Resilience assessment framework for critical infrastructure in a multi-hazard environment: Case study on transport assets
Graphical abstract
Introduction
The exposure of critical infrastructure to natural hazards such as floods, earthquakes, tsunami, landslides, hurricanes, wildfires or extreme temperatures was proven to have severe consequences on world economies and societies (Pescaroli and Alexander, 2016). For example, the heavy 2007 rainfall in the UK affected the road network, with the cost estimated at £60m, while during the 2009 floods in Cumbria, UK, at least 20 bridges had collapsed or damaged, causing one fatality, £34m of restoration costs and great societal impact (Cumbria County Council, 2010). Among the critical threats to infrastructure around the world, scour is recognised as the most common cause of bridge failure (Kirby et al., 2015). The direct and indirect economic losses due to landslides affecting road networks are of similar magnitude (Winter et al., 2016). The effects of natural hazards may be exacerbated due to climate change that causes more frequent and intense extreme weather and climatic events (Stern et al., 2013; Draper et al., 2015; Pant et al., 2018; Sarkodie and Strezov, 2019). Furthermore, infrastructure assets are exposed to multiple hazards and/or cascading effects, such as flood series over time, flood-earthquake, earthquake-induced tsunami, landslides and liquefaction, rainfall-induced landslides or earthquake-aftershock events (Akiyama et al., 2019). A well-known example of the importance of multiple hazard effects is the 2011 Tohoku, Japan earthquake and resulting tsunami. During this extreme event, the country rail and highway networks were both strongly affected, and in total 23 stations were washed away, tracks and bridge piers were either eroded or buried, passenger and freight trains were derailed (Krausmann and Cruz, 2013). During the destructive hurricanes Katrina in 2005 and Sandy in 2012 in the US, several structures were damaged due to combined wave forces and debris impact (Padgett et al., 2008). Rainfall-induced landslides are one of the most critical geohazards in the world (Zhang et al., 2019) and earthquake-induced landslides are equally detrimental. The 2008 Wenchuan earthquake in China triggered more than 15,000 landslides, caused more than 20,000 deaths and the cut-off of many towns, due to the extensive damage to highways (Tang et al., 2011). More recently, a bridge had collapsed due to flood in Italy, an area with high seismicity (Scozzese et al., 2019).
Infrastructure owners and operators are increasingly faced with the challenge of delivering resilient infrastructure and mitigating the effects of multiple hazards and climate change effects. In particular, resilience describes the emergent property or attributes that infrastructure has, which allows them to withstand, respond and/or adapt to a vast range of disruptive events, by maintaining and/or enhancing their functionality (Woods, 2015). The term is used widely over many different fields of research, but quantitative metrics of the resilience of socio-technical systems are not well established and standards and processes are still emerging (Lloyd's Register Foundation, 2015). The concept of resilient cities and infrastructure for disaster management has nowadays received more attention, and the existing approaches are mainly based on qualitative methods and index systems (Rockefeller Foundation, 2014; Rus et al., 2018). Moreover, the risk approaches for multi-hazard assessment and management of ecosystems (Furlan et al., 2018; Robinne et al., 2018), communities (Moghadas et al., 2019; Sajjad and Chan, 2019) and critical infrastructure (Giannopoulos et al., 2012; Komendantova et al., 2014; Theocharidou and Giannopoulos, 2015; Chen et al., 2019) are generally qualitative, or quantitative (e.g. Decò and Frangopol, 2011). Life-cycle design and assessment methodologies of infrastructure under multiple hazards are discussed by Yang and Frangopol (2018) and Akiyama et al. (2019). Also, the hazard interactions and cascading effects can be classified differently, while modelling of multiple hazards has not been established or agreed internationally yet (Gill and Malamud, 2014; Zaghi et al., 2016; Liu et al., 2016; Bruneau et al., 2017).
Resilience-based assessment and management are recent philosophies that are gradually being adopted in practical applications of critical infrastructure and are expected to be incorporated in the next generation of provisions and guidelines, e.g. see REDi system by Almufti and Willford (2013). However, the shift to resilience-based management should include specific methods to define and measure resilience, new modelling and simulation techniques for highly complex and interacting systems, development of resilience engineering and approaches for communication with stakeholders (Linkov et al., 2014). In this context, different frameworks and assessment tools have been proposed in the literature to assess resilience under individual or multiple hazards, at (a) asset level, (b) infrastructure network level, and (c) community or national scale (Table 1). The resilience metrics and criteria are commonly dealing with descriptive and qualitative analysis. Recently, Kong and Simonovic (2019) assessed multiple hazard spatiotemporal resilience of interdependent infrastructure systems using network theory and statistical analysis. Quantitative resilience metrics usually measure the quality or performance of the asset or system before and after the event, and the recovery rate (Hosseini et al., 2016). Resilience measures can be either static or time-dependent, and in some cases, stochastic approaches are enabled to account for the aleatoric and epistemic uncertainties (Frangopol and Bocchini, 2011; Ouyang et al., 2012; Decò et al., 2013). The majority of the abovementioned frameworks generally encompass the principles of resilience or the 4R's, as per Bruneau et al. (2003): 1) Robustness, describing the inherent strength or resistance of a system to withstand external demands, e.g. hazard actions, without degradation or loss of functionality; 2) Redundancy (Zhu and Frangopol, 2012), reflecting the system properties that allow for alternate options, choices and substitutions under stresses; 3) Resourcefulness, expressing the capacity to mobilise needed resources and services under emergency conditions, and 4) Rapidity, defining the speed at which disruption can be overcome.
