Accepted Manuscript

Fragility of transport assets exposed to multiple hazards:

State-of-the-art review toward infrastructural resilience

Sotiris Argyroudis , Stergios A. Mitoulis , Mike G. Winter ,

Amir M. Kaynia

PII: S0951-8320(18)31525-4
DOI: https://doi.org/10.1016/j.ress.2019.106567
Article Number: 106567
Reference: RESS 106567

To appear in: Reliability Engineering and System Safety

Received date: 18 December 2018

Revised date: 8 May 2019
Accepted date: 7 July 2019

Please cite this article as: Sotiris Argyroudis , Stergios A. Mitoulis , Mike G. Winter , Amir M. Kaynia ,
Fragility of transport assets exposed to multiple hazards: State-of-the-art review to-
ward infrastructural resilience, Reliability Engineering and System Safety (2019), doi:
https://doi.org/10.1016/j.ress.2019.106567

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Highlights

 Vulnerability assessment methods for transport infrastructure exposed to multihazards

 Hazard effects, asset typologies and fragility assessment methods, mitigation measures
 Challenges in fragility based on numerical models, emphasis on floods and earthquakes
 Novel concept of transport System of Assets in diverse ecosystems is introduced
 Current trends insights and future research opportunities in multiple hazard fragility

T
IP
CR
US
AN
M
ED
PT
CE
AC

1
 ACCEPTED MANUSCRIPT

Fragility of transport assets exposed to multiple hazards:

State-of-the-art review toward infrastructural resilience

Sotiris Argyroudis1, Stergios Α. Mitoulis2, Mike G. Winter3, Amir M. Kaynia4

1
Marie-Curie Research Fellow, Department of Civil and Environmental Engineering, University of Surrey,
Guildford, United Kingdom, [email protected]
2
Associate Professor, Department of Civil and Environmental Engineering, University of Surrey, Guildford,

T
United Kingdom, [email protected]
3

IP
Head of Ground Engineering and Honorary Chief Scientist, Transport Research Laboratory (TRL), Edinburgh,
United Kingdom and Visiting Industrial Professor, University of Portsmouth, Portsmouth, United Kingdom,
[email protected]

CR
4
Professor, Department of Structural Engineering, Norwegian University of Science and Technology (NTNU) and
Discipline Leader, Norwegian Geotechnical Institute (NGI), Norway, [email protected]

US
Abstract Vulnerability is a fundamental component of risk and its understanding is important for
characterising the reliability of infrastructure assets and systems and for mitigating risks. The vulnerability
AN
analysis of infrastructure exposed to natural hazards has become a key area of research due to the critical role
that infrastructure plays for society and this topic has been the subject of significant advances from new data
and insights following recent disasters. Transport systems, in particular, are highly vulnerable to natural
hazards, and the physical damage of transport assets may cause significant disruption and socioeconomic
M

impact. More importantly, infrastructure assets comprise Systems of Assets (SoA), i.e. a combination of
interdependent assets exposed not to one, but to multiple hazards, depending on the environment within which
these reside. Thus, it is of paramount importance for their reliability and safety to enable fragility analysis of
ED

SoA subjected to a sequence of hazards. In this context, and after understanding the absence of a relevant
study, the aim of this paper is to review the recent advances on fragility assessment of critical transport
infrastructure subject to diverse geotechnical and climatic hazards. The effects of these hazards on the main
PT

transport assets are summarised and common damage modes are described. Frequently in practice, individual
fragility functions for each transport asset are employed as part of a quantitative risk analysis (QRA) of the
infrastructure. A comprehensive review of the available fragility functions is provided for different hazards.
CE

Engineering advances in the development of numerical fragility functions for individual assets are discussed
including soil-structure interaction, deterioration, and multiple hazard effects. The concept of SoA in diverse
ecosystems is introduced, where infrastructure is classified based on (i) the road capacity and speed limits and
AC

(ii) the geomorphological and topographical conditions. A methodological framework for the development of
numerical fragility functions of SoA under multiple hazards is proposed and demonstrated. The paper
concludes by detailing the opportunities for future developments in the fragility analysis of transport SoA
under multiple hazards, which is of paramount importance in decision-making processes around adaptation,
mitigation, and recovery planning in respect of geotechnical and climatic hazards.

Keywords: fragility functions; reliability in quantitative risk analysis; highway and roadway infrastructure;
numerical modelling; earthquakes; landslides; liquefaction; flooding; scouring; multiple hazards

2
 ACCEPTED MANUSCRIPT

1 Introduction
Natural hazards, such as ground movements, debris flow, earthquakes, and floods are major threats to
infrastructure in many regions around the world. More importantly, societies and businesses rely heavily on
transport infrastructure. In addition to the loss of life and the physical loss of the assets themselves, damage to
transport infrastructure may cause significant socio-economic losses. For example, the heavy 2007 rainfall in
the UK affected the road network, with the cost estimated at £60 million (The Parliamentary Office of Science
and Technology, 2010); also, during the 2009 floods in Cumbria, at least 20 bridges were destroyed or
damaged, causing at least one fatality, £34m of repair and replacement costs and large societal impact
(Cumbria County Council, 2010). The 2012 flood events resulted in a total of 131 damaged bridges in the
same region mainly due to scour (Zurich Insurance Group and JBA Trust, 2016). The 2010-2011 Canterbury

T
earthquake sequence in New Zealand caused extensive damage to road networks due to liquefaction that
resulted in settlements, lateral spreading, sand boils and water ponding on the road surfaces. Moreover, rock

IP
falls led to several road closures (Kongar et al. 2017). Extensive bridge damage was reported after the 2010
Maule earthquake in Chile due to inadequate seismic design. The effects of structural irregularity and soil

CR
liquefaction were proven to be critical for the performance of bridges (Kawashima et al. 2011). In the U.S.A,
hydraulic in nature actions, such as scour and debris build-up have been established as the most catastrophic
causes of bridge collapses. They represent more than 50% of the cases (US National Bridge Inventory, Cook

US
et al. 2015), resulting in an average annual repair cost of $50m (Lagasse et al. 1995). In Europe, weather
stresses represent 30% to 50% of road maintenance cost (up to €13bn p.a.); 10% of these costs are associated
with effects of extreme weather events (Nemry and Demirel, 2012). As an example, flooding over large areas
AN
of the Danube and Elbe rivers in Central Europe on May-June 2013, caused road and rail closures, erosion of
embankments and streets, damage to bridges and landslides blocking railways. The high-speed rail links
between Frankfurt and Berlin, and between Berlin and Hannover had to be closed for repairs for several
months after the event. The total economic damage was estimated at more than €12bn (MunichRe, 2013).
M

Based on a recent international expert elicitation workshop the damage of bridges due to hydraulic causes is
strongly related to the history and accumulation of predominantly scour damage (Lamb et al. 2017). As a
result, undetected scour may lead to unexpected failures for flood events of smaller intensity. Among the
ED

critical threats to infrastructure around the world, scour is cited as the most common cause of bridge failure
(Kirby et al. 2015). In the UK and based on a record of scour-induced failures spanning over 173 years, it is
estimated that the annual probability of failure incidents is approximately 27% (van Leeuwen and Lamb,
PT

2014). Projected changes to river flows, including climate change effects (Pant et al. 2018), would increase
scour by over 8% of all the approximately 4,200 railway and 8,700 main road bridges which cross
watercourses in the UK, placing them at high risk of failure (Dawson et al. 2018). Similar vulnerabilities have
CE

been identified for transport assets at estuaries or near the sea-side affected by tidal water, as well as sea level
rise, which may be exacerbated due to climate change.
Multi-hazard and extreme weather effects on transport infrastructure is a strategic priority in European
AC

research and have been addressed by recent research projects. In particular, INFRARISK (Clarke and O’Brien
2016) developed a multi-hazard risk assessment methodology to perform stress testing for European transport
infrastructure networks due to extreme events, i.e. earthquakes, floods, landslides, based on available fragility
functions or expert judgment approaches, providing a practical and operational framework for decision
making. RAIN project (O’Brien et al. 2015) identified critical land transport infrastructure exposed to extreme
weather events, reviewed its failures and the current means of protecting them and developed an
understanding of how infrastructure failure leads to societal vulnerability and insecurity through a risk-based
decision-making framework. INTACT project (Reder et al. 2018) addressed the resilience of critical
infrastructure to extreme weather events in the form of a publicly accessible Wiki and a risk management
decision framework that facilitates cross-disciplinary and cross-border data sharing providing potential end-
users with a means to determine the impact of extreme weather events to their infrastructure. WEATHER

3
 ACCEPTED MANUSCRIPT

project (Doll et al. 2012) analysed the economic losses of extreme weather conditions, i.e. hot and cold spells,
floods, landslides, wildfires, and storms, on transport systems and the wider economy and explored adaptation
strategies for reducing them in the context of sustainable policy design. Similar efforts have been made in the
US, to deploy resilience solutions to current and predicted future extreme weather events (Wright and Hogan
2008; FHWA 2012b; 2013; NCHRP 2014). These projects facilitate the better understanding of the impacts of
natural hazards and climate change on transport systems and provide state-of-the-art knowledge on risk
analysis frameworks; however, a systematic and accurate representation of the performance of transport assets
subjected to geo-hazards is still lacking. Thus, reliable assessment of the vulnerability of, and the associated
risks to, transport infrastructure subjected to critical hazards is of paramount importance, since it will enable
the efficient allocation of resources toward resilient transport networks.
The objective of this paper is to prove a well-informed debrief of the understanding and applicability of the

T
available methodologies for the vulnerability assessment of transport infrastructure in inter-urban

IP
environments subjected to multiple natural hazards and to identify current trends and gaps in the knowledge.
This effort is directed towards enabling the enhancement of the safety of infrastructure assets toward more

CR
resilient and robust transport assets and networks. Based on the extensive literature review and to the authors’
best knowledge, the results of this paper are unique, as most of the published research related to transport
infrastructure focuses on the seismic fragility assessment of bridges. This review commences with an outline
of the critical hazards and their effects on transport infrastructure, along with an introduction to the recent

US
trends on quantitative risk analysis together with the design and assessment of transport assets exposed to
hazards. The next section describes the common damage modes of the main transport assets. Subsequently, a
review of fragility functions for transport assets under different natural hazards is provided. The review on the
AN
fragility of bridges is selective in this paper as bridges have been covered adequately in published research. In
the next part of the review, the literature is summarised in terms of how different researchers have tackled the
main modelling challenges in the generation of analytical fragility functions for assessing physical damage.
M

These include the soil structure interaction and deterioration effects, the treatment of uncertainties and the
modelling of multiple hazard effects. The following section introduces the concept of the infrastructure
System of Assets (SoA) in ecosystems as a combination of interdependent assets exposed to diverse hazards.
ED

A methodology for the development of numerical fragility functions for SoA is also proposed. The study
concludes with the gaps in the knowledge that need dire attention, and on this basis, recommendations for
future developments are provided.
PT

1.1 Natural hazards and their effects on transport infrastructure

Natural hazards and weather-related hazards have different effects on various transport assets. The impacts on
CE

the transportation system, from changes in temperature, precipitation, sea-level rise and hurricanes, along with
possible adaptation strategies in the United States, are summarised in TRB (2008) and NCHRP (2014). Table
1 summarises the effects of critical geotechnical and hydraulic hazards to road transport infrastructure and
AC

possible mitigation measures. Apart from the asset-specific mitigation measures shown on the table, the
following measures may be employed for any asset:
 Improve asset data knowledge and understanding, for example, for identification of high-risk
locations, definition of existing vulnerabilities and interdependencies of assets and networks, which is
a major challenge in the design of resilient infrastructure (Vespignani 2010);
 contingency planning measures for rapid repair and re-routing of traffic;
 monitoring of critical assets in combination with response planning;
 design of new structures to account for additional stresses exacerbated due to climate change (Stern et
al. 2013), e.g. design for extremes and multiple hazards, in the context of the resilience-based design
(e.g. Franchin 2018; Almufti and Willford 2013).

4
 ACCEPTED MANUSCRIPT

Table 1. The effects of critical hydraulic and geotechnical hazards on road infrastructure, and relevant mitigation measures.

