Help Drafting

Civil Engineering and Environmental Systems, 2015 Vol. 32, Nos. 1–2, 5–17, http://dx.doi.org/10.1080/10286608.2015.1025065

Improving resilience through vulnerability assessment and management

Jitendra Agarwal∗

Department of Civil Engineering, University of Bristol, Bristol BS8 1TR, UK

(Received 18 October 2014; accepted 27 February 2015)

The increasing complexity of infrastructure systems and the possibility of severe consequences due to interdependency and uncertain demands have led to an increased emphasis on resilience. Resilience, in simple terms, is the ability of a system to withstand adverse conditions and to recover quickly from these. Its interpretations and linkages to the related concepts of vulnerability and risk are examined. It is argued that vulnerability is an inherent characteristic of any system, hard or soft, and its identification and management is essential for improving the system’s resilience. A systems approach to identify the vulnerable failure scenarios uses the concepts of form, connectivity and hierarchical modelling. Modelling of interactions with social systems and assessing their consequences requires dealing with uncertainty and it remains a challenge.

Keywords: resilience; vulnerability; risk; infrastructure systems; networked systems

1. Introduction

Infrastructure systems such as those for water supply, transport, communication and energy are socio-technical systems of enormous complexity. Traditionally, these have been designed and operated as technical systems to achieve a target level of reliability. However, with increasing social and economic consequences after unpredictable failures, this cannot remain so. For exam- ple, the economic damage from the failure of North American power grid in 2003 was estimated to be $6.4 billion (Anderson and Geckil 2003). Similarly, the collapse of I-35W bridge over the Mississippi River in Minneapolis resulted in estimated economic losses of $71,000–$220,000 per day depending upon adjustment of destinations by the road-users (Xie and Levinson 2011). There is also increasing evidence of a knock-on effect of the failure of one infrastructure com- ponent onto another. For example, the 2007 Gloucester floods in the UK (Cabinet Office 2008) not only disrupted the water supply for nearly 350,000 people but also threatened the opera- tion of a power plant. The increasing complexity of infrastructure systems has contributed to the possibility of unintended behaviour. With climate change high on the agenda in scientific and policy domains, there is increasing realisation that infrastructure systems cannot be designed for all hidden and uncertain threats. Instead there is a move towards increasing the resilience of infrastructure systems and societies so that they are better able to cope with unknown demands, uncertainties of the modelling and the emerging consequences.

*Email: [email protected]

© 2015 Taylor & Francis

6 J. Agarwal

There is a growing body of literature on the resilience of systems in areas such as socio- ecological systems, complexity theory, disaster research and risk management (see e.g. Francis and Bekera 2014; Ouyang 2014). Different approaches for resilience are being put forward. The purpose of this paper is to argue that the identification of vulnerabilities is essential for improving the resilience of infrastructure systems and systems approaches have an important role to play. First, the interpretations and linkages between the related terms, resilience, vulnerability and risk are examined and then a systems approach for vulnerability modelling is introduced and the relevant systems issues discussed.

2. Resilience

The term ‘resilience’ is used in different disciplines. In physics, it is the ability of an object to return to its original shape after being deformed. In medicine, it refers to an individual’s ability to recover from illness or depression. In the context of ecological systems, resilience implies the persistence of systems to external influences and their ability to absorb disturbance and adapt their dynamics (Holling 1973). The latter implies that they may find a new stable state rather than trying to return to their original state. UNISDR (2009) defines resilience as

the ability of a system, community, or society exposed to hazards to resist, absorb, accommodate to and recover from the effects of a hazard in a timely manner, including through the preservation and restoration of its essential basic structures and functions. (24)

This has two important aspects, that is, minimum functionality and time to recover. In the context of infrastructure systems, Bruneau et al. (2003) (and many others) have related

resilience to the ability of a system (i) to reduce the chances of a shock, (ii) to absorb the shock and (iii) to recover quickly after a shock. They proposed robustness, rapidity, redundancy and resourcefulness as the four aspects to measure it. Robustness implies the ability to withstand an adverse event without disproportionate consequences. Rapidity refers to the speed at which the recovery occurs. Redundancy points to the presence of duplicate means. Resourcefulness is used as a measure of the ability of a system to adapt to unexpected events. Redundancy contributes to robustness, and rapidity depends upon the resources available, thus it could be argued that robustness and rapidity of recovery are two key aspects of resilience.

According to the UK government (Cabinet Office 2011), a resilient system or organisation will be able to achieve its core objectives in the face of adversity through a combination of measures. These have been placed into four categories: resistance (direct physical protection), reliability (capability to operate under a range of conditions), redundancy (concerned with design and capacity), and response and recovery (ability to respond and recover). The first three com- ponents, that is, resistance, reliability and redundancy are closely linked and routinely practised in the design of engineering systems; however, these are associated with large uncertainties for human systems.

