside the community are
ity. Such interactions are essential in resilient communities. devoted to their community and willing to engage in the activities
For example, following Hurricane Katrina in 2005, after the necessary to keep the community alive, before or after the disaster
evacuation of New Orleans, attention shifted towards the restora- strikes. Examples of performance metrics for the community
tion of the physical infrastructures. The pictures of damages have competence in normal conditions before the disaster might be
been used to communicate to the media in the world the consequen- the number of immigrants or the number of citizens involved in
ces of the hurricane and of the subsequent flood (e.g., collapse of politics.
critical facilities such as churches, schools, and hospitals). The Specific performance metrics for this dimension directly related
critical facilities were not able to provide their services without to the disaster might be the extensiveness of community warning
water and electricity. The damaged schools affected the commun- procedures and plans, measured using, for example, the number of
ity’s self-confidence to overcome the disaster and restore the initial citizens involved or the number of organizational disaster training
functionality. The roads full of debris created an obstacle to the programs. (Tierney 2009).
supply chain, therefore, the economy in the region could not restart,
because even if shops and companies re-opened they could not
Economic Development
be accessible and even if they relocated for a short term, the pre-
vious customers were having some difficulties in finding the new The economic development dimension is composed of both a static
location. and a dynamic assessment. The static assessment is the market
After a disaster, the restoration of physical infrastructures re- activity of the current economy of a community; although the
mains a technical problem that is also related to the sociopolitical dynamic assessment corresponds to the economic development,
events and the economic situation. The resilience dimension of which is the community’s ability to continuously sustaining the
physical infrastructure should also take into account the interde- economic growth. Resilient communities are characterized by
pendencies between the different types of infrastructures and the community’s capacity to replace goods, services, and shift
structures, and (3) the adequacy of plans for postdisaster commer- However, this emotional connection to the community is not nec-
cial restoration (Tierney 2009). Because of these indicators, this essarily related only to the residents in those places (Manzo and
dimension is interdependent with the population and demographics Perkins 2006). For example, several displaced residents of New
dimension. Orleans after Hurricane Katrina expressed the desire to return home
Analytically, one possible functionality-performance metric with a strong place attachment, regardless the job they had and
(QE ) for this dimension is given by the people they knew. These residents are an important resource
for the community because if they will be provided with housing
per inco þ med inco þ emply þ hsg value þ buss þ insurance and employment after the disaster, they will act to restore the
QE ¼ community to the initial condition before the disaster. The citizen
6
participation in community organizations (e.g., religious congrega-
ð6Þ
tions, school and resident associations, neighborhood watches,
and self-help groups) is a way of demonstrating one’s care for the
where per_inco = per capita income; med_inco = median house- community, one’s care for meeting and understanding one’s fellow
hold income; employ = employed civilian population; hsg_value = citizens, and it increases an individual’s circle of influence and per-
median value of owner-occupied housing units; buss = busi- ception of control (Norris et al. 2008).
ness establishments; and insurance = population with health
insurance.
Mathematical Formulation of the Peoples
Framework
Social and Cultural Capital
The social and cultural capital dimension includes numerous General Description of the Methodology and the
subcategories such as (1) education services, (2) child and elderly Community Hybrid Model
services, (3) cultural and heritage services, and (4) community
participation. The key indicators in normal condition for this di- The main part of the methodology consists of developing a com-
mension are: (1) the number of members belonging to the diverse munity hybrid model, coupling the network models, which will be
civil and community organizations and (2) the surveys of leaders used to model the physical infrastructures networks such as the
and their perception. The key indicators in emergency conditions power and the water, with the agent-based models (ABM), which
are (1) the existence of community plans targeting transportation- will be used to model the sociotechnical networks such as the emer-
disadvantaged residents, (2) the adequacy of postdisaster sheltering gency medical technicians and the fire brigade (Fig. 2). Inside the
plans, (3) the adequacy of plans for incorporating volunteers into ABM models, the emotions in the agents (Ortony 2002) will be
official response activities, (4) the adequacy of donations manage- modeled using the extended version of the belief-desire-intention
ment plans, and (5) the community’s plans to manage various (BDI) framework proposed by Zoumpanaki et al. (2010) that has
networks (Tierney 2009). been expanded and adapted to the proposed methodology (Fig. 3).