The robustness to hazard actions is usually quantified through fragility functions, which give the probability of the asset exceeding defined limit states, e.g. serviceability and ultimate, for a given hazard intensity, e.g. peak ground acceleration for earthquakes, water discharge or scour depth for floods or ground displacement for liquefaction and landslides. Fragility functions can be derived from empirical, analytical, expert elicitation and hybrid approaches (Argyroudis et al., 2019; Silva et al., 2019). An overview of the available fragility functions for critical infrastructure subjected to earthquakes is given by Pitilakis et al. (2014), while HAZUS-MH (2011) methodology provides fragility functions and loss models for buildings and infrastructure in the US, exposed to earthquakes, tsunamis, hurricanes and floods. Bridges are key assets of the transport infrastructure, and the available fragility models for earthquakes and other hazards are discussed by Tsionis and Fardis (2014), Billah and Alam (2015), Gidaris et al. (2017) and Stefanidou and Kappos (2019), while fragility functions for other transport assets are summarised by Argyroudis and Kaynia (2014) and Argyroudis et al. (2019). The fragility of other assets exposed to hazards other than earthquakes are limited and sparse, including for example electric power transmission lines and towers exposed to wind (Panteli et al., 2017), industrial plants and tanks subjected to tsunami (Mebarki et al., 2016) or critical infrastructure under volcanic hazards (Wilson et al., 2017). Few fragility models for multiple hazards are available as summarised in Section 2. Hence, despite the increase of research efforts on the vulnerability of critical infrastructure against natural, environmental and human-induced hazards, there is still a lack of systematic vulnerability assessment against multiple hazards, considering also the effects of deterioration, e.g. ageing, and mitigation measures, e.g. retrofitting, in the fragility response.
The rapidity of the recovery after disruption due to a hazard event is expressed through restoration functions for the infrastructure assets. The available restoration models correlate the recovery time with the functionality reached for a given damage level, e.g. Gidaris et al. (2017) for bridges, Galbusera et al. (2018) for port facilities, Castillo (2014) for electric power systems, Luna et al. (2011) for water distribution systems and HAZUS-MH (2011) for various infrastructure assets. They are typically based on expert judgements, following a linear, e.g. Bocchini and Frangopol (2012b), stepwise formulation, e.g. Padgett and DesRoches (2007) or normal distribution, e.g. HAZUS-MH (2011). The development of reliable restoration models is a challenge because the recovery time depends on the available resources and practices of the owner, the type of hazard and the extent of the damage. Furthermore, the functionality and restoration time of assets with multiple components, for example, bridges, is dependent on the damage of the sub-components, e.g. bearings, piers, deck, abutments, foundation. This includes different restoration tasks and uncertainties and, therefore, a probabilistic approach is more appropriate. For example, Decò et al. (2013) proposed a probabilistic evaluation of seismic resilience of bridges, accounting for the uncertainties in the recovery pattern, i.e. residual functionality, idle time, duration of recovery and target functionality, as a support tool for decision making within the bridge life-cycle. The restoration times for the different tasks and components can vary considerably, while a range of values or a mean value and a standard deviation can describe the expected recovery time (Bradley et al., 2010; Karamlou and Bocchini, 2017). In general, the restoration models are mainly available for earthquake-induced hazards, while little information for other hazards is provided, e.g. by HAZUS-MH (2011) for tsunami.
Important gaps in current resilience assessment frameworks for infrastructure assets are that they consider only single hazards and one occurrence of the hazard. A more reliable assessment of the vulnerability, risk and resilience of critical infrastructure should consider the occurrence of multiple hazard events, potentially of different natures including their temporal variability during the lifetime of the asset as well as the asset deterioration and/or improvement. The development of methods for lifetime resilience assessment (Yang and Frangopol, 2018) is an urgent need of paramount importance for infrastructure owners and operators, to enhance safety, leading to significant cost savings and efficient allocation of resources toward resilient infrastructure.