In both the above approaches, the primary focus is on the restoration of the original state of the system. Ecological resilience is concerned with finding another stable state. Organisations tend to be flexible and easy to move into a different state. Technical infrastructure projects take several years to come to fruition and they are not easy to upgrade or steer into an alternative state. Haimes (2006) related resilience not only to the ability to recover to a desired state and to the time taken, but also to the cost to achieve this. He argues that resilience can be improved through making the system robust and managing the operational factors, for example, prevention, containment, scenario planning, etc.

A report by the Institute of Public Policy Research (IPPR 2009) refers to three levels of resilience: (i) the ability of a system to absorb shocks primarily through built-in redundancy,

Civil Engineering and Environmental Systems 7

F u

n ct

io n

Time

100%

Stress

Absorb

Repair Adapt

Failure

t1 t2

Figure 1. Different approaches to achieve resilience.

(ii) community resilience through preparation and response measures and (iii) anticipating adverse situations and adapting to circumstances (rather than trying to recover to the original state). The report also highlights the risks posed by reduced redundancy, just-in-time culture and interdependent systems. Improved efficiency has magnified vulnerabilities and increased interde- pendence has created the potential for cascades of failures. Addressing these challenges requires holistic thinking. In the context of socio-ecological systems, Adger (2006) notes three aspects of resilience: (i) the magnitude of disturbance before a radical change occurs in system state, (ii) the capacity to self-organise and (iii) the capacity for adaptation to emerging circumstances. These can be extended to socio-technical systems, however, the ability to self-organise or adapt must come from society unless technical systems can be coded with more intelligence.

At a practical level, the ability to absorb a shock or to recover from an adverse state to the orig- inal state or to a new better state is important. However, understanding uncertainty and coping with change is at the heart of resilience thinking (Rees 2010). It requires creating an appropri- ate model of the system and analysing its behaviour for a range of scenarios. Given the diverse nature of uncertainties and possible changes, a flexible approach is needed. For example, pro- viding redundancy or increasing the resistance may not always be feasible and to achieve the desired outcomes the repair and recovery processes must be planned in advance for the poten- tial vulnerable scenarios. The earlier a system can sense the undesirable events approaching, the sooner it can begin to respond or to adapt. Figure 1 illustrates a conceptual model of methods to achieve resilience. Most systems have some absorptive capacity to bear additional stresses. Adaptation often tends to be a longer term process but failure can be avoided through timely repair and recovery.

3. Vulnerability

Resilience requires robustness. Robustness is the ability ‘to take a knock’ without disproportion- ate loss of functionality. In practice, it means that the system will be able to cope with small variations in demand or minor damage. Advanced optimisation techniques often lead to systems that are good for the demands considered during the design, but may result in failures should these demands change. Reliability theory gives the probability of the successful operation of a system under defined conditions over a stated period of time. However, system outcomes for vari- ations beyond design basis remain uncertain. Further, modelling issues, nonlinearities and limits to our knowledge make it difficult to certify a system to be robust. One way to gain insights into robustness is to examine how a system is vulnerable.

8 J. Agarwal

Vulnerability is commonly referred to in the resilience literature but its usage has tended to be quite different. Vulnerability captures the idea of susceptibility to damage. Such susceptibility derives from a characteristic form of the system within a given context. For example, an earth- quake in California may result in very different consequences from an earthquake of the same strength in Iran, because of the differences in the technical design and construction of the infras- tructure. Agarwal, Blockley, and Woodman (2001) defined vulnerability as the susceptibility of a system to disproportionate consequences in relation to damage or perturbation. Haimes (2006) related vulnerability to the inherent states of a system that can be exploited to adversely affect the system. The behaviour of a system is characterised by its state variables. The state variables are dynamic and so are the vulnerabilities. Infrastructure systems are arranged in a hierarchy and exhibit vulnerabilities at different levels.

Social scientists tend to view vulnerability in terms of the socio-economic factors that deter- mine people’s ability to cope with stress or change. Cutter, Boruff, and Shirley (2003) used a hazards-of-the-places model to identify vulnerability indicators in the US counties. These include personal wealth, proportion of children and elderly, density of built environment, single sector economic dependence, housing stock, race and ethnicity, occupation, infrastructure dependence, etc. Different factors were found to contribute differently to the vulnerability of a place. Morrone et al. (2011) assessed societal vulnerability based on access to different types of capital: economic – indicating poverty levels, human – corresponding to education, social – relating to support networks, and collective assets – such as essential services.