Hazard Asset affected Damage/Impact Typical mitigation measure

Fluvial/river flood Bridges over a river or Scour of piers/abutment foundations Improve existing scour protection system;
due to extreme stream (general, contraction, local scour); retrofitting of bridge foundations with
precipitation impact to the deck due to additional piles (e.g. Hung and Yau 2017);
(including overtopping; failures of bank and bridge scour monitoring (e.g. Prendergast and
overbank and flash riprap protections Gavin 2014)
flooding)
Embankments and Scour due to high river levels; culvert Improve drainage (increase ditch and culvert
cuttings washout; slope erosion and capacity); install geotextiles and geogrids to
instability, seepage of water prevent cracking

Pavements Inundation, washout, deterioration, Improve/maintain drainage (increase ditch and

T
and loss of skid resistance due to culvert capacity)
excess water

IP
Pluvial/surface Embankments and Settlement, sliding/slumping; Improve/maintain drainage (increase ditch and
flood due to cuttings swelling of clay materials culvert capacity); install geotextiles and

CR
extreme geogrids to prevent cracking
precipitation
Pavements Inundation, washout/cracking Improve/maintain drainage (Willway et al.
2008)

US
Underground water Tunnels, bridges, Corrosion of reinforcement; Improve/maintain drainage
retaining walls degradation of concrete strength
Sea level rise and Coastal roads, Scour effects; overtopping and wave Renewal programme;
storms (flood causeways over a lake erosion, softening by soil saturation, consider sea level rise in the new designs
AN
surge) or sea seepage (internal erosion), piping
Landslides (rainfall Pavements Closure by debris flows or mudflows Warning signs; protection measures (debris
or earthquake- Cuttings, embankments Slope: failures along discontinuities, shelters, barriers, fences, ditches, tunnels);
induced, including and natural slopes toppling failure and falls, stabilization measures (e.g. reduce slope
sliding, debris flow, translational failure; angles, rock anchors, shotcrete, jet grouting);
M

mudflow) Embankment: instability due to the planting of appropriate vegetation; improve

failure of the foundation, failure of drainage; removal of the exposed asset (Winter
the material 2014)
ED

Tunnel portals Rockfalls Rock removal; netting of rock

cutting/protection barriers; rock sheds

Drought Cuttings/slopes/ Ground stability impacts Removal of prone materials; vegetation

PT

embankments (desiccation, shrinkage of clay management

materials); creation of ruts
Extreme hot Bridge components Expansion of the deck Use of expansion joints; use of sliding bearings
weather
CE

Signalling and Malfunction due to overheating of Use of uninterruptable power supplies (UPS);
Intelligent Transport power lines (indirect) replacement of ageing cables
Systems (ITS)
AC

Cuttings/embankments Erosion, shrinkage due to soil Use of sun sheds (slopes)

moisture change
Pavements “Thermal fatigue”; thawing effects Use of geogrids; insertion of expansion joints
and cracking; melting of bitumen; for concrete roads to prevent “blow-ups”;
loss of skid resistance application of more deformation resistant
surfacings; trees not planted <15m from the
road edge (Willway et al. 2008; FHWA 2015)

Wildfires Pavements/ bridges/ Burning of asphalt; Installation of high-volume sprinkler systems;

tunnels/ signalling and Failure or melting of components replacement of wood poles and other structures
ITS with fire-resistant materials, e.g. steel or
concrete

5
 ACCEPTED MANUSCRIPT

Pavements/bridges Limited visibility (indirect) n/a as the risk is indirect and the hazard is
addressed for all elements/assets at risk

Natural slopes Slope degradation and soil erosion n/a as the risk is indirect and the hazard is
(indirect) addressed for all elements/assets at risk

Snow Pavements Closure due to avalanches (in n/a, i.e. not technical as the effect cannot be
mountainous areas); accumulation of mitigated prior to the hazard
snow
Cuttings/slopes/ Melting of snow; slope instability Use of snow sheds
embankments
Cold & freeze Pavements “Thermal fatigue”; frost heave; Sufficient thickness of pavement and non-frost
asphalt cracking susceptible base course; use of granular rock
caps; remove and replace frost-susceptible

T
subgrade; use of geotextiles/frost blankets
(Willway et al. 2008; AASHTO 1993)

IP
Embankments Thermal erosion; creep; heave Backslope protection blankets using gravel or
crushed stone protection layer over a geotextile

CR
Bridge components Contraction of the deck Use of expansion joints; use of sliding bearings
Signalling and ITS Malfunction due to low temperatures Use of uninterruptable power supplies (UPS);
use of electric heaters; replacement of ageing

Slopes US
Instability of rock slopes
cables

Backslope protection blankets

AN
Wind Cable-stayed and Aerodynamic effects (vortex Damper systems and stiffeners; spoilers
suspension bridges shedding, galloping, flutter);
turbulence
Signs and signals Collapse Strengthening equipment
M

Pavements Closure due to windblown and Wind warnings

damaged trees
Earthquake (ground Bridges, tunnels, Different damage modes to structural For bridges: strengthening/replacement of
shaking, ground retaining walls, elements (piers, abutments, bearings, bearings; restrainer cables; seat extension;
ED

failure due to pavements, foundations) and geotechnical assets steel, fiber composite or steel jacketing of
liquefaction or fault embankments, cuttings (settlement, heave, rotational/slump piers; pier cap strengthening or replacement;
rupture) failures etc). See section 3. energy dissipation devices (Buckle et al. 2006).
PT

For approach fills to bridge abutments:

structural approach slabs; alternative materials,
such as rubber-sand mixtures (Mitoulis et al.
2016; Argyroudis et al. 2016).
CE

For tunnels in rock: rock bolts; shotcrete or

replacement of weak lining. For tunnels in soft
soil: spot repairs; contact grouting; ground
AC

improvement; liner replacement; construct

special joints.
For embankments/pavements: compact soft or
loose soils; improve foundation drainage; add
berms or struts (Power et al. 2004).
For slopes/cuttings (see Landslide hazard).

Any hazard that Pavements Closure due to collapse/failure of Increase the distance between buildings and
leads to impacts overpass bridges or adjacent roads; replace pole lines with buried cables;
due to geographic buildings and/or overturn of utility use of durable materials for the utilities;
interdependencies poles (power, communication etc) or separation of underground utility installations
(mainly in urban signalling systems from roadways facilities; encasement of
environments) Damage/closure due to failure, pipelines; increase of cover depth; provide

6
 ACCEPTED MANUSCRIPT

leakage or explosion/fire of gas, oil, adequate coating and wrapping of pipes/cables;

water, sewerage pipes buried under joint highway and utility planning and
the roadway development
Bridges Damage of cables (electric power, Protection of pipes through coating, wrapping
fibre-optic communication) or pipes or fiberglass shields; provision for shut-off
(water, gas) carried by the bridge systems for gas, oil and hazardous material
pipes

It is observed that for certain assets the fragility assessment requires a cross-disciplinary expert judgement
including input from structural, geotechnical or mechanical/electrical engineering. Similarly, the mitigation
measures are not solely of an engineering nature, as the contingency planning and preparedness may need the

T
involvement of engineers and experts from other disciplines (e.g. economists, foresters, geologists),

IP
stakeholders, operators, and owners.
1.2 The concept of vulnerability and quantitative risk analysis (QRA)

CR
Hazards refer to events related to geological, meteorological and hydrological phenomena that are
characterized by intensity, spatial variability and a probability of occurrence in time. Natural hazards are
independent accidental actions (EN1991-1-7, 2006). In case of transport networks, hazards and their

US
interactions strongly depend on the geomorphological and topographical surroundings (see also section 6.1).
The concept of multi-hazard design and assessment has been introduced by Bruneau et al. (2017) among
others. Hitherto, no common nomenclature has been established for the phenomenally similar meaning in
AN
engineering terms between multiple hazards, multi-hazard effects, cascading, cross-hazards among others
(Kappes et al. 2012). The vulnerability of transport systems is commonly assessed in terms of physical
vulnerability of its components depending on the physical characteristics of the infrastructure assets, e.g. age,
material, structural types, and functional vulnerability depending on the functional characteristics of the
M

network, e.g. capacity and speed. The risk analysis of a network includes hazard identification, vulnerability
evaluation of the infrastructure exposed to the given hazards and risk assessment in terms of economic,
functional and social losses. The vulnerability is a fundamental component in risk analysis under any natural
ED

or climatic hazard, and its accurate estimation is essential in making reasonable predictions of losses and
consequences. Risk analysis is distinguished in three levels, depending on the input data, procedures of the
analysis and risk output: qualitative, semi-quantitative and quantitative (Eidsvig et al. 2017). All approaches
PT

aim to classify the most vulnerable parts of the network that require detailed analysis and to provide support
for planning, preparedness, and prioritization of risk-reduction measures. In the first approach, hazard and
vulnerability are described through qualitative estimates using descriptive ranks, e.g. high, moderate and low.
CE

In the second approach, the risk is estimated based on semi-quantitative vulnerability indicators using
numerical thresholds (ranking) and quantitative estimates of the frequency of the natural hazard (e.g. Petrucci
and Gulla 2010; Eidsvig et al. 2017). The concept of the quantitative risk analysis (QRA), which quantifies
AC

the probability of a given level of loss and the associated uncertainties, has also been touched by Eurocodes
(EN1991-17, 2006). Thus, QRA quantifies the risk in an objective and reproducible manner, providing a
robust basis for the prioritisation of mitigation actions, efficient risk management for stakeholders and owners,
and prediction of losses for the insurance industry (Corominas et al. 2014); recent examples of such QRA
approaches to debris flow risk on a road network, in this case relating to the probability of fatalities amongst
road users, are given by Wong and Winter (2018) and Winter (2018). Based on the above, it is clear that
predictions of losses and associated impacts on the asset, e.g. bridge, tunnel, and in extension at the network
level, as in for example highways, are realistic only if the vulnerability is estimated based on advanced
approaches that reliably predict the damageability of the assets. The latter is commonly expressed through
vulnerability and/or fragility functions, which are discussed in detail in section 4. Risk analysis is performed
for a single component, e.g. a bridge or a road cut, linear features, e.g. part of a highway or a network in

7
 ACCEPTED MANUSCRIPT

regional or national level or areas, e.g. counties (Suh et al. 2011; CEREMA 2014; Jenelius and Mattsson,
2015).
1.3 QRA at a network level
Different approaches have been adopted for the performance assessment of transport infrastructure at the
network level and the quantification of the consequences of the disaster events. Different levels of analysis,
e.g. connectivity, capacity, integrated loss estimation, have been applied depending on the time frame
considered such as emergency phase or economic recovery phase, the scale and type of system, that is urban,
regional, and national, the objectives of the analysis and the needs of stakeholders (emergency planning,
mitigation or network extension planning, insurance) and the information available. An overview of the
modelling techniques on the transport infrastructure system performance in disasters is given by Faturechi and

T
Miller-Hooks (2015). The concepts and measures of different approaches are generally categorized as risk,

IP
vulnerability, reliability, robustness, flexibility (also known as adaptability), survivability, and resilience.
Performance metrics include the travel time, flow/capacity, accessibility, topological measures, e.g.
connectivity, direct and indirect economic losses. These quantitative measures are used in disaster

CR
management for the prioritisation of mitigation, preparedness, and adaptive actions. The modelling of possible
disasters and associated uncertainties includes specific scenarios, simulation of a wide range of scenarios, use
of probability distributions, identification of worst-case performance, or historical scenarios. Mathematical

US
models of system performance are classified as analytical, e.g. risk matrix, event tree analysis, fault tree
analysis, analytical hierarchy process, simulation, e.g. through Monte Carlo simulation, or optimisation by
deterministic or stochastic models. Khademi et al. (2015) reviewed the methods related to the vulnerability of
AN
transport networks due to natural disasters, concluding that accessibility indexes often serve as indicators of
network vulnerability. Muriel-Villegas et al. (2016) classified the available approaches for transport network
reliability to natural disasters in three main areas, namely connectivity reliability, performance reliability, and
vulnerability. An overview of network vulnerability analysis, classified to scenario-specific, strategy-specific,
M

simulation, and mathematical modelling approaches is provided in Murray et al. (2008), while the methods
and challenges in modelling and simulation of interconnected infrastructure are discussed by Eusgeld et al.
(2011) and Ouyang (2014).
ED

In the case of earthquake hazards, most of the efforts have addressed the direct seismic shaking effects,
focusing on bridges, which is the most critical asset (e.g. Miller and Baker 2015). The interactions of the
urban road network with the built environment in post-earthquake conditions have been examined by Goretti
PT

and Sarli, 2006; Argyroudis et al. 2015; Ertugay et al. 2016; and Zanini et al. 2017, considering the effect of
building collapses to the connectivity of the network. The extent of the debris of the collapsed buildings that
affects the functionality of the road is estimated through simplified geometric models. The damage estimation
CE

using fragility functions has been used in the design of new tunnels and in the implementation of earthquake
early warning systems for high-speed railways (Fabozzi et al. 2018).
AC

The effects of multiple hazards on a network level have been studied by Hackl et al. (2018) who proposed and
applied a modular approach to couple rainfall, runoff, flood, mudflow, physical damages of bridges and
pavements, functional loss, traffic, and restoration modelling. Consequences were monetized into direct and
indirect costs, considering restoration interventions, prolongation of travel time, and lost trips. This model has
been used by Lam et al. (2018) to conduct a stress test on a road network affected by floods and rainfall-
triggered mudflow, using fragility functions and functional capacity loss functions.
1.4 Current policies, strategies and guidelines for assessment of transport infrastructure
The importance of risk assessment is proven by the recent research interest in quantitative risk analysis, which
is related to the protection of critical infrastructure assets subjected to natural hazards. This is in line with the
current strategies for adapting infrastructure to climate change and natural disasters as reflected in various

8
 ACCEPTED MANUSCRIPT

governmental decisions and documents in Europe, (e.g. Council Directive 2008/114/EC; SWD 2013/137;
COE 2011), USA (FHWA 2013) and other countries as for example in UK (e.g. Cabinet Office UK 2011;
Highways England 2016) and New Zealand (NIU 2011). These frameworks emphasize that the design and
assessment should integrate extreme weather events and climate change induced risks into asset management
practices toward more resilient infrastructure. For example, practices for predicting 100-year floods in the
design may no longer be valid, while greater extremes and more frequent events should be assumed.
Furthermore, transportation systems have several vulnerabilities, which are poorly understood and difficult to
quantify (Markolf et al. 2019). These vulnerabilities include direct physical, direct non-physical, related to
travellers’ behaviour and system operators’ decision making, indirect physical, due to physical or geographic
interconnections and indirect non-physical, due to cyber or logical interdependencies with other infrastructure.
In this regard, a risk-based asset management system should include accurate inventories and mapping of

T
assets, sound maintenance practices, hierarchical prioritisation of critical assets and assessment based on a

IP
probability and impact assessment. In this context, results of the assessment will not only support planning
prevention, adaptation and mitigation of disruptive events, but will also inform the recovery processes
required to maintain functionality immediately following a severe event. Therefore, the adaptation strategies

CR
go beyond risk management to a resilience-based management concept that determines how a system can
adapt to and recover from shocks, and not just avoiding or mitigating them (Cimellaro et al. 2010; Meyer and
Weigel 2011; Schweikert et al. 2014; Mattsson and Jenelius 2015; Espinet et al. 2016).