In relation to disaster phases’ activities, the sociocultural capital Both types of models will be integrated in a hybrid framework
dimension can be measured using the following six components and a matrix approach will be used to describe the interdependen-
suggested in the literature (Mayunga 2009): cies between the different layers. Each layer represents an infra-
1. Participation in voluntary organizations (volunteerism): this com- structure (Fig. 3) and is described by an adjacency matrix A,
ponent was measured using registered non-profit organizations; whereas a D matrix will describe the interdependencies between
2. Involvement in social groups (association densities): the invol- the nodes of the different layers (e.g., DWater → Power) and it
vement in social groups was measured using recreational centers will be obtained using an extended version of the Haimes input-
(bowling centers and fitness centers), golf clubs, and sport output inoperability matrix (IIM) (Haimes et al. 2005). For exam-
organizations; ple, in Fig. 2, the hospital is a node of the EMT layer and it is
3. Civic and political participation: this social capital component interdependent with the power and the water network. Therefore,
was measured using three indicators including registered voters, a D matrix describing the interdependencies between the EMT
civic and political organizations, and census response rates for layer and the water and power layer will be determined using
the decennial population and housing survey; the Haimes model.
4. Religious participation: it was measured using religious The matrix D is composed of constant scalars terms if the sys-
organizations; tem does not change through time. In reality, the interdependent
5. Community attachment: the community attachment component networks might change through the time their interconnectedness
was measured using owner-occupied housing units; and as shown in some applications (Fantini et al. 2014), however the
HYBRID MODEL
vALIDATION
Gas EMT
B.1
Plant
Non- B.2
Physical
lifelines
(ABM) B.2 Gas
Fire
Plant
B.1 brigade
Scenario
analysis
Performance
Robustness measures
Resilience actions
Rapidity P Environmental/
Ecosystems
Performance levels
Redundancy
Economic Physical
Development Infrastructures
Yes
Fig. 2. Methodology for RBD, based on control (feedback loop) approach and hybrid layered model
proposed approach can also be applied in this case. The proposed sociotechnical networks operating in the community will be af-
approach will require substantial computational power if the spatial fected. For example, the road transportation disservice, might limit
and temporal dimensions of the problem increase, therefore, the the capacity of the emergency rescue teams to move and operate in
use of parallel computing is recommended in these cases. Below the community. For example, a water network disservice might
is shown in simple terms how the agent base models and the limit the capacity of the fire brigade to extinguish fires. To study
network model interact in the proposed methodology. Once the these interdependencies the network models and the ABMs need to
hazard is determined (e.g., earthquake event), the corresponding run simultaneously. So the output of the damage analysis in the
damage in the infrastructure networks is determined using fragility network models should be used as input to modify the character-
analysis combined with graph theory to identify the nodes of istics of the ABMs, such as the extension of the environmental
the network that will not be functional following the extreme space (e.g., roads) and the capacity to perform certain actions
event. Because of physical infrastructure disservice, also the (e.g., extinguish fire) of the agents.