Miller and Nigg (1993) distinguished between ‘event vulnerability’ and ‘consequence vul- nerability’. McEntire, Crocker, and Peters (2010) related the former to the ‘proneness issues’ and the latter to the ‘capabilities’. The former refers to susceptibility of a system to a par- ticular event or action and the latter where consequences tend to be higher due to inadequate strength. Higher consequences derive from inherent weaknesses in the form of a system or due to severe action. Where this is due to inadequacies in form, this can be referred to as ‘inter- nal vulnerability’ and that due to severe action (or lack of capacity) can be called ‘external vulnerability’. In social systems, it is difficult to make a distinction between proneness and capa- bility and this remains a challenge. Indeed, some authors (e.g. Alwang, Siegal, and Jorgensen 2001) argue that proneness and capability are interwoven and both contribute to the vulner- ability of the system. The JCSS framework on risk (JCSS 2008) refers to susceptibility of a component to an action as being vulnerable, and the ensuing disproportionate consequences due to component failure as a lack of robustness. For example, a steel structure in a harsh marine environment is vulnerable to corrosion (i.e. external vulnerability) and if a small reduction in cross-sectional area due to corrosion causes the structure to fail, then it is not robust (i.e. internal vulnerability).

Thus, a broad distinction exists between those who see vulnerability as the potential damage to a system and those who see vulnerability as a state that exists within a system before it encounters a hazardous event. The latter view is that vulnerability exists independently of external hazards. Social vulnerabilities of this type might include poverty, inequality and housing quality. Material defects or a lack of corrosion resistance in physical systems also fall in the latter category. It could be argued that this type of vulnerability is a particular form of hazard – a hazard that is internal to the system.

The need to reduce vulnerability has been emphasised in disaster research (e.g. Blakie et al. 2003) and it has influenced the current research on climate change. The Intergovernmental Panel on Climate Change (IPCC 2007) regards vulnerability as a function of ‘climate varia- tion to which a system is exposed, its sensitivity and its adaptive capacity’. Interestingly, this encompasses most of the elements of resilience defined by UNISDR (2009). Adger (2006) also considered vulnerability as having three components that include exposure, sensitivity to pertur- bations and the capacity to adapt. Exposure here essentially refers to the stresses or perturbations.

Civil Engineering and Environmental Systems 9

It is the sensitivity, that is, the degree to which a system is affected by perturbations, that relates to vulnerability. Adaptive capacity, that is, the ability of a system to evolve to accom- modate perturbations, is taken to contribute to resilience rather than vulnerability. Turner et al. (2003) considered vulnerability in terms of exposure, sensitivity and resilience. They analysed these at a particular spatial scale and sought to establish links to other scales. Berkes (2007) argued that a resilience approach helps to understand uncertainty and to reduce vulnerability. A multi-disciplinary analysis of infrastructure systems requires a shared understanding of different points-of-views. These differences seem to arise from when the impact is measured. All systems are dynamic and it is important to consider the time, for example, time t1 or t2 in Figure 1, at which performance is measured. Vulnerability assessment using consequences at time t2 would include resilience.

Whatever domain one considers, quantification of vulnerability is not easy. A probabilistic measure of vulnerability, defined as the ratio of the failure probability of the damaged system to the failure probability of the undamaged system (Lind 1995), can be applied to almost any sys- tem. Luers et al. (2003) argued for assessing the vulnerability of important state variables using generic metrics such as sensitivity to stress, state relative to defined threshold and exposure to stress. Alwang, Siegal, and Jorgensen (2001) decomposed vulnerability into three compo- nents: the risk, the risk responses (i.e. managing risk), and the outcome in terms of welfare loss. They note that ‘the tautological nature of these definitions – risk determines vulnera- bility, but vulnerability also determines risk – invites confusion’. This seems to derive from hazards being viewed as risks and as discussed earlier, hazards could be external or within the system.

In engineering, as compared to human ecology research, there is not enough emphasis on political and structural causes of vulnerability. Disasters resulting from Hurricane Katrina and the Fukushima tsunami have exposed such vulnerabilities of modern infrastructure systems. However, the consequences of such events in a given context can be reduced through planning using vulnerable failure scenarios. There is a need to identify such scenarios addressing physical, environmental, social and economic dimensions.

McEntire, Crocker, and Peters (2010) discussed four schools of thought for vulnerability reduction – physical, engineering, structural and organisational. The focus of the physical school is on reducing the exposure to hazards, whereas the engineering school aims to increase the resistance through better design and construction practices. The structural school stresses the importance of socio-economic factors and demographic characteristics. The organisational school deals with the effectiveness of response and recovery including the ability to adapt. Both engineering and structural schools refer to inherent characteristics of their respective systems (i.e. engineering and social) which can increase or decrease vulnerability. The other two schools mainly contribute to risk reduction rather than vulnerability reduction. However, certain sys- tem characteristics that govern vulnerability can contribute to increased exposure to hazards and hence increased risk. For example, low socio-economic conditions may lead to settlements in hazardous zones.