US
Risk-based management approaches are widely applied by transport infrastructure owners and stakeholders to
prioritise the assets with a higher risk that require more detailed assessments or mitigation measures. These
approaches are usually given in the form of guidelines and provisions by national transport departments and
AN
governmental organisations. The risk assessment is commonly based on screening methods to calculate a risk
score using different criteria and factors that describe the hazard conditions, the vulnerability of the assets and
their importance. For example, guidelines to identify and prioritise seismically deficient bridges in the US are
M

provided by FHWA (Buckle et al. 2006). The screening is based on seismic rating methods using indices and
expected damage. The indices describe the structural/geotechnical vulnerability, such as connections,
bearings, piers, foundation, and soil liquefaction, and the hazard intensity. Rating using expected damage is
ED

based on fragility functions and estimation of economic losses for given seismic hazard levels. Prioritisation
includes bridge importance, network redundancy, non-seismic deficiencies, remaining useful life, and other
socioeconomic issues. Seismic screening and evaluation criteria for retaining structures, engineered slopes and
embankments, tunnels, culverts, and pavements are also provided by FHWA (Power et al. 2004). In Europe,
PT

the seismic assessment of bridges will be based on the on-going update of Part 3 of Eurocode 8 (EN 1998-3,
2005).
CE

With regard to guidelines for the design and assessment of bridges under hydraulic actions, the ones by Kirby
et al. (2015) and BD97/12 (2012) are available in the UK, whilst in the US relevant documents are provided
by NCHRP (2010a,b), NCHRP (2011) and FHWA (2012a). Multiple factors are considered to calculate a risk
AC

score including the scour history, the characteristics of the bridge structures and the watercourses that they
cross. The scour depth is estimated for given design return periods based on closed-form solutions.
Vulnerability and risk of transport assets exposed to extreme weather effects are aggravated by climate change
and are assessed on the basis of transportation system sensitivity and exposure to weather effects and adaptive
capacity (FHWA 2012b). Vulnerabilities are assessed through a combination of quantitative measures and
qualitative judgments, based on impact rating scale scorecards, multi-criteria decision analysis or risk matrix
approaches (WSDOT 2011; Yang et al. 2013). Thus, based on the international literature there does not exist a
well-established methodology for quantifying the losses of transport infrastructure exposed to weather effects.
Existing national and international landslide guidelines are reviewed and evaluated by Wang et al. (2012).
Some of these focus on certain topics and issues, e.g. landslide risk management and zoning, mitigation and

9
 ACCEPTED MANUSCRIPT

remediation, slope design, and others are more generic, e.g. geotechnical assessment, land use planning. A
summary of available publications, codes and design practices for earthworks associated with transport
infrastructure is provided by Griffiths and Radford (2012).
Resilience-based assessment and management are the new philosophies that are gradually being adopted in
practical applications of transport assets and are expected to be incorporated in the next generation of
provisions and guidelines (Linkov et al. 2014). In this context, different frameworks and assessment tools
have been proposed in the literature, e.g. Bruneau et al. 2003; Hughes and Healy 2014; Ayyub 2014; Dong
and Frangopol 2015; Chan and Schofer 2015; Rattanachot et al. 2015; Kiel et al. 2016, among others, while
EU projects on this topic have been recently implemented as already presented in the Introduction of this
paper.

T
IP
2 Brief description of the main typologies of transport assets

CR
Important transport assets include bridges, tunnels, culverts, retaining walls, embankments, trenches, slopes
and pavements. The secondary assets include information and communication technology (ICT), signalling,
lighting, and safety (e.g. fences, barriers) components, and buildings, such as tolls or warehouses. Railway

US
systems also include tracks, electric power and communication systems, stations and workshops. Another
distinction of transport assets can be made on the basis of urban and inter-urban networks. Some components,
such as embankments, slopes or trenches, are mainly encountered in inter-urban networks. A significant
AN
difference is the geographic interdependencies of urban systems with other infrastructure, e.g. buried pipelines
or cables underneath or buildings in the proximity of the roads. Moreover, due to the lower redundancy of the
network compared to the urban ones, the consequences and indirect losses of natural hazards and weather
M

stressors have significantly different impacts on inter-urban transport infrastructure, whilst urban networks
have higher redundancy, yet, greater interdependencies with other interacting networks. For example, closure
of a highway tunnel or bridge can potentially cause higher total losses compared to closure of a main urban
ED

street, as it is easier to follow alternative routes in the second case. However, there are examples of significant
losses in case of failures in urban networks, such as the collapse of the Hansin Expressway during the 1995
Kobe earthquake or the consequences of the flash floods in central European cities in 2013. The focus of this
paper is on inter-urban roads, whilst additional literature would be required for urban and strongly
PT

interdependent networks.
The variation of bridge typologies is greater compared to other transport infrastructure; therefore, the
CE

available classification schemes are diverse, particularly focusing on the seismic behaviour of bridges (e.g.
Applied Technology Council 1985; NIBS 2004; Hancilar and Taucer 2013). The bridge typologies are
commonly based on the number of spans and length, particular design considerations, material, type of pier
AC

and abutment and deck continuity. The SYNER-G taxonomy (Hancilar and Taucer 2013) includes the
following structural characteristics: material, type of superstructure, type of deck, deck structural system, pier
to deck connection, type of pier, number of columns per pier, cross section of pier, spans, type of connection
to the abutments, bridge irregularity, skew, foundation type, seismic design level. Due to the peculiarity of the
bridge abutment, its typology is examined here separately, and its typology is related to the structural type of
the bridge, e.g. stub, partial or full height, integral. Other characteristics are the depth and the soil conditions
of the foundation and the fill material behind the abutment. The depth is dependent on the surrounding
topography and geometry of the abutment, while a critical factor for the backfill material is its degree of
compaction.
The basic parameters of the typology of tunnels are the construction method (bored or mined, cut-and-cover,
immersed), the cross-section shape (circular, rectangular, horseshoe), the depth (surface, shallow, deep), the

10
 ACCEPTED MANUSCRIPT

geological conditions (rock, alluvial) and the supporting system (concrete, masonry, steel, etc.). For example,
ALA (2001) classifies tunnels into four categories according to the quality of construction and the ground
conditions.
The typology of retaining walls is related to the construction, and the most common types are gravity,
cantilevered, sheet piling, bored pile and anchored retaining walls. In addition, the soil material, slope angle,
and water content are relevant parameters in the typology definition of retaining walls.
The main typology characteristics of embankments, trenches and slopes are the geometrical parameters of the
construction, that is, slope angle and height as well as the ground conditions (soil material, water level etc.).
Usually, the transport assets are grouped within classes based on the typology properties, and the vulnerability
is calculated for a model that represents the entire class. This approach is applied for the risk analysis of a

T
large portfolio of assets as it would be very time consuming and computationally expensive to calculate asset-

IP
specific vulnerability models. However, this approach may not be acceptable for some assets within a class
due to inevitable differences and peculiarities of each asset. In addition, significant variabilities exist across

CR
different countries and different classes of assets are encountered depending on the classification of the
transport system. A diversity of assets is also imposed across different transport networks, such as highways,
railways and underground transport systems. Table 2 summarises the main characteristics and typological
parameters for the road infrastructure assets in non-urban environments. Urban road infrastructure has

buildings or poles, the cover depth of pipelines. US

additional characteristics that describe their interactions with the built environment, such as the distance from

Table 2. Main parameters of road assets’ typology.

AN
Asset Typology
High capacity and speed roads Horizontal alignment: variable, mainly depends on the design speed
(e.g. Controlled access Vertical alignment: 3% (desirable max grade)
M

motorways) Standard lane width: 3.65m

Standard hard shoulder width: 3.65m
Standard median strip width: 1.0m
Standard total width per direction (incl. shoulders and median strip): 11.95m for 2 lanes, 15.6m
ED

for 3 lanes, 19.3m for 4 lanes.

Speed limit: 110-120 kmph
Lower capacity and speed roads Horizontal alignment: variable, mainly depends on the design speed
(e.g. Single carriageways) Vertical alignment: 6% (desirable max grade; in hilly terrain steeper gradients may are present)
PT

Standard lane width: 3.65m

Standard hard strip width: 1.0m
Standard total width (including strips): 9.3m (new design), as low as 6.8m (for old design)
Speed limit: <=90 kmph
CE

Embankment /Slope/Cutting Variable height, depending on local geomorphology;

Typical height classification: 0-2.5m, 2.5-5.0m, >5.0m
Typical slope angle: 1.5(H):1(V) - 2(H):1(V), in some cases 2.5(H):1(V) - 3(H):1(V)
depending on the material and design specifications
AC

Drainage type: None, French drain, Open ditch

Bridge Commonly based on the number of spans and length, particular design considerations,
material, type of pier and abutment and deck continuity.
Geometry is variable depending on bridge type and local geomorphology.
Typical pier height: 5.0 to 20.0 m.
Typical deck cross section height: 1.0 to 2.0 m.
Typical span length: 15.0 m to 35.0 m.
Bridge abutment Based on the structural type of the bridge (e.g. stub, partial or full depth, integral abutment).
Other features: depth and soil conditions of the foundation
Geometry is variable depending on bridge type and local geomorphology.
Typical abutment height: 2.0 to 10.0 m.
Tunnel Commonly based on construction method (bored or mined, cut-and-cover, immersed), cross-
section shape (circular, rectangular, horseshoe, etc.), depth (surface, shallow, deep), geological

11
 ACCEPTED MANUSCRIPT

conditions (rock, alluvial), supporting system (concrete, masonry, steel, etc.)

Retaining wall Common rigid types: gravity, cantilevered, sheet piling, bored pile, anchored,
Flexible types: reinforced soil
Variable height depending on retained soil mass, commonly 3.0 to 15.0 m.

Backfill (bridge abutment, Soil material, ground angle, and water content are of main interest
retaining wall)/
Embankment/Slope/Cutting

3 Damage description
The performance levels of an asset are defined through damage thresholds called limit states, which define the

T
boundaries between different damage conditions or damage states. Various damage criteria have been used

IP
depending on the typology of the asset and the method used for the fragility analysis. In analytical methods,
the damage is measured through engineering demand parameters (EDPs), which represent an observable
response parameter of the asset. The number of damage states is variable, e.g. none, minor, moderate,

CR
extensive, complete, depending on the type of asset. Damage states are usually correlated to the traffic
capacity of the assets. In some cases, the damage is correlated to the replacement, repair and enhancement
costs as well as to restoration time and delays due to repairs (NIBS 2004; Werner et al. 2006; Mackie and

US
Stojadinovic 2006; Bradley et al. 2010; Tsionis and Fardis, 2014; D’Ayala et al. 2015). For railway
infrastructure assets, the same damage measures are used as in highway assets, but with different thresholds
between the damage states.
AN
3.1 Bridges
Bridge damage is related to the response of bridge components, i.e. the deck, the piers and foundation,
bearings, abutments and expansion joints (Deng et al. 2016). For piers, the damage measures used in practice
M

are the drift ratio, the curvature, rotation, and displacements. The response of the abutments is usually
described based on its displacement, i.e. abutment gap, and rotation, while the damage measure for bearings is
its longitudinal and transverse shear deformations and/or displacements and for bridge foundations are the
ED

sliding and soil bearing capacity. Damage states have been defined for the specific bridge components and for
the whole bridge (Tsionis and Fardis 2014; D’Ayala et al. 2015). Most studies consider bridges as serial
systems; hence, their damage states are defined by the most vulnerable components (Nielson and DesRoches
PT

2007; Padgett and DesRoches 2009).

Common failure modes due to hydraulic actions include pier or/and abutment settlement or/and tilting due to
loss of support to the foundation or/and hydraulic loading aggravated by debris accumulation, damage to
CE

superstructure or deck falling off abutment or pier, scouring or washout of the embankment behind abutment
(JBA Trust 2014). In case of river crossings, failure mechanisms of rock bank protections include slope
instabilities, sliding, movement of rock cover, migration of sub-layers, etc (Melville and Coleman 2000;
AC

CIRIA et al. 2007). Most of these mechanisms are related to flow characteristics, such as discharge, flow
velocity, and water levels and also to geotechnical characteristics, such as density of materials or pore water
pressure (Roca and Whitehouse 2012).
3.2 Tunnels
Earthquake effects on tunnels include slope instability leading to tunnel collapse, portal failure, roof or wall
collapse, invert uplift, spalling, cracking or crushing of the concrete lining, slabbing or spalling of the rock
around the opening, bending and buckling of reinforcing bars, pavement cracks, wall deformation, local
opening of joints and obstruction at the tunnel portals due to rock falls. Non-seismically induced landslides
can cause similar damage modes. Flooding is not considered as a crucial hazard for tunnels; however,

12
 ACCEPTED MANUSCRIPT

underground water can have a damaging effect on the tunnel lining during its lifetime due to corrosion of
reinforcement or degradation of concrete strength (ITA 1991).
In terms of fragility assessment, damage states commonly describe the response of the main tunnel
components, i.e. liner, portal and support systems. Different damage states and damage measures have been
proposed in the literature depending on the method of fragility analysis and the typology of the tunnel. In
empirical approaches damage states are defined based on the extent of lining cracks (e.g. NIBS 2004; ALA
2001), while in numerical methods damage states are defined based on the exceedance of lining capacity
(Argyroudis and Pitilakis 2012; Argyroudis et al. 2017), number of activated plastic hinges in the liner (Lee et
al. 2016), lateral displacement (Huh et al. 2017) or permanent rotations of longitudinal joint (Fabozzi et al.
2017).

T
3.3 Embankments

IP
Failure modes of embankments subjected to earthquakes are related to ground failures due to soil liquefaction
or dynamic loading. Main failure modes include sliding or slumping of the embankment, cracking at the

CR
surface and settlement of the embankment. Damage states are defined in the literature based on the extent of
settlement or ground offset (NIBS 2004; Werner et al. 2006; JRA 2007; Maruyama et al. 2010; Argyroudis
and Kaynia 2015).