Fig. 3. Methodology to model the interdependency and the human behavior within the community hybrid model
Resilience Index and Performance Metrics determined using time series analysis (Cimellaro et al. 2014c) or
from linear algebra manipulation of the Dx→y matrix. Then, the
Once the hybrid model in Fig. 2 is built, it is necessary to identify indices I can be grouped into an infrastructure interdependency ma-
the performance metrics to estimate the resilience of each trix (IM) (Fig. 4). The infrastructures considered in the analysis of
infrastructure. Several approaches exist in literature for hospitals the community are listed in the rows and the columns, whereas each
(Cimellaro et al. 2011), lifeline structures (Ouyang et al. 2012; cell shows the degree of interdependency (from 0 to 1) between
Cimellaro et al. 2014c), and cities (Chang et al. 2014). Once the them. The sum over the columns gives the dependent factor of
proper performance metric is selected, the degree of interdepend- the specific lifeline, and the sum over the rows gives the importance
ency between an infrastructure x and y is described using a factor of a specific lifeline. Ideally, the target is to realize a com-
matrix, Dx→y , which is able to identify the exact location of the munity in which all lifelines are independent, so IM will be an
interdependency in the network (e.g., node or link). However, identity matrix. As discussed previously, the IM can also be used
sometimes it is also useful to identify a global index, I, that to have an unbiased estimation of the weight coefficients to assign
measures the degree of interdependency between the different in- to each infrastructure considered in the layered approach as shown
frastructures, to have a global evaluation of the community perfor- in Fig. 2. Once the spatial and temporal boundaries of the problem
mance and to assign an unbiased evaluation of the weight (or at hand are defined, the performance metrics of all the resilience
importance factor) to each infrastructure. This index can be dimensions are aggregated following the procedure described in
dex
on
es
m
Financial system
Leadership ind
Telecomunicatio
Building service
Transportation
Commodities
Water supply
Food supply
Government
Natural Gas
Wastewater
Oil delivery
Health care
Emergency
Education
Electricity
Business
services
tratment
delivery
the
an affects in this points
Electricity 1.00 0.78 0.79 1.00 0.73 0.73 1.00 0.80 0.80 1.00 0.66 0.87 0.80 0.80 0.60 1.00 13.34
Transport
Oil delivery 0.53 1.00 0.90 0.60 0.40 0.45 0.40 0.45 0.45 0.40 0.60 0.50 0.80 0.60 0.00 0.65 8.73
Leeadership index
Transportation 0.50 0.76 1.00 0.80 0.70 0.53 0.65 0.90 0.87 1.00 0.60 0.80 1.00 0.63 0.60 1.00 12.34
Telecomunication 0.65 0.45 0.30 1.00 0.45 0.30 0.30 0.80 0.30 0.87 1.00 0.54 1.00 0.80 0.30 0.50 9.56
Road
Rail
Air
Natural Gas delivery 0.65 0.00 0.30 0.00 1.00 0.30 0.00 0.60 0.60 0.60 1.00 0.45 0.80 0.63 0.60 0.60 8.13
Maritime
Water supply 0.80 0.87 0.60 0.60 0.80 1.00 0.00 0.60 0.77 0.83 0.70 1.00 0.60 0.40 0.60 0.73 10.89
Wastewater tratment 0.40 0.60 0.53 0.30 0.60 0.60 1.00 0.30 0.30 0.45 0.53 0.60 1.00 0.50 0.30 0.00 8.01
M
Financial system 0.60 0.73 0.60 0.60 0.73 0.60 0.60 1.00 1.00 1.00 0.60 0.60 1.00 0.60 0.00 0.60 10.87 the
Building services 0.45 0.45 0.60 0.45 0.45 0.45 0.45 1.00 1.00 0.60 0.73 0.45 0.80 0.45 0.65 0.70 9.68 Road 1.00 0.78 0.79 1.00 3.57
Downloaded from ascelibrary.org by University Of Florida on 12/04/17. Copyright ASCE. For personal use only; all rights reserved.
Business 0.60 0.60 1.00 0.60 0.60 0.60 0.60 0.60 0.80 1.00 0.30 0.60 1.00 0.00 0.00 0.60 9.50 Rail 0.53 1.00 0.90 0.60 3.03
Emergency services 0.65 0.65 0.87 0.65 0.65 0.65 0.65 0.77 0.77 0.77 1.00 0.63 1.00 0.80 0.60 0.00 11.10 Air 0.50 0.76 1.00 0.80 3.06
Food supply 0.65 0.65 0.65 0.65 0.65 0.65 0.45 0.77 0.50 0.50 0.77 1.00 0.80 0.80 0.60 0.45 10.53
Maritime Transport 0.65 0.45 0.30 1.00 2.40
Government 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.60 1.00 0.65 1.00 1.00 1.00 0.87 1.00 0.53 14.65
Health care 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.80 0.60 0.92 1.00 0.60 0.60 10.52
Education 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.60 0.30 0.60 0.80 0.60 0.60 0.80 1.00 0.60 10.10
Index of subordination 2.68 2.99 2.99 3.40
Commodities 0.63 0.63 0.60 0.75 0.63 0.60 0.60 0.60 0.75 1.00 0.30 0.53 0.60 0.60 0.30 1.00 10.13
Index of
10.3 10.3 10.9 10.2 10.5 10.9 10.8 11.8 11.3 10.7 13.7 10.2
subordination 9.66 8.90 7.75 9.57
1 7 3 0 9 8 0 6 9 7 2 7
((dependence)
p )
Z Z Z tOE þT LC
the paper of Cimellaro et al. (2014c). The global resilience indi-
Rcom ¼ Rð~rÞ=AC dr ¼ QTOT ðtÞ=ðAC T LC Þdtdr
cator at the community level is evaluated using the following AC AC tOE
equation:
ð8Þ
X
R¼ ðRi × wi Þ ð7Þ where Ac = area of the selected region. The contour plot of each
i dimension can be combined with the other plots using a layered
approach. Then, a radar graph is built (Fig. 2) and the internal area
where Ri = calculated using Eq. (1). For example, if it is considered will determine the final score of the resilience index that will be
the physical infrastructure dimension: Ri = resilience indicator of a used to recognize the priority resilience actions to be taken in
specific infrastructure; and wi = weight factor describing the inter- the community.