Two essential lines of defence against adverse demands are robust internal form and robust management. A system that is vulnerable in any way cannot be robust. An inappropriate form if damaged may lead to consequences that are disproportionate to the initial damage. It gains importance for low chance–high consequence events. Both vulnerability and resilience are not concerned with the likelihood of external actions. Vulnerability is an inherent property of the system and resilience is a performance attribute. Elms (1999) noted that many engineers tend to focus on capacity (strength) rather than vulnerability. Capacity relates to survival and vulnera- bility to failure. The two views can lead to different results, but the ones based on failure could be more useful because an identification of weaknesses can provide protection against several hazards, thus beneficial for improving resilience also.

10 J. Agarwal

4. Risk

Risk and resilience share certain characteristics and it is useful to examine how risk management can be used to enhance resilience. In engineering, risk is understood as the combination of the chance of an event occurring and the consequences of an event in a context. In disaster reduction literature, it is taken as the combination of hazard, exposure and vulnerability. The combination of exposure and vulnerability determines the consequences, and hazard defined in probabilis- tic terms gives the chance. So the two definitions are not very different, however each has its usefulness in connecting risk to resilience.

Risk refers to a potentially dangerous situation that might or might not exist in the future with consequences we want to avoid. Depending upon the context, consequences may be measured in terms of fatalities, injuries, cost of repairs, losses due to unavailability of infrastructure and so on. Risk changes with time and is affected by what we do. It is recognised that risk cannot be eliminated altogether, but measures must be taken to reduce it. These could include reducing exposure to hazards or hardening the system to reduce the consequences. The measures taken to cope with the consequences or recover from these are generally not part of risk management but they are needed to improve resilience. In the Eliminate, Reduce, Inform, Control (ERIC) approach to risk management, eliminating or reducing risks, while helpful, does not always imply that the system has become less vulnerable. This is because risk can be reduced by limiting the exposure also. The Inform and Control aspects of the ERIC approach are useful for resilience. By knowing what the remaining risks are, contingency plans can be prepared and where possible measures can be put in place to control the consequences. These plans or measures need not be limited to credible risks but can also be developed for the inherent vulnerabilities in the system.

Sarewitz, Pielke, and Keykhah (2003) noted that public policies to mitigate the impact of extreme events depend upon whether the focus is on reducing risk or on reducing vulnerability. Reducing vulnerability does not require data about extreme events but risk reduction does. Here vulnerability relates to inherent characteristics of a system that are independent of any particular hazard or event, and the risk of an event is distinguished from the risk of an outcome. The risk of an outcome is based on the risk of an event and the vulnerability of the system. Sarewitz, Pielke, and Keykhah (2003) also refer to the tension between individual action and collective conse- quences. Such tensions could be identified through systems thinking. For example, the growth of megacities due to increased opportunities for economic gains has increased the vulnerability of a city due to a wide variety of hazards. Extreme events are created by context that is, not simply by a set of characteristics inherent in physical phenomenon but by interactions with other systems. For example, the eruption of the Eyjafjallajökull volcano in 2010 became an extreme event due to atmospheric currents (environmental system), heavy use of the air space (social and economic system) and the design of aircraft engines (technical system). In this case, road and rail transport contributed to resilience to some degree, but the robust design of aircraft engines would have reduced the risk of failure and improved resilience.

5. Preparedness and adaptability

All systems are dynamic and they change in time and space, albeit at different scales. Expect- ing the unexpected requires preparedness in advance. This not only includes specific measures but elicitation and integration of local knowledge because global behaviour emerges from local interactions. Such knowledge may be in form of simple rules or may have been gained from past events. Models of many nonlinear engineering systems are known to result in multiple states and the system could reside in one of these states or jump from one to the other depending upon external input. Such behaviour, typical of low degree of freedom systems, is characterised by the

Civil Engineering and Environmental Systems 11

sensitivity of the system to parameters and initial conditions. In the design of such systems, the aim usually is to keep them away from sensitive regions to reduce the uncertainty of behaviour rather than let them move to an alternative state.

For infrastructure systems, it is the adaptability of people who use or interact with the infras- tructure that can contribute to resilience. For example, having different modes of transport is only good if people can adapt quickly in case of the failure of any one mode. Adaptability of ecologi- cal systems contributes significantly to their resilience. Such behaviour results from a collective response that emerges after a shock. However, this is not the case for technological systems. Often man-made systems are tailored to their environment and they require human interven- tion to change their behaviour unless pre-programmed. The latter requires scenario planning and artificial intelligence to provide some degree of adaptability.