US
The failure mechanisms commonly encountered during flooding involve hydrostatic and hydrodynamic forces
that result from overtopping, seepage forces and the lateral pressure caused by headwater elevation. Common
failure modes in coastal and riverine environments include overtopping erosion, softening by soil saturation,
AN
underseepage, and piping, through seepage (internal erosion) and piping, wave erosion, lateral sliding on
foundations, other failure modes including culvert failures and pavement failures (ALA 2005; Briaud and
Maddah 2016). Damage states are not provided in the literature; however, the ones proposed in case of
earthquake damage can be adopted for floods. The effects of climate change, resulting in excess water, high
M

soil moisture and high temperatures on highway pavements are described by Willway et al. (2008).

3.4 Slopes and Trenches

ED

Earthquake or rainfall-induced landslides and rock falls can cause partial or complete closure of the road or
railbed as well as potential structural damage of the pavement or the rail track. Roads and railbeds constructed
on slopes are subjected to potential failure mechanisms due to large movements of the slopes or slumping of
PT

the sides of the road or railbed. Damage states are defined according to the extent of settlement or ground
offset (NIBS 2004; Argyroudis and Kaynia 2015) and in some cases they are correlated to the permanent
ground deformation as well as to restoration time and traffic capacity (Winter et al. 2014; Argyroudis and
CE

Kaynia 2014; D’Ayala et al. 2015).

3.5 Bridge abutments and Retaining walls

AC

The main form of seismic failure of backfills behind bridge abutments or retaining walls is the backfill
settlement or heaving (White et al. 2007). Structural damage of the abutment wall includes permanent
dislocation, i.e. sliding, rotations. In addition, pounding of the deck to the abutment can seriously affect the
overall response of the bridge due to collision forces. Damage states have been defined (Argyroudis et al.
2013).

4 Fragility analysis methods and intensity measures

13
 ACCEPTED MANUSCRIPT

4.1 General
The degree to which an asset exposed to a hazard can be damaged is commonly expressed through the
damage functions that correlate the severity of the hazard with the level of the expected damage. The most
common types of damage functions used in QRA and reliability analysis are the fragility and vulnerability
functions. Other simplified approaches include indicator-based methodologies, which asses the vulnerability
of an asset or system based on a weighted scoring system for ranking and evaluating the critical characteristics
of the assets (Kappes et al. 2012).
Fragility functions express physical damage and give the probability that the asset exceeds some undesirable
limit state, e.g. serviceability for a given level of environmental excitation, such as force, deformation, or
other forms of loading to which the asset is subjected (Figure 1a). In other words, a fragility function

T
expresses the reliability of a structure as a function of a defined dominant stress variable. The excitation or

IP
stress variable is commonly related to an engineering demand parameter (EDP), which depends on the type of
asset and the hazard that the asset is subjected to (Porter 2015). The fragility functions are usually described
by a lognormal probability distribution, as follows (Eq. 1)

CR
Eq. 1

US
where Pf () is the probability of exceeding a particular limit state, LS, for a given intensity level defined by the
intensity measure, IM, e.g. peak ground acceleration-PGA for earthquake or peak flow discharge for flood
hazard, Φ is the standard cumulative probability function, IM mi is the median threshold value of the intensity
AN
measure, required to cause the ith limit state, and βtot is the total lognormal standard deviation, as per Eq. 2.
Vulnerability functions describe the losses to a given asset or system of assets as a function of environmental
actions (Figure 1b). The losses are commonly expressed in terms of damage repair costs, usually normalised
M

by replacement cost, casualties, commonly given as a fraction of the occupants or travellers, or down-time in
terms of days or fractions of a year, during which the asset or system is not operating. The vulnerability
functions can be generated using the fragility functions by applying consequence analysis that provides
ED

uncertain loss conditioned on damage state. Another means for measuring damage is to express the
functionality loss, such as the reduction of traffic capacity due to a given intensity measure (as per Figure 1c).
Practically, the fragility and vulnerability functions can be derived from empirical, analytical, expert
PT

elicitation and hybrid approaches (Pitilakis et al. 2014; Porter 2015; Silva et al. 2019). Analytical approaches
validated by experimental data and observations from recent events have become more popular, in particular
for earthquake hazard (e.g. Banerjee and Shinozuka 2008; Argyroudis and Pitilakis 2012; Argyroudis and
CE

Kaynia 2015), as they are more readily applied to different structure types and geographical regions, where
damage records are insufficient. Furthermore, the improvement of computational tools, methods, and skills
allow comprehensive parametric studies and better control of the associated uncertainties. The fragility
AC

functions express the vulnerability of assets in quantitative terms and can be directly integrated into the QRA.
Fragility functions encapsulate the concepts of the factor of safety and reliability index, and they are used to
evaluate the reliability of an asset based on a probabilistic approach. In particular, the traditional deterministic
approach to define the safety factor of an asset, i.e. ratio between the design strength and the applied load, is
not representative due to the inherent uncertainties in strength, loading and modelling assumptions adopted.
The reliability index introduces the concepts of uncertainty in capacity and demand but provides information
only about reliability relative to a specific design. On the contrary, fragility functions characterise the system
reliability over the full range of loads, to which an asset might be exposed, thus, provides a more
comprehensive perspective of infrastructural reliability (Schultz et al. 2010). Apart from that, fragility
functions have also been proposed to be used in the design process (Mangalathu et al. 2018) as they provide
information for the performance of an asset under diverse hazards and as a function of different hazard

14
 ACCEPTED MANUSCRIPT

magnitudes and/or frequency design levels. Thus, they provide means of resilient designs because they specify
the likelihood of intermediate damage levels that affect the functionality and restoration of service (Bruneau et
al. 2003).
The generation of fragility functions hinges on the definition of representative intensity measures, IM, which
describe the severity and characteristics of the hazard and are used to correlate the response of each asset with
the hazard. The selection and use of specific IM in the fragility analysis is related to the adopted hazard
model, the typology of the asset, the considered damage modes and the method of fragility analysis. Optimum
IMs are defined based on practicality, effectiveness, efficiency, sufficiency, robustness, and computability
(Mackie and Stojadinovic 2005). In the case of earthquake, several measures of the strength of the ground
motion have been proposed that describe different properties of the motion. They include peak ground

T
acceleration/velocity/displacement, spectral acceleration/velocity/displacement, Arias intensity, etc. Most
common intensity measure types used are the peak ground acceleration (PGA) when ground shaking is the

IP
cause of damage and the permanent ground deformation (PGD) when ground failure, e.g. due to liquefaction,
fault rupture or slope failure, is the trigger of damage. Representative intensity measures for slow-moving

CR
landslides and debris flows are the permanent ground displacement and landslide volume respectively
(Corominas et al. 2014; Winter et al. 2014). In the case of floods, the main parameters are the peak flow
discharge and velocity, flood height (water depth) and hydrograph defined by discharge as a function of time
(Kirby et al. 2015; Lamb et al. 2017; Pregnolato et al. 2017). Scour depth, i.e. at bridge foundation, has been
widely used as intensity measure; however, it is recognised that it is a consequence of the flood hazard and
US
doesn’t explicitly represent the source of the hazard or the load to the structure (Yilmaz et al. 2016). In coastal
environments, wave parameters, such as run-up elevation and significant wave height are also considered. The
AN
rain intensity expressed in mm/day (Jasim and Vahedifard 2017), and the lahar depth (Dagá et al. 2017) have
been considered as intensity measures for transport infrastructure exposed to extreme precipitation and lahar
flows, respectively.
M

Recently, a substantial increase in interest in the seismic fragility assessment of transport infrastructure is
evident in the literature. The studies concern mainly bridge assets (Tsionis and Fardis 2014; Billah and Alam
2015; Gidaris et al. 2017; Stefanidou and Kappos 2018). The available fragility models for railway and
ED

highway infrastructure other than bridges, i.e. tunnels, embankments/cuts, slopes, retaining walls, subjected to
seismic shaking are summarized by Argyroudis and Kaynia (2014). With regard to the available fragility
models for transport assets exposed to ground failures, these were also found to be limited. Generic fragility
PT

functions for tunnels, roads, and bridges subjected to ground failure due to liquefaction and fault displacement
are provided by NIBS (2004), yet not accounting for the typology of assets or the soil conditions. The
following subsections summarise the available fragility functions for transport assets for different hazards.
CE

These fragility models provide measurable means for expressing physical damage, e.g. structural and/or
geotechnical failures, of transport assets subjected to multiple hazards. Thus, these fragility models do not
refer to the loss of non-structural capacity, e.g. the functionality loss of a road due to icy pavement, unless
AC

otherwise stated.
1.0 1.0 1.0
Loss of Functionality (Traffic capacity)

Minor damage
Mean Physical Damage Ratio

Moderate damage
Exceedance Probability

0.8 Extensive damage 0.8 0.8

Complete damage
0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 Intensity
1.0 1.2 1.4
Measure 0.0 0.2 0.4 0.6 0.8 Intensity
1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 Intensity
1.0 1.2 1.4
Measure Measure

(a) (b) (c)

Figure 1. Examples of fragility functions (a), vulnerability function (b), functionality loss function (c).

15
 ACCEPTED MANUSCRIPT

4.2 Bridges
Empirical fragility curves for bridges have been developed based on post-earthquake damage observations,
such as 1994 Northridge and 1995 Kobe earthquakes, using different statistical approaches (e.g. Basoz and
Kiremidjian 1998; Shinozuka et al. 2001). Analytical methods have been widely applied, that include various
simulation techniques such as nonlinear static analysis (e.g. Stefanidou and Kappos 2017), nonlinear time
history analysis (e.g. Kwon and Elnashai 2010; Avşar et al. 2011), incremental dynamic analysis and Bayesian
approaches (e.g. Gardoni et al. 2002).
Soil-structure interaction (SSI) effects on fragility analysis of bridges have been addressed in several studies
(e.g. Stefanidou et al. 2017), while liquefaction-sensitive fragility functions were developed based on
numerical modelling accounting for SSI effects (Brandenberg et al. 2011; Aygün et al. 2011). The combined

T
effect of flood-induced scouring and earthquake to the fragility of bridges has been studied by Dong et al.

IP
(2013), Banerjee and Prasad (2013), Prasad and Banerjee (2013), Kameshwar and Padgett (2014), Guo et al.
(2016), Yilmaz et al. (2016, 2017). Gehl and D’Ayala (2016) developed multihazard fragility functions
through the use of system reliability methods and Bayesian networks. The influence of deterioration effects,

CR
such as corrosion on the seismic fragility has been investigated by Zhong et al. (2012), Choine et al. (2013)
and Ghosh and Sood (2016) among others. The effect of retrofitting measures has also been studied (e.g.
Padgettt and DesRoches 2009). More recently, Karamlou and Bocchini (2017) proposed a methodology to

US
develop probabilistic functionality-fragility surfaces by integrating fragility and restoration functions. Tanasic
et al. (2013) developed analytical fragility functions for multiple span continuous RC bridges considering
degradation of the elastic and plastic soil parameters over time due to scour and Kim et al. (2017) obtained
AN
flood fragility estimates for a case study bridge, considering multiple failure modes due to scour and corrosion
effects to piles and steel reinforcement. Peduto et al. (2018) generated empirical fragility curves for bridge
settlement-induced damage in Amsterdam (Holland) using damage surveys and remote sensing measurements
of settlements.
M

4.3 Tunnels
Expert-based fragility models were included in ATC-13 (Applied Technology Council 1985) and HAZUS
ED

(NIBS 2004), while empirical fragility curves were proposed by ALA (2001) and Corigliano (2007) based on
damage observations in past earthquakes, mainly in Japan and USA. Analytical fragility functions were
developed for tunnels and underground structures under seismic shaking (Argyroudis and Pitilakis 2012;
PT

Mayoral et al. 2016; Huh et al. 2017; Qiu et al. 2018; Avanaki et al. 2018; Nguyen et al. 2019). These studies
considered uncertainties in structural parameters, e.g. tunnel depth, tunnel cross-section, lining thickness, local
soil conditions, e.g. strength of soil or rock mass, and ground motion characteristics (use of a range of input
CE

motions), which are not captured in the empirical or expert-based models. More recently, the effect of
corrosion of the lining has been studied by Argyroudis et al. (2017), whilst Kiani et al. (2016) proposed
experimental fragility functions for circular tunnels as a function of fault rupture.
AC

4.4 Embankments
Empirical fragility curves for road embankments have been generated by Maruyama et al. (2010) and
Nakamura (2015) as functions of peak ground acceleration (PGA) or peak ground velocity (PGV) based on
actual damages observed in Japan. Argyroudis et al. (2013) and Argyroudis and Kaynia (2015) used nonlinear
dynamic analyses and developed analytical fragility functions for cantilever bridge abutments-backfill systems
and embankments and cuts under seismic shaking. Lagaros et al. (2009) proposed analytical fragility functions
for embankments based on pseudo-static slope stability analyses, while Yin et al. (2017) investigated the
influence of retaining walls on embankment’s seismic fragility following an Incremental Dynamic Analysis.
Tsubaki et al. (2016) developed fragility functions for railway embankment fill and track ballast scour based
on recorded observations of railway damage in Japan and simulated overtopping water depth. Lozano-

16
 ACCEPTED MANUSCRIPT

Valcárcel and Obregón (2017) proposed a fragility surface for river levees as a function of the overflow
duration and the height difference between the water level and the levee crest (overtopping failure) and the
river water level and water level behind the levee (piping failure) based on closed form equations and Monte
Carlo simulations. Analytical fragility functions for levees subjected to extreme precipitation as a function of
rain intensity for various rain duration and return periods have been developed by Jasim and Vahedifard
(2017).
4.5 Slopes
The semi-empirical fragility model provided by Pitilakis et al. (2010) as a function of PGA considers the slope
characteristics through the yield coefficient. Wu (2015) developed fragility functions by modelling the
combined effects of infiltration and seismic conditions for a combination of slope geometries based on

T
reliability methods. Fragility curves for roads subjected to debris flow were developed by Winter et al. (2014)
as a function of the landslide volume based on an expert judgement approach. Martinović et al. (2016)

IP
developed generic fragility functions for rainfall-triggered shallow landslides for a range of slope angles in a
railway network as a function of rainfall duration.