dependencies between the different indicators. The coefficient wi is
determined using a time-series analysis approach borrowed from
Resilience Performance Levels
the economic field, which is based on the analysis of the cross-
correlation function (CCF). The procedure can be applied to all The objective of performance-based seismic engineering (PBSE) is
the components and subcomponents of the PEOPLES framework to design, construct, and maintain facilities with better damage con-
to take into account the interdependencies between the different trol, coupling the expected or desired performance levels with the
variables. Further details about the methodology can be found in levels of seismic hazard. Generally, the levels focus on the perfor-
Cimellaro et al. (2014c). mances a structure can hold during the shaking and are related
The selection of the proper performance metric for the critical to engineering demand parameters such as deformations. More re-
infrastructures plays a key role in the analysis. Even if a realistic cently, the San Francisco Planning and Urban Research Associa-
and predictive model is developed, the results might be affected by tion (SPUR) (Bonowitz 2009) introduced other definitions of
the selection of the final performance function adopted to evaluate performance levels for infrastructures based on recovery target
the community resilience index using the methodology shown in states combining safety and recovery time. Five performance mea-
Fig. 2. Different innovative approaches to measure functionality sures for buildings have been identified: (1) safe and operational;
are available in the literature and they include ABM, input-output (2) safe and usable during repair; (3) safe and usable after repair;
models, mathematical models, and game theory (Pederson et al. (4) safe but not repairable; and (5) unsafe.
2006). Therefore, once the approach and the geographic scale is The proposed resilience performance levels (RPLs) focus on
selected, the global performance indicator QTOT can be plotted over building performance after the earthquake, recognizing the impor-
the region of interest using a contour plot at a given instant of time tance of the temporal dimension (Recovery time TRE) in the as-
t, so the time-dependent functionality maps can be obtained. When sessment of the RPLs of structures and communities in general.
also the control time T LC is defined, the resilience contour map of In this paper, a two-dimensional (2D) performance domain, con-
the region of interest can also be plotted. The resilience contour sisting of performance levels PLði; jÞ, defined by the combination
maps are obtained by integrating the functionality maps over time of functionality (Index j) and recovery time (Index i) are proposed.
using Eq. (1), therefore, the resilience maps will be time indepen- By accounting for the effect of the temporal dimension, a three-
dent, but they will vary in space from point to point in the selected dimensional (3D) performance matrix (Fig. 5) can be visualized
region. Finally, the community resilience index, Rcom , is given by as a set of predefined joined performance domains (masks) for
the double integral over time and space as follows: different seismic intensity levels, IM, and RPLs. The RPLs can
Fig. 5. 3D resilience performance levels matrix for structures, communities, and systems
be defined using expert opinions and public interviews, which will applications of this approach can be found in the literature: to
allow identification of the acceptable and desired performance lev- hospitals (Cimellaro et al. 2011), natural gas distribution networks
els by citizens for different type of infrastructures, for example. (Cimellaro et al. 2015b), water distribution network (Cimellaro
et al. 2015a); but they are not reported in this paper attributable
to the lack of space.