6. Infrastructure criticality

Many physical infrastructure systems are spatially distributed. They are continuously evolving while interacting with other systems. These infrastructure systems are expected to have a mini- mum level of resilience. But how does one arrive at these levels? Should these be based on risk calculations or the continuity of services or something else? The UK government has divided national infrastructure into nine sectors: communications, emergency services, energy, finance, food, government, health, transport and water (Cabinet Office 2010). Within each, critical ele- ments are identified according to the impact of their loss. This has three dimensions: (a) the impact on delivery of the nation’s essential services, (b) the economic impact and (c) the impact on life. Severe impact in any dimension could lead to criticality. The severity of impact is dis- tinguished using three factors: (i) the degree of disruption of service, (ii) the extent of disruption in terms of population or geographical region and (iii) the duration of disruption. Approximate ranges are given for the geographical regions and population affected (Table 1) but there are hardly any indicators of the degree of disruption or the duration of disruption, both of which are essential to the recovery plans. Also, there is a lack of clarity on the indicators of life quality

Table 1. Description of criticality scales (adapted from Cabinet Office 2010).

Description

Criticality scale

Degree of disruption

Extent of disruption – geographical

Extent of disruption – population

Impact on the delivery of essential services

Infrastructure importance

CAT 5 Across a number of sectors

Catastrophic impact, national long-term effects

Of unique national importance

CAT 4 Impact on essential services across the nation

Millions of citizens affected

Severe Highest impor- tance to the sector

CAT 3 Affects a large geographic region

Many hundreds of thousands of people affected

Substantial importance to the sector

CAT 2 Affects whole counties

Tens of thousands of people affected

Significant impact

CAT 1 Mostly localised Thousands of people affected

Moderate

CAT 0 Minor (on national scale)

12 J. Agarwal

F u

n ct

io n

Time

100%

Shock

Too much loss (R > R*) & slow recovery (t > t*)

R*

t*

R*,t* acceptable standardR

t

Figure 2. A resilience assessment framework (adapted from Chang and Shinozuka 2004).

and economic impact. In structural safety, a life quality index based on life expectancy and gross domestic product per person has been proposed (e.g. Pandey and Nathwani 2004).

Chang and Shinozuka (2004) proposed a resilience assessment framework based on a com- parison of loss of system performance to predefined standards as shown in Figure 2. Here, R* corresponds to robustness standard (but does not measure robustness in itself) such as less than 5% of the population loses water service and t* corresponds to rapidity standard such as 99% of population has service restored within one week.

Whatever the approach, the minimum levels are largely governed by the societal expectations and the resources available. It is important that consistent standards are set across all sectors and any deviations are communicated across the sectors because of the interconnected nature of infrastructure systems. This leads to the important issue of models of infrastructure systems, discussed in the next section, so that their performance could be simulated for likely future events and compared against the minimum agreed standards. This requires all stakeholders to come together for the benefit of the society and systems professionals have an important role to play.

7. A systems model for vulnerability and resilience

Infrastructure systems are complex and they consist of not just the physical elements such as roads and bridges but also include the organisations that run these, the public who use these and the environment they are part of. They can be considered as a set of interacting objects that are arranged together in an appropriate form to fulfil a purpose. They can be modelled as holons that is, they are both wholes and a part of the whole. The nature of objects may differ from one system to the other. Figure 3 presents a model of the system at different levels of definitions and an example at a detailed level is shown in Figure 4. Any weakness in one part of the system can propagate to other parts of the system or different processes may join together to provide the necessary resistance to stop that happening.

In literature, infrastructure networks such as water supply and electricity grids have been mainly analysed from two points of view: topological analysis and response analysis (also referred to as flow models). Topological analysis focuses on connectedness and accessibility. Properties of links are sometimes taken into account by assigning weights to the links or the nodes but the analysis is independent of the demands. Response analysis requires a model of demands that changes with time.

Civil Engineering and Environmental Systems 13

Figure 3. Modelling infrastructure systems at different levels of definition.

Holyhead

Wrexham

Bangor

Aberystwyth

Fishguard

Cardiff Swansea

Milford Haven Newport

Pembroke

Bristol

Liverpool

Highway Rail Airport Ferry City/Town

Figure 4. A simplified model of the transport network of Wales (adapted from http://www.traffic-wales.com/media/ 33513/welsh-government-strategic-road-network.pdf and http://www.nationalrail.co.uk/static/documents/content/Net work_Rail_national_map(1).pdf).