CR
4.6 Road pavements
Road pavements are constructed along the entire road network and form the surface for traffic, e.g. passenger

US
and goods vehicles. Therefore, they are constructed on and in bridges, embankments, tunnels and other civil
works. Available research that studies weather impacts, i.e. rainfall, flooding, snowfall, ice, wind, fog, or
temperature, on roads is summarised in Pregnolato et al. (2017). This study was based on observations and
AN
data from past events, modelling and simulations or experiments and is focused on particular occurrences or
regions. The objective was to examine the effect of weather stressors, such as the rainfall intensity, on traffic
conditions, such as the vehicle speed, and not the physical vulnerability, i.e. structural or geotechnical, of the
assets in the form of fragility functions. This paper also contains a function that correlates the floodwater
M

depth with the vehicle speed. Additional functional capacity models for high-speed and local roads sections
inundation have been suggested by Lam et al. (2018).
ED

4.7 The missing fragility and functionality loss models

It is evident that numerous studies have assessed the physical vulnerability of individual transport assets, such
as embankments, tunnels, and mainly bridges exposed to earthquakes. Regarding other hazards, past studies
PT

have focused on the effects of liquefaction, landslides, debris-earth flow and flood and the combined effects of
scouring and earthquakes. Again, these studies mainly concern bridges, and this is also the case for those
investigating the effects of potential mitigation measures, deterioration due to previous hazard events or
CE

ageing effects on the fragility of the assets, the majority of which refer to structural or geotechnical damage.
Additionally, functionality loss models, which quantify induced, i.e. non-structural related, effects hindering
mobility are very limited or completely missing from the literature. Therefore, there is a need to develop well-
AC

informed models that correlate the level of functionality, i.e. traffic capacity, with an appropriate hazard
metric, e.g. floodwater or snow depth, volume and depth of debris. These include, for example, the closure or
partial obstruction due to: i) inundation depth, snow, ice or debris accumulation on the pavement, ii) rockfall,
ice or debris on the road surface at bridge decks, iii) debris or water flow and accumulation on the
road/embankment surface originating from the slope. Such models can be derived based on the available
observational, experimental and modelling analysis, as well as expert elicitations and safety criteria for
vehicles.

5 Modelling challenges in the development of analytical fragility functions

17
 ACCEPTED MANUSCRIPT

5.1 Soil-structure interaction (SSI) effects on transport assets

SSI effects can influence the performance of structures under earthquake shaking or combinations of hazards
(Mylonakis and Gazetas 2000). The numerical modelling of the structure and subsoil is commonly performed
using numerical methods such as finite difference method, finite element method and boundary element
method. According to Bowles (1996) and Dutta and Roy (2002), numerical techniques can incorporate the
effects of material nonlinear behaviour, heterogeneous material conditions, stress anisotropy, hysteretic and
radiation damping as well as changes in geometry of the supporting soil medium in the dynamic soil-structure
interaction analysis. Generally, the selection of the SSI modelling technique significantly affects the
vulnerability assessment of an asset (Kwon and Elnashai 2010). The increase in computational power and
resources in the last decade resulted in significantly more intensive research efforts using analytical methods

T
and more sophisticated models to enable more accurate assessments of structural components and SSI effects.
The approaches found in the literature can be categorised in: (a) models with emphasis on soil-foundation

IP
elements and simplified structural components (e.g. Argyroudis et al. 2013), (b) models with emphasis on
structural details and simplified soil-foundation elements or methods of analysis (e.g. Nielson and DesRoches

CR
2007; Aygun et al. 2011), (c) computationally complex models with detailed structure and soil-foundation
elements and methods of analysis (e.g. Kwon and Elnashai 2010), reflecting an escalation of accuracy on the
basis of high fidelity computational models.

US
Bridge models including SSI effects commonly simulate the deck using linear-elastic elements, while for the
piers inelastic beam-column elements are employed. Abutments are modelled with spring elements (linear,
multi-linear, non-linear) that simulate the response of the abutment and backfill soil. For seat type abutments,
AN
gap elements are added to the model to represent the opening and closure in expansion joints (Mitoulis 2012).
The foundation is usually modelled using bilinear (p-y) springs along the length of the piles, representing the
non-linear force-deformation relationship of the foundation and soil by the inertial and kinematic SSI effects
(Kappos et al. 2012). In more simplified SSI models, linear springs or fixity conditions are adopted for the
M

foundation (e.g. Stefanidou et al. 2017). A full SSI model includes kinematic and inertia interaction, i.e.
masses, non-linear stiffness and damping. The modelling may incorporate either two-dimensional (2D), three-
dimensional (3D) or combined approaches with simplifications of the structural or soil-foundation elements.
ED

For example, a 3D bridge superstructure, 2D soil domain and one-dimensional (1D) lateral coupling p–y, t–z,
and q–z springs was developed by Aygun et al. (2011) and Nielson and DesRoches (2007) in OpenSees to
produce fragility functions for bridges. The bridge deck and piers are modelled using linear elastic elements in
PT

a lumped model comprising a 3D ‘spine’ model of the bridge, which is adequate for the needs of parametric
vulnerability analysis. Other researchers employ SDOF models when the bridge can be adequately
approximated by an equivalent column and a lumped mass at its top, a simplification that is allowable by the
CE

codes in the longitudinal direction of regular bridges (e.g. Anastasopoulos et al. 2015).
The response of the soil-structure coupled system is commonly analysed by 2D FEM for the seismic fragility
analysis of tunnels (Argyroudis et al. 2017, Argyroudis and Pitilakis 2012), embankments and cuts
AC

(Argyroudis and Kaynia 2015) or retaining walls (Argyroudis et al. 2013) using a variety of software
platforms, such as ABAQUS, PLAXIS or FLAC. The soil behaviour is typically modelled with the Mohr-
Coulomb criterion to account for soil non-linearities. Interface elements are employed to model the interface
between the structure, e.g. tunnel, abutment wall, footings, and the soil. In addition, the analyses are usually
conducted assuming total stresses and undrained soil conditions, which are most representative during rapid
earthquake loading. To account for the accumulation of excess pore water pressures, the definition of case-
dependent parameters would be required, such as the water table level or the degree of saturation, leading to
increased uncertainty of the SSI simulation. Yet, the effect of the soil saturation is expected to affect the
response of the structure and should be considered when studying the effects of flood and scour on the
fragility of transport infrastructure (Argyroudis et al. 2018a).

18
 ACCEPTED MANUSCRIPT

5.2 Treatment of uncertainties

Uncertainties in the analytical fragility modelling are related to:
 the structure and soil parameters, i.e. uncertainty in geometric properties, mechanical and structural
parameters, structural modelling, which represent the variability in the capacity of the soil-structure
system,
 the hazard parameters, e.g. selection of intensity measure, uncertainty in hazard actions, such as the
seismic shaking or the scour depth, which represent the variability in the demand, and,
 definition of thresholds used as damage or limit states (Rossetto et al. 2014; Tsionis and Fardis 2014).
To account for uncertainties in the capacity (βC), sampling approaches, such as the Latin Hypercube or Monte

T
Carlo techniques, are commonly used to generate random combinations of key uncertain and geotechnical
parameters (e.g. Guo et al. 2016; Huh et al. 2017; Karamlou and Bocchini 2017; Stefanidou et al. 2017).

IP
These parameters include concrete compressive strength, yield strength of reinforcing bars, unit weight of soil,
friction angle of soil, among others. For example, based on a sensitivity study for a bridge under the combined

CR
effect of earthquake and flood-induced scour, Yilmaz et al. (2017) indicated that the most significant
parameters for the performance of the bridge are the compressive strength of concrete, the yield strength of
the reinforcing steel, the mass of the bridge, the abutment stiffness and the friction angle of the subsurface

US
soil. Uncertainties in soil unit weight, friction coefficient of sliding bearings and shear modulus of elastomer
of bridge isolators are found to have an insignificant impact on the seismic performance of the structure.
Depending on the hazard, certain design parameters might become dominant. For instance, Padgett et al.
(2013) concluded that the effect of soil properties is important for evaluating bridge damage due to
AN
liquefaction. Means for decreasing this type of epistemic uncertainties may be sought in structural health
monitoring (SHM) techniques, which may assist in identifying the key properties of the assets, such as the
modal and structural parameters of bridges during their lifetime, aiming at updating fragility functions in an
M

effort to represent more realistically the vulnerability of the degraded assets (e.g. Torbol et al. 2013). Based on
the study by Stefanidou and Kappos (2017), the uncertainty in capacity for bridge piers under ground shaking
varies between 0.14 and 0.50, depending on the limit state and pier type, with an average value equal to 0.35.
ED

In case of other assets such as tunnels or embankments, βC is commonly assigned based on engineering
judgment with a representative value being equal to 0.3 (Argyroudis and Kaynia 2015; Qiu et al. 2018). To the
best of the authors’ knowledge, the uncertainty in the capacity of geotechnical components such as soil
PT

embankments has not been investigated before.

The uncertainty in the demand (βD) in case of earthquake hazard, is taken into account by using a suite of
ground motions either from real or artificial seismic records scaled to different intensity levels (e.g.
CE

Argyroudis et al. 2013; Karamlou and Bocchini 2017). For flood hazards, probabilistic approaches are
employed to account for the uncertainty in the hydraulic characteristics, e.g. flood discharge, flow intensity,
shape of the flood hydrograph etc, that are associated with the scour depth or water pressure estimation (e.g.
AC

Kim et al. 2017; Tubaldi et al. 2017; Yilmaz et al. 2017). Vulnerability assessment of infrastructure, vehicles,
and people to landslide hazards involve large uncertainties and complexities; therefore, most of the
approaches are based on empirical data and expert judgment (Corominas et al. 2014; Wong and Winter 2018;
Winter 2018). The uncertainty in demand is commonly estimated based on the variability in the response of
the structure (simulated EDPs) due to the variability of the hazard characteristics, as the lognormal standard
deviation about the estimated median in the analysis results (Baker and Cornell 2006; Porter 2015). Therefore,
it is dependent on the properties of the structural components as well as the hazard intensity and its
characteristics, such as the selection of seismic ground motions. Stefanidou and Kappos (2017) have
calculated βD values between 0.38 and 0.71 for different bridge components and seismic intensities, while in
HAZUS methodology (NIBS 2004), an uncertainty factor for seismic demand equal to 0.5 is suggested.

19
 ACCEPTED MANUSCRIPT

The uncertainty in limit states and damage thresholds (βLS) is frequently neglected or is considered directly in
the lognormal standard deviation of the fragility function, which is commonly represented by a lognormal
cumulative distribution (Argyroudis and Kaynia 2014). In some cases, a Monte Carlo simulation is performed
to sample damage thresholds from a uniform distribution in their confidence intervals (Selva et al. 2013).
Uncertainties in the definition of limit states for bridge components are discussed by Stefanidou and Kappos
(2017), suggesting a uniform value equal to 0.35 for piers, 0.20 for bearings and 0.47 for abutments. A value
that is commonly used is 0.4 as per HAZUS (NIBS 2004) recommendations for buildings seismic fragility.
The total uncertainty (βtot) is usually introduced in the fragility functions as the summation of the lognormal
variances deriving from each component of uncertainty assuming that they are probabilistically independent
(Eq. 2).

T
βtot = √𝛽𝐶 2 + 𝛽𝐷 2 + 𝛽𝐿𝑆 2 Eq. 2

IP
It is realised that different sources of uncertainty are associated with the fragility analysis and achieving an
adequate level of modelling fidelity and treatment of uncertainty is a challenge (Silva et al. 2019). The

CR
propagation of the various uncertainties and the effect of modelling parameter variation, e.g. material or
geometric uncertainty, on the fragility estimates require the assessment of the significance of the modelling
parameters on the response of the components within an asset through sensitivity analysis (Padgett and

US
DesRoches 2007). This, on one hand, will facilitate defining the significant parameters and produce more
reliable fragility functions, and, on the other hand, will reduce the computational cost for statistical sampling
and additional simulations that have insignificant effects on the fragility assessment. The treatment of
AN
uncertainties is also related to the scope of the study, for example, when the aim is the fragility analysis of a
class of assets, the variation of parameters is larger as opposed to the fragility analysis of a single asset.
Notwithstanding this, there is a gap in understanding the significance of a number of parameters in different
transport assets exposed to diverse hazards, such as the properties of the soil and structures and the definition
M

of hazard actions.
5.3 Deterioration effects
ED

Numerical models for fragility assessment are usually created to cater for the design needs of the assets, which
may neither be accurate nor representative of the current condition of the structure. The time-dependent
deterioration effects, which are usually not taken into account, can considerably increase the vulnerability of
PT

the assets. Substantial research efforts have been performed on the mechanisms and modelling of the
deterioration of structural elements and earthworks. The degradation of structural strength may be attributed to
multiple factors, such as corrosion, erosion, other forms of chemical deterioration and fatigue (Melchers and
CE

Frangopol, 2008; Andisheh et al. 2016). In particular, the corrosion of steel due to the ingress of chlorides is
crucial, and more recent research efforts have focused on the effect of corrosion of reinforcing bars and steel
bearings on the seismic fragility (Ghosh and Padgett 2011; Alipour et al. 2011) or reliability (Frangopol et al.
AC

1997) assessment of bridges. The time-dependent deterioration effects on the fragility of other transport assets
are limited and further research is required. For example, Argyroudis et al. (2017) showed how the seismic
fragility of shallow tunnels is altered when ageing effects due to corrosion are considered following available
approaches for over ground structures. The corrosion is commonly modelled by the reduction in the cross-
section area of reinforcement as a function of time and the characteristics of the chlorides (e.g. CEBFIB-Task
Group 5.6, 2006; Andisheh et al. 2016). In some studies, the reinforcement reduction is estimated using a
Monte-Carlo simulation to account for the uncertainty in the factors that affect corrosion (Melchers and
Frangopol, 2008; Kallias et al. 2017). It is clear that more reliable deterioration models are needed for the
analysis and fragility assessment of corroded RC structures under- and over-ground, including large-scale
experimental tests. There is also a need for time-dependent fragility models of deteriorated transport assets
subject to hazards other than earthquakes. This includes the change in soil properties, e.g. due to the presence

20
 ACCEPTED MANUSCRIPT

of water, and boundary conditions, e.g. due to scour or erosion, as well as the accumulation of damage, e.g.
due to evolving ground movements.
5.4 Modelling of multiple hazards and cascading effects
Hazard interactions and cascading effects can be classified differently, while modelling of multiple hazards is
a relatively new endeavour (Ayyub 2014; Gill and Malamud 2014; Zaghi et al. 2016; Liu et al. 2016; Bruneau
et al. 2017). The available fragility models that account for hazard interactions at the vulnerability level are
limited and mainly focused on bridges. In this section, we provide selected examples to highlight some
modelling issues for:
 Uncorrelated hazards of different nature, including for example floods caused by different weather

T
phenomena, flood preceding an earthquake or the opposite. The time between the occurrence of the
two hazards, their sequence and their intensities can vary considerably.