Restoration Models
After the definition of the performance indices provided in previous
sections, the key elements for the quantification of the resilience Concluding Remarks
index are the restoration phase and the recovery time, but they
are also the most uncertain and difficult quantities to be computed. After the recent disasters worldwide, the general public became
Several recovery models are available in literature to describe the aware that resilience is the solution to face natural and artificial
restoration functions, and they can be grouped (as empirical or ana- threats. This paper presents a holistic framework to evaluate the
lytical) depending on the type of analysis and data (Cimellaro resiliency of a community at various geographical and temporal
et al. 2010a). scales and identifies the gaps in the definitions and quantification
Empirical recovery models are based on test or field data inter- of resilience at the community level. The suggested framework,
pretation and engineering judgment. They can be built using Monte summarized with the acronym PEOPLES, combines different di-
Carlo simulations based on data from past events or maximum like- mensions of resilience together using a layered approach. The main
lihood method. Because the complexity of the problem changes contribution in the field is the development of a community hybrid
case by case, no specific model is presented in this part. model combining network models to simulate the physical infra-
Analytical recovery models are developed using response data structures (e.g., electric power, water, gas) with agent-based models
from numerical simulations (e.g., nonlinear time history analysis, to simulate the sociotechnical networks (e.g., emergency medical
response spectral analysis) of system models. Few examples of ana- technicians, fire brigade, police). Furthermore, special attention is
lytical recovery models (e.g., discrete event simulation models, given to human behavior and emotions, which play a key role dur-
metamodels) for critical facilities like hospitals can be found in ing an emergency; they have been modeled using the extended
Cimellaro et al. (2011, 2014a). version of the BDI framework.
Each dimension of the framework is made of components and sub-
components with their respective performance indicators. These
Step-by-Step Procedure for Resilience Evaluation indicators can be grouped according to their difficulty in evaluating
Finally, to clarify the methodology previously described, a sche- them, their complexity, and spatial and temporal scales. Some of them
matic step-by-step procedure of the framework (shown in Fig. 2) might be valid on a multihazard approach, although others might be
is given below. valid only for certain types of hazard. The PEOPLES framework is the
1. Define the extreme event scenarios [e.g., Probabilistic seismic result of a project developed in 2009 through the national institute of
hazard analysis (PSHA) and ground motion selection]. standards and technology (NIST), and since then, several applications
2. Definition, calibration and validation of the hybrid model of the of the framework have been developed therefore the paper is showing
community. some examples of indicators for this methodology and is making
3. Run the analysis and evaluate the response of the model. reference to existing applications in literature.
4. Evaluate the performance metrics (e.g., losses, restoration time, In the long term, the proposed framework can be used as deci-
performance index, resilience index) for different scenarios and sion support software by decision makers, planners, and engineers
compare with different performance levels. to help implement RBD techniques. The goal is to make individual
5. Recognize remedial mitigation actions (e.g., advanced technol- structures and communities safe and resilient with both advanced
ogies such as base isolation, passive dampers) and/or resilience technologies (e.g., base isolation, passive dampers) and resilience
actions (e.g., resourcefulness, redundancy). actions that allow each system to recover its functionality in a short
The proposed design approach has analogies with the feedback time by selecting the optimal restoration strategy and enhancing
loop taken from control theory, and it can be applied both to com- the community resilience index by comparing it with the resilience
munities and single structures (e.g., hospital, city hall). Several levels targets.