Scenario planning for improving resilience requires that vulnerable failure scenarios be iden- tified. As discussed before, vulnerability is susceptibility to some kind of damage due to certain characteristics of the form. The system model (Figure 3) is used to systematically look for such vulnerabilities. First, technical infrastructure systems such as water supply or road networks are

14 J. Agarwal

Vulnerability of scenarios

E vi

d e

n ce

f o

r ra

p id

ity o

f re

co ve

ry

Very low

Low

Medium

High

Resilience

Figure 5. Mapping of scenarios for relative resilience.

modelled independently and analysed for damage using the vulnerability theory reported else- where (Agarwal, Blockley, and Woodman 2003 for structures; Pinto et al. 2010 for water supply). The key concepts are that of modelling a system as a graph with nodes and links (Figure 4). The links are the channels of communication between the nodes and associated with each link is a parameter called well formedness that describes the quality of the form of the link. This param- eter is used to form clusters (or communities) of neighbouring nodes and links so that they are able to provide the best functionality to the system. These clusters are grown by including neigh- bouring nodes and links (or clusters) using a set of criteria until there is one single cluster, the whole system. This leads to a hierarchical representation of the system that is systematically searched by introducing damage to identify vulnerable failure scenarios. Each failure scenario has a vulnerability index based on a measure of damage demand and a measure of consequences to the form of the system. These can be used to improve the form of the system or to prepare recovery plans.

Next the impact of failure scenarios of one infrastructure system on the interacting systems (such as the other utility networks, social/community networks, economy and environment) is considered (Agarwal, Liu, and Galvan 2014). The interdependency between them can lead to cascades of failure. For example, the road and rail networks (Figure 4) have their own vulnera- bilities but are impacted by each other. These vulnerabilities also impact the towns they serve. While impact on physical infrastructure can be quantified it is not so easy for social or environ- mental systems. Causal loops (Milke 2013) could help identify positive and negative interactions. The use of various schemes that recognise incompleteness and enable evidence to be combined, for example, ‘Italian flags’ (Blockley and Godfrey 2000) can be useful.

An assessment of the rapidity of recovery after damage to a physical or social system also poses challenges. However, evidence can be gathered from an analysis of the past failures and an examination of the current system processes to make recovery predictions in the future. A map of the vulnerability of identified failure scenarios and the corresponding predictions of recovery (Figure 5) is useful to compare the likely resilience of infrastructure systems. However, such an assessment of resilience must be treated with caution because of the large uncertainties and the nonlinear dynamics of the systems.

8. Discussion and conclusion

Systems philosophy requires that social and environmental impact of potential failure scenarios of technical systems be examined so that appropriate measures can be taken to increase the

Civil Engineering and Environmental Systems 15

resilience of the systems. Infrastructure systems with social dimensions are complex systems where component interactions are difficult to define or poorly understood and behaviour models are not fully known. In such a setting, identifying their vulnerabilities and the ability to recover has to be at the core of resilience building.

There are lessons to be learnt from the complex dynamics of ecological systems where resilience may derive from structural changes, even though these occur at a much longer time scale as compared to those for infrastructure systems with a social dimension. Hence resilience- building measures for the latter are aimed at maintaining the function and the structure, at least in the short term. Identifying the structural weaknesses, such as through an analysis of form, and taking remedial actions can provide a powerful method to increase the resilience.

There exists a tension between the improvement in the level of resilience and competing demands on resources. If the potential consequences are serious enough not to be ignored, action has to be taken irrespective of resource constraints. Hence, it is necessary to define the accept- able levels of vulnerability and recovery for all interacting systems. These would invariably vary depending upon the economic state of the society and its value system but should be at least aimed at avoiding cascading effects and must be agreed through a dialogue between all the stakeholders. Development of robust models for spatial and temporal analysis of infrastructure systems that would allow a range of social behaviours and recovery options to be explored is very much needed.

Systems thinking for resilience also requires considering all kinds of threats whether techno- logical, climate, terrorism, cyber or pandemics. Information technology (IT) is now embedded in all systems. It is difficult to imagine the consequences of sudden loss of IT infrastructure. Even localised malfunctions are known to cause significant disruptions. Similarly, spread of disease has become a much greater threat in recent years.

All catastrophic modelling tools determine losses in monetary terms. The impact of the dis- ruptions to the supply of goods or services is often also measured in economic terms. However, the social costs (e.g. the lives lost, psychological effects, lost livelihoods, disintegrated com- munities, etc.) cannot be measured – not just because these are difficult to estimate but due to ethical issues also. Infrastructure owners assess damage to their infrastructure and restore supply as quickly as possible. But who should assess the damage done to the communities – infras- tructure owners, local government or communities themselves? Time and economic constraints make it difficult to collect all the information needed for assessing the social consequences of an infrastructure failure. If this remains so, measures to improve resilience may not be fully effective.