IP
 Correlated or cascading hazards, where the secondary hazard is triggered by the primary hazard,
including for example, liquefaction, landslide and tsunami triggered by earthquakes, or flood,

CR
landslides, extreme wind and debris flow triggered by a hurricane. In this case, the two hazards are
concurrent or successive within a short period of time.
 Correlated or uncorrelated hazards of the same nature that may have cumulative effects on the

US
structure, e.g. main-shock and aftershocks, or minor hazard effects occurring before a major stressor
over a short or longer period of time. For example, scour holes might be forming at bridge
foundations throughout the life of the bridge, of minor or moderate extent, and then followed by an
extensive flood that causes extensive scouring, debris accumulation and hydraulic forces on the
AN
structure.
The combined effect of uncorrelated hazards such as earthquake and flood-induced scour on the performance
of bridges has been researched by Prasad and Banerjee 2013; Banerjee and Prasad 2013; Dong et al. 2013;
M

Guo et al. 2016; Yilmaz et al. 2016. Scour of bridge foundations is a major cause for failure, as deepening of
scour around piers and/or abutments during the lifetime of a bridge can coincide with other hazards, such as
seismic excitations. Most of the studies consider identical scour depths at all bridge piers associated with
ED

specific flood events usually with a return period up to 100-years (e.g. Banerjee and Prasad 2013, Guo et al.
2016) or analysing a range of scour depths, which leads to a large computation effort. The potential flood
hazard at bridge sites is commonly not evaluated; however, more recently Yilmaz et al. (2016) assumed a
PT

variation of scour depth across multiple piers based on streamflow statistics and regional regression equations.
A deterministic scour depth is commonly adopted, while in some cases the uncertainty of scour hazard and its
time dependency is considered (Guo et al. 2016). The combined effect of earthquake and flood hazards is
CE

represented through fragility surfaces (Yilmaz et al. 2016; Guo et al. 2016), providing the failure probability
of the bridge as a function of the corresponding intensity measures, commonly the PGA for earthquake and
scour depth or flow discharge for flood. The surfaces are derived based on the fragility functions of each
AC

individual hazard, considering the intensity measures as statistically independent random variables. Gehl and
D’Ayala (2016) developed fragility surfaces for concrete bridges as a function of PGA and flow discharge
based on system reliability methods and Bayesian networks. As expected, scouring increases the probability
of damage, however, in some cases it was found there was no further change after a certain scour depth
(Prasad and Banerjee 2013). Similar approaches may be applicable for other combinations of hazards, e.g.
permanent ground movements preceding dynamic loading, such as earthquakes.
With respect to the modelling of sequences of uncorrelated hazard effects a reasonable approach would be to
consider the consequences of the first hazard effects on the structure and subsequently the second hazard
effect acting upon the modified and potentially more vulnerable system. In the absence of validated models
simulating a sequence of hazards, simplified approaches may be employed. For example, to account for the

21
 ACCEPTED MANUSCRIPT

sequence of flood-induced scour and subsequent earthquake on bridges, the springs that model the resistance
of the soil are removed around the piles or shallow foundation down to the scour depth. This is a common
approach followed by Dong et al. (2013), Prasad and Banerjee (2013), Banerjee and Prasad (2013), and Guo
et al. (2016). However, in this way, the effect of scour geometry and the modification of the soil properties
due to saturation and scour is not considered. Tanasic et al. (2013) and Tanasic and Hajdin (2017) estimated
the bridge damage probability considering the degradation of soil parameters over time due to scouring using
a simplified approach. Bridge failure is the result of either geotechnical or structural failure mechanisms, with
the first being the excessive settlement and the second being the failure of the deck (ultimate capacity). At the
present time, a comprehensive numerical model that accounts for the scour size and its effect to the soil
properties, as well as other hydraulic actions due to flood such as hydraulic forces, and debris accumulation is
yet to be reported in the literature.

T
An example of cascading hazards is the case of seismic excitations and liquefaction effects. Bridge fragility

IP
for this case has been studied by Aygun et al. (2011) using dynamic nonlinear p-y elements to model pile-soil
interaction along with constitutive models available in OpenSees for liquefiable soils. Also, Brandenberg et al.

CR
(2011) developed numerical fragility functions for bridges in liquefied soil as a function of free-field lateral
ground displacement by applying a reduction factor in p-y capacity associated with liquefaction.
The cumulative structural damage of transport infrastructure due to cascading hazards of the same nature,

US
such as mainshock-aftershock sequences can be significant. The cascading effects on the seismic fragility of
bridges have been studied by Franchin and Pinto (2009), Alessandri et al. (2013), Dong and Frangopol (2015),
Ghosh et al. (2015), Kumar and Gardoni (2014). The structural model is commonly subjected to mainshock-
AN
aftershock sequences, a challenging issue is the selection of aftershock ground motions that are consistent
with the mainshock. One approach is to adopt the same set of records used to represent the mainshock for the
aftershocks (Franchin and Pinto 2009; Alessandri et al. 2013). Another approach is to develop probabilistic
models to predict the effects of past earthquakes on the structural properties and to assess the effects of
M

degradation on the seismic vulnerability (Kumar and Gardoni 2014). There is a number of other hazard effects
of the same nature that cause cumulative effects and thus increase the vulnerability of the structure, which has
not yet been researched, such as the fragility of SoA for cumulative flood-induced scour or evolving ground
ED

movements.

6 A new methodology for vulnerability assessment of transport infrastructure to

PT

multiple hazards
6.1 Transport Infrastructure System of Assets (SoA) in diverse ecosystems
CE

Based on the literature review conducted it was realised that the available vulnerability and risk assessment
frameworks typically consider individual assets of the transport infrastructure, exposed to one hazard, and
they are static in the sense that they neglect changes of the asset performance during its life. Additionally, in
most cases, the available models are simplified, and they focus on bridges. Moreover, they usually ignore the
AC

geomorphological and topographical conditions of the surrounding environment as well as the classification of
the assets in terms of road capacity or speed limits. Nevertheless, infrastructure comprises Systems of Assets
(SoA), i.e. a combination of interdependent assets exposed to multiple hazards, depending on the environment
within which these reside, whilst their performance changes due to deterioration or improvements that take
place during their life. In addition, the SoA performance depends on the classification and typology
characteristics of the infrastructure.
Herein, the newly introduced concept of the transport infrastructure SoA in ecosystems refers to inter-urban
roads and illustrate the different elements that comprise the system and the geotechnical and climatic hazards
to which the system is subjected. In this respect, the infrastructure is classified based on:

22
 ACCEPTED MANUSCRIPT

(i) the road capacity and speed limits, i.e. high capacity and speed roads, such as interstate highways,
motorways and dual-carriageways, and lower capacity and speed roads, such as single carriageways, and,
(ii) the geomorphological and topographical conditions, i.e. mountainous or lowland areas.
This classification covers the majority of the existing inter-urban road networks, exposed to potential hazards,
such as earthquakes, floods, landslides including slides, debris flow and rock fall, extreme temperatures and
shrink/swell phenomena. Figure 2 and Figure 3 display sketches of this concept. The transport infrastructure
ecosystem approach provides the basis for realising the need for an integrated assessment of the fragility of
SoA, as opposed to the examination of the individual assets independently.
The landforms, geomorphological processes, and surface geology are different in mountainous and lowland
areas leading to different hazard actions. Stiff soil and rock formations are more common in mountainous

T
areas, while softer alluvial deposits and sediments are predominantly met in lowland areas and valleys.

IP
Earthquake or rainfall triggered landslides (slides, rockfalls, debris flows) are common in hilly and
mountainous areas. Also, the dynamics of riverine flooding vary with terrain. Floods may manifest within
minutes after a heavy rain with fast-flowing of water due to steeper slopes leading to erosion, washout of

CR
roads and scour of foundations (Figure 2a, Figure 3a). Lowland areas may stay covered with shallow, slow-
moving floodwater for days or even weeks, e.g. overbank flooding. As a result, the floodplain is wider and the
amount of water is greater, causing scour of foundations, softening by soil saturation and so on (Figure 2b,
Figure 3b).
US
Moreover, the typology of transport infrastructure varies due to geomorphological conditions, for example,
rock tunnels are common in mountainous areas (Figure 2a) and cut & cover tunnels in lowland or urban areas
AN
(Figure 3a). Foundations of bridges are shallow in rock/stiff ground conditions and deep, i.e. pile supported, in
soft soils. Cuttings and embankments are usually of greater height in steeper geomorphological settings
compared to those in flatter terrains. The classification of roads affects also the typology and geometry of the
M

infrastructure (see Table 1). Motorways for high-speed traffic require grade-separated interchanges (Figure 2a,
Figure 2b), while lower speed single carriageways typically have at-grade junctions without median strip to
separate opposing flows (Figure 3a, Figure 3b).
ED
PT
CE
AC

23
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
(a)
M
ED
PT
CE
AC

(b)
Figure 2. Transport infrastructure in diverse ecosystems exposed to multiple hazards: High capacity and speed roads (e.g.
motorways) in (a) mountainous areas, (b) lowland areas

24
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
(a)
M
ED
PT
CE
AC

(b)
Figure 3. Transport infrastructure in diverse ecosystems exposed to multiple hazards: Lower capacity and speed roads (e.g.
single carriageways) in (a) mountainous areas, (b) lowland areas

25
 ACCEPTED MANUSCRIPT

To highlight the complexity of transportation SoA and the effects due to diverse hazards Figure 4 illustrates
two typical sections that can be encountered in either of the ecosystems illustrated in Figures 2 and 3. The first
is a transverse and the second one is a longitudinal section.

Debris flow (rainfall

or earthquake induced)

Slope
Small cutting or/and

T
Translational retaining wall Road or rail bed
slide Heave
Rotational/Slump

IP
failure
Embankment
Retaining

CR
structure
Settlement ground shaking or
liquefaction induced

Earthquake
US Scouring
Sea, lake or river
(deep and/or fast water)
AN
(a)
Multi span bridge Road pavement
(continuous or simply supported) or rail bed Heave Approach Embankment
slab Approach fill or natural ground
M

Bearings (for not integral bridge) Settlement* Settlement and/or

Pier
local sliding*
ED

Abutment
River
(deep and/or fast water) Foundation
PT

Pile cap Scouring

Scouring
*ground shaking or
Settlement* liquefaction induced
Piles
CE

Earthquake

(b)
AC

Figure 4. Multiple hazard effects on representative transport System of Assets (SoA): (a) embankment, slope, retaining
structure, (b) bridge, abutment, foundations, backfill.

The SoA illustrated along with common hazard effects in Figure 4a includes slopes responding together and
interacting with a road pavement or railway tracks on embankments and supported by retaining structures,
exposed to landslides, potentially triggered by precipitation or earthquakes (ground shaking or/and
liquefaction), flooding effects or/and ground shaking. Rotational or slump failure of embankments may occur
due to the same hazards. Degradation of the SoA, in this case, may be the result of embankment erosion or
foundation scour over flooded sea, lakes or rivers and potential residual dislocations of the retaining
structures. The stability of the SoA may deteriorate during its lifetime as a result of an increase in the stresses
or traffic loads, decrease of soil shear strength due to changes in pore water pressure and presence of organic

26
 ACCEPTED MANUSCRIPT

materials. Potential improvement measures include shotcreting, soil anchors, nailing, vegetation, and
improved drainage (FHWA 2014; FHWA 2015).
With reference to Figure 4b, the multiple hazard scenarios may include settlements, heave or/and local sliding
of the embankment and approach fill due to ground shaking or liquefaction among other hazards. Bridge
components such as the deck, abutment, piers and foundations, may suffer damage due to seismic shaking,
settlements, scouring and liquefaction. Degradation, in this case, may occur due to corrosion of the reinforced
or prestressed concrete elements, scouring of the foundation soil and residual dislocations of the abutments.
Similarly, degradations of the approach fill can be due to traffic loads and residual deflection of the backfill,
such as settlement or heave. Improvements include strengthening of the piers and/or the abutments,
improvement of the compacted state of the backfill or some means of reinforcement (Power et al. 2004;

T
Buckle et al. 2006).