© ASCE
demographics and ecosystem governmental a. Facilities competence a. Financial Services capital
a. Distribution a. Water quality services (1) Residential a. Collective action (1) Asset base of financial a. Child and elderly
and density and quantity a. Executive and (a) Housing units and decision making institutions services
(1) Urban b. Air quality administrative (b) Shelters (1) Conflict resolution (2) Checking account balances b. Commercial
(2) Suburban c. Soil quality (1) Emergency (2) Commercial (2) Self-organization (personal and commercial) centers
(3) Rural d. Biodiversity response (a) Distribution b. Collective efficacy (3) Consumer price index c. Community
(4) Wild land e. Biomass (2) Health and facilities and empowerment (4) Insurance participation
b. Composition (vegetation) hygiene (b) Hotels and c. Quality of life (5) Number and average d. Cultural and
(1) Age f. Other natural b. Judicial accommodations amount of loans heritage services
(2) Gender resources c. Legal and (c) Manufacturing (6) Number of bank and e. Education
(3) Immigrant status security facilities credit union members services
(4) Race and ethnicity (d) Office buildings (7) Number of bank and credit f. Non-profit
c. Socioeconomic status (3) Cultural unions organizations
(1) Educational (a) Entertainment venues (8) Saving account balances g. Place attachment
attainment (b) Museums (personal and commercial)
(2) Income (c) Religious institutions (9) Stock market
(3) Poverty (d) Schools b. Industry—Employment services
(4) Home (e) Sports and (1) Agriculture
ownership recreation venues (2) Construction
(5) Housing b. Lifelines (3) Education and health services
vacancies (1) Communications (4) Finance, insurance, and real state
(6) Occupation (a) Internet
(b) Phones (5) Fortune 1,000
(c) TV (6) Fortune 500
(d) Radio (7) Information, professional
04016063-11
(e) Postal business, others
(2) Health care (8) Manufacturing
(a) Acute care (9) Leisure and hospitality
(b) Long-term acute care (10) Number of corporate
J. Struct. Eng.
Acknowledgments Cimellaro, G. P., and Solari, D. (2014). “Considerations about the optimal
period range to evaluate the weight coefficient of coupled resilience
The research leading to these results has received funding from index.” Eng. Struct., 69, 12–24.
the European Research Council under the Grant Agreement n° Cimellaro, G. P., Solari, D., and Bruneau, M. (2014c). “Physical infrastruc-
ERC_IDEal reSCUE_637842 of the project IDEAL RESCUE— ture Interdependency and regional resilience index after the 2011
Integrated DEsign and control of Sustainable CommUnities during Tohoku earthquake in Japan.” Earthquake Eng. Struct. Dyn., 43(12),
1763–1784.
Emergencies and from the European Community’s Seventh Frame-
Cimellaro, G. P., Tinebra, A., Renschler, C., and Fragiadakis, M. (2015a).
work Programme—Marie Curie International Outgoing Fellow- “Resilience-based design of an urban water distribution system.”
ship (IOF) Actions-FP7/2007-2013 under the Grant Agreement J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X.0001433, C4015014.
n°PIOF-GA-2012-329871 of the project IRUSAT—Improving Cimellaro, G. P., Villa, O., and Bruneau, M. (2015b). “Resilience-based
Resilience of Urban Societies through Advanced Technologies. design of natural gas distribution networks.” J. Infrastruct. Syst,
10.1061/(ASCE)IS.1943-555X.0000204, 05014005.
Comfort, L. K. (1999). Shared risk: Complex systems in seismic response,
References Pergamon, New York.
Downloaded from ascelibrary.org by University Of Florida on 12/04/17. Copyright ASCE. For personal use only; all rights reserved.
tance of place attachment to community participation and planning.” Silva, J. D., and Morera, B. (2014). City resilience framework, Rockefeller
J. Plann. Lit., 20(4), 335–350. Foundation/Arup, New York.
Mayunga, J. S. (2009). “Measuring the measure: A multi-dimensional scale Simoniello, R. A., and Quarantelli, E. L. (1985). “Emergent citizen groups
model to measure community disaster resilience in the U.S. Gulf Coast and emergency management.” Publ. Administration Rev., 45(1), 93–100.
region.” Ph.D. dissertation, Texas A&M Univ., College Station, TX. Simoniello, T., Lanfredi, M., Liberti, M., Coppola, R., and Macchiato, M.
Miles, S., and Chang, S. (2006). “Modeling community recovery from (2008). “Estimation of vegetation cover resilience from satellite time
earthquakes.” Earthquake Spectra, 22(2), 439–458. series.” Hydrol. Earth Syst. Sci., 12(1), 1053–1064.
Mileti, D. (1999). Disasters by design: A reassessment of natural hazards Stallings, R. A., and Quarantelli, E. L. (1985). “Emergent citizen groups
in the United States, Joseph Henry Press, Washington DC. and emergency management.” Public Admin. Rev., 45, 93–100.
Miller-Hooks, E., Zhang, X., and Faturechi, R. (2012). “Measuring and Tamvakis, P., and Xenidis, Y. (2013). “Comparative evaluation of resil-
maximizing resilience of freight transportation networks.” Comput. ience quantification methods for infrastructure systems.” Procedia Soc.