Risk calculations are based on chance of failure and do not deal satisfactorily with human and organisational issues. Moreover, low probability – high consequence events may be missed out. Risk is often defined as a triplet: what can go wrong, how likely it is and what are the consequences. Similarly, resilience can be defined as a triplet: what can go wrong, what are the consequences and how quickly the system can recover. So risk management contributes to resilience building but more needs to be done.

Two phases can be identified for resilience building. In the first phase, before the event, vulner- abilities are identified considering all interactions and managed through redundancy, diversity, protection, and monitoring. In the second phase, after the event, resourcefulness, including the preparatory measures, social and economic capital, determines the rate of recovery through repair and/or gradual adaptation.

Disclosure statement

No potential conflict of interest was reported by the author.

16 J. Agarwal

References

Adger, W. N. 2006. “Vulnerability.” Global Environmental Change 16: 268–281. doi:10.1016/j.gloenvcha.2006.02.006. Agarwal, J., D. I. Blockley, and N. J. Woodman. 2001. “Vulnerability of Systems.” Civil Engineering and Environmental

Systems 18: 141–165. doi:10.1080/02630250108970297. Agarwal, J., D. I. Blockley, and N. J. Woodman. 2003. “Vulnerability of Structural Systems.” Structural Safety 25:

263–286. doi:10.1016/S0167-4730(02)00068-1. Agarwal, J., M. Liu, and G. Galvan. 2014. “Vulnerability and Resilience of Networked Infrastructure Systems.” In Vul-

nerability, Uncertainty and Risk: Quantification, Management and Mitigation, edited by M. Beer, S. K. Au, and J. W. Hall, 2811–2820. Reston, VA: ASCE. doi:10.1061/9780784413609.283.

Alwang, J., P. B. Siegal, and S. L. Jorgensen. 2001. “Vulnerability: A View from Different Disciplines.” SP discussion paper 0115. Washington: The World Bank. http://www-wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/ 2002/01/17/000094946_01120804004787/Rendered/PDF/multi0page.pdf.

Anderson, P. L., and I. K. Geckil. 2003. Northeast Blackout Likely to Reduce US Earnings by $6.4 Billion. Lansing, MI: Anderson Economic Group. http://www.andersoneconomicgroup.com/Portals/0/upload/Doc544.pdf.

Berkes, F. 2007. “Understanding Uncertainty and Reducing Vulnerability: Lessons from Resilience Thinking.” Natural Hazards 41: 283–295. doi:10.1007/s11069-006-9036-7.

Blakie, P., T. Cannon, I. Davis, and B. Wisner. 2003. At Risk: Natural Hazards, People’s Vulnerability and Disasters. London: Routledge.

Blockley, D. I., and P. S. Godfrey. 2000. Doing it Differently. London: Thomas Telford. Bruneau, M., S. E. Chang, R. T. Eguchi, G. C. Lee, T. D. O’Rourke, A. M. Reinhorn, M. Shinozuka, K. Tierney, W. A.

Wallace, and D. von Winterfeldt. 2003. “A Framework to Quantitatively Assess and Enhance the Seismic Resilience of Communities.” Earthquake Spectra 19 (4): 733–752. doi:10.1193/1.1623497.

Cabinet Office. 2008. The Pitt Review: Lessons Learned From the 2007 Floods. London: Crown copyright. http://webar chive.nationalarchives.gov.uk/20100807034701/http://archive.cabinetoffice.gov.uk/pittreview/thepittreview/ final_report.html.

Cabinet Office. 2010. Strategic Framework and Policy Statement on Improving the Resilience of Critical Infrastructure to Disruption from Natural Hazards. London: Crown copyright. https://www.gov.uk/government/uploads/system/ uploads/attachment_data/file/62504/strategic-framework.pdf.

Cabinet Office. 2011. Keeping the Country Running: Natural Hazards and Infrastructure. London: Crown copyright. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/61342/natural-hazards-infrastructure. pdf.

Chang, S. E., and M. Shinozuka. 2004. “Measuring Improvements in Disaster Resilience of Communities.” Earthquake Spectra 20: 739–755. doi:10.1193/1.1775796.

Cutter, S. L., B. J. Boruff, and W. L. Shirley. 2003. “Social Vulnerability to Environmental Hazards.” Social Science Quarterly 84: 242–261. doi:10.1111/1540-6237.8402002.

Elms, D. G. 1999. “Achieving Structural Safety - Theoretical Considerations.” Structural Safety 21: 311–333. doi:10.1016 /S0167-4730(99)00027-2.

Francis, R., and B. Bekera. 2014. “A Metric and Frameworks for Resilience Analysis of Engineered and Infrastructure Systems.” Reliability Engineering and System Safety 121: 90–103. doi:10.1016/j.ress.2013.07.004.