IP
6.2 Methodology for the development of numerical fragility functions for transport SoA
exposed to multiple hazards

CR
The proposed methodology is described in the following six steps and illustrated in a flowchart (Figure 5).
This approach is practically applicable for evaluating the physical damage, i.e. structural or geotechnical, and
not for the loss of functionality, as discussed in section 4.7.

US
(i) Definition of the basic configurations of the SoA, including geometry and material of the assets and the
components and properties of the soil. The properties are strongly dependent on the local geomorphology and
typology of structures and can be selected on the basis of representative assets (see Table 2) considering their
variation. A common approach is to consider typical soil profiles on the basis of common engineering practice
AN
and code specific classification, for example the Uniform Building Code (UBC) or Eurocode 8 (EC8) use the
shear wave velocity (VS,30) to classify the soil types (Argyroudis et al. 2013, Argyroudis and Kaynia 2015).
Depending on the hazard, the initial soil properties may be altered (for example the strength characteristics
M

can be reduced due to saturation, Argyroudis et al. 2018a), while the shear modulus and viscous damping
could change in accordance with the increase in shear strain levels during seismic excitation (Argyroudis et al.
2013, 2017). A sampling technique may be applied by considering the main soil and asset material and
ED

geometric properties as random variables to generate a series of SoA samples. The uncertainty in capacity, βC,
can be quantified based on the distribution of the above variables or on the basis of an expert judgement
approach.
PT

(ii) Selection of engineering demand parameters (EDPs) for each asset or component and relevant limit
states and thresholds for the definition of damage states (see section 3). With reference to Figure 4b, the EDPs
for bridge components can include the curvature of the piers, the gap between the deck and the approach slab
CE

at the abutments-width of the expansion joint, displacement of the bearings and maximum moment on the
deck, while the EDP for the backfill can be described by the permanent ground displacement. Relevant limit
states and thresholds for corresponding damage states (e.g. minor, moderate, extensive, complete) are given in
AC

the literature or can be defined by sectional analysis based on which the capacity of the deck or pier can be
defined. The uncertainty in limit states, βLS, is usually estimated on the basis of expert judgment.
(iii) Definition of hazard actions and intensity measures (see also section 4), which depends on the type of
assets and scope of the analysis, including the envisaged accuracy and the number of assets that are under
examination. For example, for seismic hazard action and when a time history or incremental dynamic analysis
is chosen to be performed, a suite of strong ground motions should be selected for different intensity levels.
The latter can either cover a range of possible intensities, e.g. PGA from 0.1 to 1.0g at bedrock, or can be
correlated to annual exceedance probabilities of seismic events, having various intensity levels through
regional seismic hazard curves. A common approach for the selection of earthquake records is spectral
matching, using the target spectrum provided by the codes (Argyroudis and Kaynia 2013; Katsanos and

27
 ACCEPTED MANUSCRIPT

Sextos 2013). Α suite of 7 to 10 motions is usually selected from available databases of earthquake records to
cover different frequency contents, duration or seismotectonic conditions (Iervolino et al. 2011). The approach
differs when a quasi-static analysis is adopted as for example for underground structures. In this case, the
induced seismic ground deformations are applied to the soil-structure model (Argyroudis and Pitilakis 2012;
Huh et al. 2017).
For other hazards such as floods, the related actions include scour, debris accumulation and hydraulic forces
and these can be calculated based on simplified approaches given by closed-formed solutions and guidelines
(e.g. FHWA 2012a; BD97/12, 2012) or well-informed hydraulic models. Again, the IMs, such as the peak
water discharge, are correlated with the annual exceedance probabilities of particular flood events through
regional flood curves or can cover a range of possible intensity levels (Yilmaz et al. 2016). For structures of

T
great length, such as bridges, the actions by the hazard can be taken as identical along the structure, e.g. at all
bridge piers, or temporal and/or spatial variability may be considered on the basis of local effects, e.g. Sextos

IP
and Kappos (2008) for earthquake loads, and Yilmaz et al. (2016) for scour effects on piers. Combination of
hazards may include a set of subsequent natural actions that are more or less obvious, as for example the

CR
sequence of a flood followed by an earthquake, or earthquake excitation and subsequent flood due to tsunami;
ground movement and earthquake or the opposite, and finally, two subsequent hazard events of the same
nature, for instance main earthquake and aftershock or two floods in a short time frame. The selection and

US
combinations of hazards and their intensity should be decided by the engineer in consultation with experts in
other relevant fields as appropriate and in agreement with the stakeholder or owner upon temporal and spatial
characteristics and local effects (Marzocchi et al. 2009; Liu et al. 2016).
AN
(iv) 2D or 3D numerical models are employed to analyse the response of the SoA defined in step (i)
subjected to different hazards or combination of hazard actions of a given sequence defined in step (iii).
Potential key challenges may include the simulation of soil-structure interaction that can be described by
elastic or inelastic models considering kinematic and inertial interaction, through equivalent springs or as a
M

continuum layered model, the type of analysis, e.g. non-linear static, incremental dynamic, time history, as
well as the definition of boundary conditions and interfaces between the structure components and
surrounding soil (see also section 5 for modelling issues). The results of the numerical analyses provide the
ED

required EDP for each component or/and asset for the fragility analysis described in the following steps.
(v) Evolution of damage and uncertainty in demand (βD). The results of the analyses conducted in step (iv)
in terms of EDPs are plotted versus the selected IM (e.g. PGA or peak flow discharge) for each asset or
PT

component representing the evolution of damage with increasing hazard intensity, usually on a logarithmic
scale. A regression model that describes the correlation between the IM and EDP is then used. The uncertainty
in demand, βD, is calculated based on the dispersion of the logarithms of IM-EDP simulated data with respect
CE

to the regression fit.

AC

28
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED

Figure 5. Flowchart for multiple hazard fragility functions of transport systems of assets
PT

(vi) Generation of component, asset and SoA fragility curves and surfaces for single and multiple
hazards correspondingly. Each fragility function requires the definition of two parameters (see Equation 1,
section 4): IMmi, that is the median threshold value of IM required to cause the i th limit state) and βtot, which is
CE

the total lognormal standard deviation. The total uncertainty is calculated at asset level assuming that the
uncertainties in demand (βD) as calculated in step (v), capacity (βC) as per step (i) and definition of limit states
(βLS) as per step (ii), are statistically independent (see Equation 2, section 5.2). The median value of IM mi is
AC

obtained using the regression model defined in step (v) and the definitions of damage states for each
component/asset defined in step (ii) (Cornell et al. 2002). Another approach for the estimation of the fragility
parameters based on the IM-EDP pairs is the maximum likelihood method (Shinozuka et al. 2001; Selva et al.
2013). The fragility of the asset (e.g. bridge) or SoA, e.g. bridge, foundations, backfill, can be calculated
based on the fragilities of the components or assets respectively, assuming a series connection (Padgett and
DesRoches 2009; Stefanidou and Kappos 2017). The combined effect of two hazards can be visualised
through fragility surfaces, where the intensity measures are plotted along the two horizontal axes and the
damage probability is indicated by the surface. In this case, the IMs, are considered as statistically
independent random variables and their joint cumulative probability distribution provides the failure
probability under the multi-hazard scenario (Yilmaz et al. 2016). It is noted that it is essential to indicate the
order of hazards as the fragility of the SoA is strongly dependent on the sequence of events. These fragility

29
 ACCEPTED MANUSCRIPT

models will enable risk and resilience-based assessments and designs of SoA as well as their reliability and
safety characterisation, to facilitate consultants and owners, and to enhance decision-making and risk
management delivered by stakeholders.

6.3 Case study

The methodology described in the previous section is briefly discussed herein for a representative SoA
exposed to combined effects of foundation scouring due to flooding followed by seismic excitation. This
example aims at illustrating succinctly the subsequent steps of the proposed methodology. Yet, the application
of the methodology will be included in a future publication, due to the length limitations of this paper.
Step (i) Definition of the SoA and its properties: A three-span pre-stressed integral bridge with its

T
components, i.e. deck, abutment, piers and foundations, together with the backfill and the foundation soil
(ground type B, EN 1998-3, 2005) is defined. The properties of the saturated soil layers are modified due to

IP
flooding conditions (Argyroudis et al. 2018a). The stiffness and damping soil properties are dependent to the
shear strain level during the earthquake (Argyroudis and Kaynia 2014).

CR
Step (ii) Selection of EDPs and definition of damage states: the maximum bending moment (M max) for
critical sections of the deck, pier and abutment and the maximum permanent ground deformation (U y) of the
backfill behind the abutment. Damage states are defined based on the exceedance of the yielding bending

US
moment of the bridge components, and the variation of Uy for the backfill.
Step (iii) Definition of hazard actions: A progressing scour depth at the right abutment is analysed
AN
corresponding to 1.0Df, 1.5Df and 2.0Df, where Df=2.0 m is the foundation depth. Five real acceleration time
histories from earthquakes recorded on rock or very stiff soil were selected as outcrop motion for the analyses
scaled to PGA =0.2, 0.4 and 0.6g. The seismic excitations are applied separately for each scour depth in order
to simulate the combination of the two hazards.
M

Step (iv) Numerical model: A 2D FEM was developed in PLAXIS ver.2017 (Argyroudis et al. 2018b). All
analyses included initial stages simulating both the initial geostatic stresses and the construction of the bridge.
An elasto-plastic soil behaviour was assumed (i.e. Mohr-Coulomb criterion), while the bridge components
ED

followed a linear-elastic behaviour. The scouring effect was modelled by gradually removing soil elements
around and under the foundation reaching the maximum scour depth, while the seismic input was uniformly
applied at the basis of the model.
PT

Step (v) Evolution of damage: For each component of the SoA and each scour scenario, the EDPs are plotted
versus the PGA in a logarithmic scale and a regression curve is fitted.
CE

Step (vi) Multiple hazard fragility functions: The fragility parameters are defined and the fragility
curves/surfaces for each component are generated. The median PGA is obtained for each damage state using
the regression models and the definitions of damage states (step ii). The total variability (β tot) includes three
AC

sources of uncertainty. The one associated with the definition of damage states (β ds) was taken 0.4, while the
uncertainty due to the capacity (βC) was taken 0.3. The third uncertainty is associated with the seismic demand
and was calculated by the dispersion in response due to the variability of the seismic input motion. Examples
of component fragility functions are shown in Figures 6a, 6b, 6c, 6d for the scenario of scour depth equal to
2Df. It is seen that the vulnerability of the components may be very different for given scour conditions and
seismic excitations. The fragility of the SoA is then extracted assuming a series connection between
components and defining an upper and lower bound (Figure 6e). The estimated system fragility can be
considerably increased for different scour conditions, for example for a PGA of 0.4g the probability of
exceeding moderate and complete damage (lower bound) is increased from 0.51 to 0.97 and from 0.05 to 0.51,
when the scour depth increases from 1.5Df to 2Df, indicating the importance of the deterioration of the system
prior to the seismic action, signifying the importance of the quantification of risk for multiple hazards. The

30
 ACCEPTED MANUSCRIPT

combined effect of the two hazards can also be visualised through fragility surfaces as it is shown in Figure 6f
for the case of the bridge pier, i.e. component level, as well as for the entire asset, i.e. bridge, and potentially
the entire network, e.g. part of the road network.

T
IP
(a) (b)

CR
US
AN
(c) (d)
M
ED
PT
CE

(e) (f)
Figure 6. Example of multiple hazard fragility functions for a representative System of Assets (SoA) exposed to combined
scour and seismic effects: (a) deck, (b) pier, (c) abutment, (d) backfill, (e) fragility surface for minor damage of the bridge pier,
(f) SoA, for scour equal to two foundation depths (2Df) and earthquake excitations.
AC

7 Summary and recommendations for future developments

This paper provides a comprehensive state-of-the-art review of natural, geotechnical and weather hazards on
transport infrastructure, including the main failure modes, EDPs, and typologies for roads, bridges, tunnels,
embankments, retaining walls and backfills. This information is relevant to the vulnerability assessment of the
aforementioned assets when subjected to multiple hazards, which is part of the quantitative risk analysis of
transport networks. A comprehensive review of the available fragility models for the above assets exposed to
multiple hazards was conducted and combinations of hazards, such as earthquakes, floods, liquefaction or
landslides among others were examined. Subsequently, the main modelling challenges for the generation of
analytical fragility functions, including soil-structure-interaction, deterioration and multiple hazard effects as
well as the treatment of uncertainties were discussed. It is recognised that transport assets exist in systems,

31
 ACCEPTED MANUSCRIPT

therefore the concept of System of Assets (SoA) in diverse ecosystems is introduced and classified
infrastructure based on (i) the road capacity and speed limits and (ii) the geomorphological and topographical
conditions. In this context, a methodological framework for the development of numerical fragility functions
of SoA under multiple hazards was also presented. The information provided in this paper can facilitate the
quantitative risk analysis of transport assets and the development of new analytical fragility models.
Furthermore, fragility functions are essential components in the reliability analysis of transport systems, as
they provide a probabilistic characterisation of system reliability over the full range of loads to which a system
might be exposed considering the associated uncertainties in structural capacity and demand.
The literature review with respect to the fragility of assets revealed that the majority of the studies focus on
either individual transport assets or entire networks, typically considering only one hazard at a time. These
studies shed light mainly on the vulnerability of bridges, and secondarily that of tunnels, and the main

T
emphasis had been placed on the ground shaking due to earthquake excitations. Very limited research has

IP
been conducted for other assets such as embankments, slopes, retaining walls and abutments. The existing
models for these assets are based on empirical data or expert judgment mainly for the earthquake hazard,

CR
while they cover limited typologies. Detailed numerical models are rarely used due to a large number of assets
and hazards, while simplifications and assumptions are considered along with categorisation of the assets in
groups on the basis of similarities in terms of their engineering characteristics, as a means to reduce the
computational time. Hence, future research should focus on the improvement of the available models for all