This study aims at filling this urgent knowledge gap by (1) providing a sound classification of multiple hazards affecting critical infrastructure, (2) reviewing existing approaches and techniques for dealing with the effect of multiple hazards in the infrastructure resilience assessment, and (3) developing and applying a resilience assessment framework for critical infrastructure assets exposed to a sequence of individual and/or multiple natural, environmental and human-induced hazard events. This framework considers the factors that reflect redundancy and resourcefulness in infrastructure, i.e. (i) the robustness to hazard actions, based on realistic fragility functions, and (ii) the rapidity of the recovery after the occurrence of different levels of direct damage and induced consequences, based on realistic restoration and reinstatement functions respectively, enabling adjustments to the time of initiation of restoration after the hazard event (idle time), the type of the restoration actions and the sequence of hazards. In Section 2 below, a classification of multiple hazards is given, by also including relevant examples from real systems subjected to hazard sequences. Subsequently, the proposed conceptual framework for resilience assessment is described. The output of the framework is a resilience index, which is a function of the time-variant functionality of the infrastructure over the restoration time for the hazard scenarios. In Section 3, an application of the proposed framework is given by analysing the resilience of a typical highway bridge under two realistic multi-hazard scenarios, both involving the occurrence of a flood and an earthquake event. In the first scenario, it is assumed that the bridge is fully restored after the occurrence of the flood event and before the earthquake strikes the bridge. For the second scenario, the earthquake is assumed to occur during the restoration process following the occurrence of the flood. The results of the resilience assessments for the two cases are presented and discussed in Section 4. The proposed framework and application contribute to the enhancement of current practices for resilience-based management of infrastructure assets by shifting toward the multi-hazard lifetime resilience assessment. The paper concludes with a demonstration of the importance of the framework and how this can be utilised to estimate the resilience of networks to provide a quantification of the resilience at a regional and country scale.
Section snippets
Resilience assessment framework for infrastructure exposed to multi-hazard
This section describes the proposed resilience framework for infrastructure assets exposed to multiple hazards. It is recognised that due to the diversity of infrastructure assets and the diversity of hazards and combinations, it will only be realistic if a number of critical scenarios are described, yet, an effort was given for the framework to be holistic and representative for a wide range of critical infrastructure. Section 2.1 introduces a classification of multiple hazards for critical
Description of the case study
This section illustrates the application of the framework described above to a realistic case study, consisting of a three-span prestressed concrete bridge, shown in Fig. 6, exposed to a sequence of hazard effects (flood and earthquake), which are independent hazards different in nature (category I in Section 2.1). Although the case study does not correspond to any real bridge, it is representative of a very common bridge class. This is a typical fully integral bridge, i.e. has no expansion
Conclusions
This paper proposes an integrated framework for the resilience assessment of infrastructure assets exposed to multiple hazards characterized by diverse nature, impact and occurrence time. The framework accounts for (i) the robustness of the assets to hazard actions, based on realistic fragility functions for individual and multiple hazards, and (ii) the rapidity of the recovery, based on realistic reinstatement and restoration models after individual and multiple hazard events. The framework
Acknowledgments
This study has received funding by the European Union H2020-Marie Skłodowska-Curie Research Grants Scheme MSCA-IF-2016 (grant agreement No 746298: TRANSRISK-Vulnerability and risk assessment of transportation systems of assets exposed to geo-hazards).
References (133)
- et al.
Effects of SSI and lining corrosion on the seismic vulnerability of shallow circular tunnels
Soil Dynamics Earthquake Engineering
(2017) - et al.
Fragility of transport assets exposed to multiple hazards: state-of-the-art review toward infrastructural resilience
Reliab. Eng. Syst. Saf.
(2019) - et al.
Preparedness against mobility disruption by floods
Sci. Total Environ.
(2019) - et al.
Probabilistic seismic performance and loss assessment of a bridge–foundation–soil system
Soil Dyn. Earthq. Eng.
(2010) Risk analysis and management in power outage and restoration: a literature survey
Electr. Power Syst. Res.
(2014)- et al.
A methodology for quantitative vulnerability assessment of coupled multi-hazard in Chemical Industrial Park
J. Loss Prev. Process Ind.
(2019) - et al.
Framework for analytical quantification of disaster resilience
Eng. Struct.
(2010) - et al.
Risk and resilience assessment of bridges under mainshock and aftershocks incorporating uncertainties
Eng. Struct.
(2015) - et al.
Travel time resilience of roadway networks under disaster
Transp. Res. B Methodol.
(2014) - et al.
Evaluating and optimizing resilience of airport pavement networks
Comput. Oper. Res.
(2014)