Haimes, Y. Y. 2006. “On the Definition of Vulnerabilities in Measuring Risk to Infrastructures.” Risk Analysis 26: 293– 296. doi:10.1111/j.1539-6924.2006.00755.x.

Holling, C. S. 1973. “Resilience and Stability of Ecological Systems.” Annual Review of Ecology and Systematics 4: 1–23. http://www.jstor.org/stable/2096802.

IPCC. 2007. IPCC Fourth Assessment Report: Climate Change 2007. Geneva: Intergovernmental Panel on Climate Change. http://www.ipcc.ch/publications_and_data/ar4/syr/en/contents.html.

IPPR. 2009. Shared Responsibilities: A National Security Strategy for the UK. London: Institute for Public Policy Research. http://www.ippr.org/publications/shared-responsibilitiesa-national-security-strategy-for-the-uk.

JCSS. 2008. Risk Assessment in Engineering – Principles, System Representation and Risk Criteria. Lyngby: Joint Committee on Structural Safety. http://www.jcss.byg.dtu.dk/Publications/Risk_Assessment_in_Engineering.aspx.

Lind, N. C. 1995. “A Measure of Vulnerability and Damage Tolerance.” Reliability Engineering and System Safety 48: 1–6. doi:10.1016/0951-8320(95)00007-O.

Luers, A. L., D. B. Lobell, L. S. Sklar, C. L. Addams, and P. A. Matson. 2003. “A Method for Quantifying Vulnerability, Applied to the Agricultural System of the Yaqui Valley, Mexico.” Global Environmental Change 13: 255–267. doi:10.1016/S0959-3780(03)00054-2.

McEntire, D., C. G. Crocker, and E. Peters. 2010. “Addressing Vulnerability through an Integrated Approach.” Interna- tional Journal of Disaster Resilience in the Built Environment 1 (1): 50–64. doi:10.1108/17595901011026472.

Milke, M. W. 2013. “Casual Loops for Analysis of the Social Dimension to Complex Systems.” Civil Engineering and Environmental Systems 30: 243–248. doi:10.1080/10286608.2013.853751.

Miller, K., and J. M. Nigg. 1993. “Event and Consequent Vulnerability: Effects on Disaster Recovery Process.” Pro- ceedings of Eastern Sociological Society, Boston. http://udspace.udel.edu/bitstream/handle/19716/350/PP217.pdf? sequence = 3.

Morrone, A., K. Scrivens, C. Smith, and C. Balestra. 2011. “Measuring Vulnerability and Resilience in OECD Countries.” Paper prepared for the IARIW-OECD conference on economic insecurity, Paris, November 22–23. http://www.iariw.org/papers/2011/morronepaper.pdf.

Civil Engineering and Environmental Systems 17

Ouyang, M. 2014. “Review on Modelling and Simulation of Interdependent Critical Infrastructure Systems.” Reliability Engineering and Systems Safety 121: 43–60. doi:10.1016/j.ress.2013.06.040.

Pandey, M. D., and J. S. Nathwani. 2004. “Life Quality Index for the Estimation of Willingness-to-pay for Safety.” Structural Safety 26: 181–199. doi:10.1016/j.strusafe.2003.05.001.

Pinto, J., H. Varum, I. Bentes, and J. Agarwal. 2010. “A Theory of Vulnerability of Water Pipe Network.” Water Resource Management 24: 4237–4254. doi:10.1007/s11269-010-9655-3.

Rees, W. E. 2010. “Thinking Resilience.” Post Carbon Institute. http://www.postcarbon.org/publications/thinking- resilience/.

Sarewitz, D., R. Pielke, and M. Keykhah. 2003. “Vulnerability and Risk: Some Thoughts from a Political and Policy Perspective.” Risk Analysis 23: 805–810. doi:10.1111/1539-6924.00357.

Turner, B. L., R. E. Kasperson, P. A. Matson, J. J. McCarthy, R. W. Corell, L. Christensen, N. Eckley, et al. 2003. “A Framework for Vulnerability Analysis in Sustainability Science.” Proceedings of National Academy of Sciences US 100: 8074–8079. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC166184/pdf/1008074.pdf.

UNISDR. 2009. Terminology on Disaster Risk Reduction. Geneva: UNISDR. http://www.unisdr.org/files/7817_UNISDR TerminologyEnglish.pdf.

Xie, F., and D. Levinson. 2011. “Evaluating the Effects of the I-35W Bridge Collapse on Road-Users in the Twin Cities Metropolitan Region.” Transportation Planning and Technology 34 (7): 691–703. doi:10.1080/03081060. 2011.602850.

Copyright of Civil Engineering & Environmental Systems is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.