Oper. Res., 39(7), 1633–1643. Behav. Sci., 74, 339–348.
Norris, F. H., Stevens, S. P., Pfefferbaum, B., Wyche, K. F., and Pfeffer- Tierney, K. (1997). “Business impacts of the Northridge earthquake.”
baum, R. L. (2008). “Community resilience as a metaphor, theory, set of J. Contingencies Crisis Manage., 5(2), 87–97.
capacities, and strategy for disaster readiness.” Am. J. Community Tierney, K. (2009). “Disaster response: Research findings and their impli-
Psychol, 41(1–2), 127–150. cations for resilience measures.” CARRI Research Rep. 6, Community
Olofsson, P., Eklundh, L., Lagergren, F., Jonsson, P., and Lindroth, A. & Regional Resilience Institute, Boulder, CO.
(2007). “Estimating net primary production for Scandinavian forests Tierney, K., and Bruneau, M. (2007). “Conceptualizing and measuring
using data from Terra/MODIS.” Space Res., 39(1), 125–130. resilience.” TR News, 250(1), 14–17.
Ortony, A. (2002). “On making believable emotional agents believable.” Tyler, S., and Moench, M. (2012). “A framework for urban climate resil-
Chapter 6, Emotions in humans and artifacts, R. Trappi, P. Petta, ience.” Clim. Dev., 4(4), 311–326.
and S. Payr, eds., MIT Press, Cambridge, MA, 189–211. United Nations. (2013). “Plan of action on disaster risk reduction for
Ouyang, M., and Duenas-Osorio, L. (2011). “An approach to design inter- resilience.” New York.
face topologies across interdependent urban infrastructure systems.” U.S. DHS (U.S. Department of Homeland Security). (2008). “DHS risk
Reliab. Eng. Syst. Saf., 96(11), 1462–1473. lexicon, risk steering committee.” Washington, DC.
Ouyang, M., Dueñas-Osorio, L., and Min, X. (2012). “A three-stage resil- Vugrin, E. D., Warren, D. E., Ehlen, M. A., and Camphouse, R. C. (2010).
ience analysis framework for urban infrastructure systems.” Struct. Saf., A framework for assessing the resilience of infrastructure and economic
36, 23–31. systems, Springer, Berlin.
Pederson, P., Dudenhoeffer, D., Hartley, S., and Permann, M. (2006). Weick, K. E. (1995). Sensemaking in organizations (foundations for
“Critical infrastructure modeling: A survey of U.S. and international organizational science), SAGE, Thousand Oaks, CA.
research.” Idaho National Laboratory, Idaho Falls, ID. Weick, K. E., et al. (2005). “Organizing and the process of sensemaking.”
Pettorelli, N., et al. (2005). “Using the satellite-derived NDVI to assess Organiz. Sci., 16(4), 409–421.
ecological responses to environmental change.” TRENDS Ecol. Evol., Werner, E. E., and Smith, R. S. (1982). Vulnerable but invincible: A study
20(9), 503–510. of resilient children, McGraw-Hill, New York.
Queiroz, C., Garg, S. K., and Tari, Z. (2013). “A probabilistic model Wildavsky, A. B. (1991). Searching for safety, Oxford Transaction
for quantifying the resilience of networked systems.” Ibm J. Res. Dev., Publisher, NB, Canada.
57(53:1–3:9). Woods, D. D. (2006). “Essential characteristics of resilience.” Resilience
Renschler, C., Frazier, A., Arendt, L., Cimellaro, G. P., Reinhorn, A. M., engineering: Concepts and precepts, Ashgate, Aldershot, U.K., 21–23.
and Bruneau, M. (2010). “Framework for defining and measuring resil- Zoumpoulaki, A., Avradinis, N., and Vosinakis, S. (2010). “Multi-agent
ience at the community scale: The PEOPLES resilience framework.” simulation framework for emergency evacuations incorporating
MCEER Technical Rep.–MCEER-10-006, State Univ. of New York, personality and emotions.” Artificial intelligence: Theories, models and
Buffalo, NY, 91. applications, Springer, Berlin, 423–428.