US
assets and for a broader range of typologies and hazards including landslides, earthquakes, flooding, sea level
rise and weather stresses. Emphasis should be given to the implementation of more advanced numerical
modelling to address SSI and other aspects, such as soil behaviour when subject to multiple hazards. Another
AN
challenge is the determination of the optimum IM for each asset and hazard, and the quantification of the
uncertainty in capacity and limit state definition, especially where variability in soil, structural material, and
local hazard potential, e.g. scour or liquefaction, are jointly present.
M

Furthermore, research on multiple hazards, such as scouring due to flooding following a strong earthquake or
cascading hazards, such as mainshock-aftershock sequences, is very limited and focused only on bridges.
Most of the approaches applied so far place emphasis on the structural details (e.g. of bridges) and adopt a
ED

simplified model for the foundations and soil behaviour. A common approach to account for the effects of
scour due to flood is the removal of spring elements around the foundation down to a depth equal to the scour
depth. In addition, the failure modes considered for scoured bridges are the same as the ones used in seismic
fragility assessment, while other hydraulic actions, such as the accumulation of debris and stream pressures
PT

are usually neglected. Nevertheless, a more advanced modelling approach is needed to account for the scour
geometry, the alteration of the flooded/scoured soil properties, the hydraulic actions as well as the potential
CE

damage modes for a broader range of bridge and other asset typologies. In some cases, reliability methods and
Bayesian networks have been applied to assemble the multiple hazard fragility of a system (e.g. bridge) based
on the fragilities of its components. This probabilistic approach allows treating hazard interactions of
component-specific fragility functions that have been derived based on various techniques. However, the
AC

treatment of uncertainties and the assumptions in the definition of failure modes are based on judgment and
not on numerical modelling, which can provide a more realistic articulation of the propagation of the damage
within the asset.
Moreover, deterioration or improvements that take place during the life of the assets have only been
considered in a limited number of fragility models, predominantly for bridges subject to earthquake
excitations. Hence, there is a need to investigate the effect of potential structural (e.g. due to corrosion) and
soil (e.g. due to saturation or settlement) degradation on fragility models. Also, the effect of mitigation
measures (e.g. drainage of structures, use of gabions, rockfall protection and snow barriers, erosion control
systems or geogrid soil reinforcement) on the fragility of the different transport assets exposed to diverse
hazards should be investigated.

32
 ACCEPTED MANUSCRIPT

The focus of this paper was on inter-urban transport infrastructure, while future research will encompass
fragility models for urban networks exposed to multiple hazards. In the latter case, interdependencies with the
built environment should be taken into consideration, e.g. interactions with energy and utility networks or
other transport systems. Additionally, validated functionality loss models, which quantify induced effects, i.e.
not losses and damages of structural nature, due to diverse hazards impeding the mobility are very scarce in
the literature, and this needs urgent attention for improving infrastructural resilience.
In summary, the available risk assessment frameworks typically consider individual assets of the transport
infrastructure, exposed to one hazard only and are not evolving with time, i.e. they neglect the temporal
variations and changes to, or deterioration of, the asset during its life that lead to the degradation of its
performance. Notwithstanding, assets exist within systems of assets (SoA), within diverse ecosystems,
exposed to multiple hazards, such as earthquakes, floods, landslides (including slides, debris flow, and rock

T
fall), extreme temperatures and shrink/swell phenomena. The proposed transport infrastructure ecosystem

IP
approach put forward by this paper forms the basis for an integrated assessment of the fragility of the SoA,
rather than the individual elements, from which it is formed. This approach has the potential to support well-

CR
informed, more accurate and comprehensive risk and resilience assessment of the transport network that will
contribute toward adaptation, mitigation and recovery planning for multiple hazards.

Acknowledgments

US
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).
AN

References
M

AASHTO (1993). AASHTO Guide for Design of Pavement Structures. Publication by the American Association of State
Highway and Transportation Officials, Washington DC.
Alessandri S, Giannini R, Paolacci F (2013). Aftershock risk assessment and the decision to open traffic on bridges.
ED

Earthquake Eng. Struct. Dyn. 42(15):2255–2275.

Alipour A, Shafei B, Shinozuka M (2011). Performance evaluation of deteriorating highway bridges located in high
seismic areas, Journal of Bridge Engineering 16(5):597-611.
Almufti I, Willford MR (2013). Resilience-based earthquake design (REDi) rating system, version 1.0. Arup.
PT

American Lifelines Alliance (ALA) (2001). Seismic fragility formulations for water systems, Part 1– Guideline, ASCE-
FEMA, Reston, VA2001, 104 p.
American Lifelines Alliance (ALA) (2005). Flood‐ resistant local road systems: a report based on case studies, American
CE

Lifelines Alliance.
Anastasopoulos I, Sakellariadis L, Agalianos A (2015). Seismic analysis of motorway bridges accounting for key
structural components and nonlinear soil-structure interaction. Soil Dynamics & Earthquake Engineering 78:127-141.
Andisheh K, Scott A, Palermo A (2016). Seismic behavior of corroded RC bridges: review and research gaps.
AC

International Journal of Corrosion, Vol. 2016, Article ID 3075184, http://dx.doi.org/10.1155/2016/3075184

Applied Technology Council (1985). Earthquake Damage Evaluation Data for California (ATC-13), Redwood City, CA.
Argyroudis S, Pitilakis K (2012). Seismic fragility curves of shallow tunnels in alluvial deposits. Soil Dyn. Earthq. Eng.
35:1-12.
Argyroudis S, Kaynia AM, Pitilakis K (2013). Development of fragility functions for geotechnical constructions:
application to cantilever retaining walls. Soil Dynamics and Earthq Eng 50:106-116.
Argyroudis S, Kaynia AM (2014). Fragility functions of highway and railway infrastructure. In: Pitilakis K, Crowley H,
Kaynia AM (eds) SYNER-G: Typology definition and fragility functions for physical elements at seismic risk. GGEE
27, Springer.
Argyroudis S, Kaynia AM (2015). Analytical seismic fragility functions for highway and railway embankments and cuts.
Earthquake Engineering and Structural Dynamics 44(11):1863–1879.

33
 ACCEPTED MANUSCRIPT

Argyroudis S, Selva J, Gehl P, Pitilakis K (2015). Systemic seismic risk assessment of road networks considering
interactions with the built environment. Computer-Aided Civil and Infrastructure Engineering 30(7): 524-540.
Argyroudis S, Palaiochorinou A, Mitoulis S, Pitilakis D (2016). Use of rubberised backfills for improving the seismic
response of integral abutment bridges. Bull Earthq. Eng. 14(12):3573-3590.
Argyroudis S, Tsinidis G, Gatti F, Pitilakis K (2017). Effects of SSI and lining corrosion on the seismic vulnerability of
shallow circular tunnels. S Dyn Earthq Eng 98:244-256.
Argyroudis S, Mitoulis S, Winter MG, Kaynia AM (2018a). Fragility of critical transportation infrastructure systems
subjected to geo-hazards. 16th European Conference on Earthquake Engineering, June 18-21, Thessaloniki, Greece.
Argyroudis S, Mitoulis S, Kaynia AM, Winter MG (2018b). Fragility assessment of transportation infrastructure systems
subjected to earthquakes. Proceedings GEESD V, Geotechnical Special Publication (GSP 292).
Avanaki MJ, Hoseini A, Vahdani S, de Santos C, de la Fuente A (2018). Seismic fragility curves for vulnerability
assessment of steel fiber reinforced concrete segmental tunnel linings. Tunnelling and Underground Space

T
Technology 78:259-274.

IP
Aygün B, Duenas-Osorio L, Padgett JE, DesRoches R (2011). Efficient longitudinal seismic fragility assessment of a
multi-span continuous steel bridge on liquefiable soils. ASCE J. Bridge Eng 16: 93–107.
Ayyub BM (2014). Systems resilience for multihazard environments: Definition, metrics, and valuation for decision

CR
making. Risk Analysis, 34(2), 340-355.
Baker JW, Cornell CA (2006). Vector-valued ground motion intensity measures for probabilistic seismic demand
analysis. University of California Press, Berkeley, CA.
Banerjee S, Shinozuka M (2008). Experimental verification of bridge seismic damage states quantified by calibrating

US
analytical models with empirical field data. Earthquake Engineering and Engineering Vibration 7(4):383-393.
Banerjee S, Prasad GG (2013). Seismic risk assessment of reinforced concrete bridges in flood-prone regions. Structure
and Infrastructure Engineering 9: 952–968.
AN
Basöz NI, Kiremidjian AS (1998). Evaluation of bridge damage data from the Loma Prieta and Northridge, California
earthquakes. Technical Rep. MCEER, Buffalo, NY.
BD97/12 (2012). Design manual for roads and bridges: Vol. 3: Highway structures: inspection and maintenance, Section
4: Assessment, Part 21: The assessment other hydraulic actions at highway structures. The Highways Agency, UK.
M

Billah AHM, Alam MS (2015). Seismic fragility assessment of highway bridges: a state-of-the-art review. Structure and
Infrastructure Engineering 11(6):804-832.
Bowles JE (1996). Foundation analysis and design. 5th ed. Civil Engineering Series. New York, NY: McGraw-Hill
ED

International Editions.
Bradley BA, Cubrinovski M, Dhakal RP, MacRae GA (2010). Probabilistic seismic performance and loss assessment of
a bridge–foundation–soil system. Soil Dynamics and Earthquake Engineering 30:395–411.
Brandenberg SJ, Kashighandi P, Zhang J, Huo Y, Zhao M (2011). Fragility functions for bridges in liquefaction-induced
PT

lateral spreads, Earthquake Spectra 27(3):683–717.

Briaud J-L, Maddah L (2016). Minimizing roadway embankment damage from flooding. A synthesis of highway
practice, NCHRP Synthesis 496, National Academy of Sciences, Washington D.C, ISBN 978-0-309-38973-0, doi:
CE

10.17226/23604.
Bruneau M, Barbato M, Padgett J, Zaghi AE, Mitrani-Reiser J, Li Y (2017). State of the art of multihazard design.
Journal of Structural Engineering 143(10):03117002.
Bruneau M, Chang SE, Eguchi RT, Lee GC, O’Rourke TD, Reinhorn AM, Shinozuka M, Tierney K, Wallace WA, von
AC

Winterfeldt D (2003). A framework to quantitatively assess and enhance the seismic resilience of communities.
Earthquake Spectra 19(4):733-752.
Buckle IG, Friedland I, Mander J, Martin G, Nutt, R, Power M (2006). Seismic Retrofitting Manual for Highway
Structures: Part 1-Bridges, Publication No. FHWA-HRT-06-032, Federal Highway Administration, US Department
of Transportation.
Cabinet Office (2011). Keeping the Country Running: Natural Hazards and Infrastructure. Available online at:
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/61342/natural-
hazards-infrastructure.pdf
CEB-FIB Task Group 5.6 (2006). Model for service life design. Fédération Internationale du Béton (fib).
CEREMA (2014) National climate change adaptation plan: transportation infrastructures and systems. Interim report,
available at http://www.infra-transports-materiaux.cerema.fr/action-3-report-in-english-a6010.html

34
 ACCEPTED MANUSCRIPT

Chan R, Schofer JL (2015). Measuring transportation system resilience: response of rail transit to weather disruptions.
Natural Hazards Review 17(1).
Choine MN, O’Connor A, Padgett JE (2013). A seismic reliability assessment of reinforced concrete integral bridges
subject to corrosion. Key Engineering Materials 569–570:366–373.
Cimellaro GP, Reinhorn AM, Bruneau M (2010). Framework for Analytical Quantification of Disaster Resilience. Eng.
Struct. 32:3639-3649.
CIRIA, CUR and CETMEF (2007). The Rock Manual. The use of rock in hydraulic engineering. 2nd edition. C683,
CIRIA, London.
Clarke J, O’Brien E (2016). A multi-hazard risk assessment methodology, stress test framework and decision support
tool for resilient critical infrastructure. Transportation Research Procedia 14:1355-1363
COE (2011). Climate change adaptation and disaster risk reduction in Europe: A review of risk governance. United
Nations Office for Disaster Risk Reduction - Regional Office for Europe, Council of Europe, European and

T
Mediterranean Major Hazards Agreement. 73p.

IP
Cook W, Barr PJ, Halling MW (2015). Bridge failure rate. Journal of Performance of Constructed Facilities 29(3).
Corigliano M (2007). Seismic response of deep tunnels in near-fault conditions. PhD dissertation. Politecnico di Torino,
Italy.

CR
Cornell CA, Jalayer F, Hamburger RO, Foutch DA (2002). Probabilistic basis for 2000 SAC Federal Emergency
Management Agency steel moment frame guidelines. J. Struct. Eng. 128(4):526-533.
Corominas J, van Westen C, Frattini P, Cascini L, Malet J-P, Fotopoulou S, Catani FM, Van Den Eeckhaut N, Mavrouli
O, Agliardi F, Pitilakis K, Winter MG, Pastor M, Ferlisi S, Tofani V, Hervas J, Smith JT (2014). Recommendations

County Council (2010). Cumbria US

for the quantitative analysis of landslide risk. Bull. Eng. Geol. Envir. 73:209-263.
Cumbria floods November 2009: an impact
http://www.cumbria.gov.uk/eLibrary/Content/Internet/536/671/4674/4026717419.pdf, last access: February 2019.
assessment.
AN
D’Ayala D, Gehl P, Martinovic K, Gavin K, Clarke J, Tucker M, Corbally R, Avdeeva YV, van Gelder P, Salceda MT,
Segarra MJ (2015). Fragility Functions Matrix. Deliverable D3.2. EU FP7 research project No 603960: INFRARISK,
https://cordis.europa.eu/project/rcn/110820_en.html
Dagá J, Chamorro A, de Solminihac H, Echaveguren T (2018). Development of fragility curves for road bridges exposed
M

to volcanic lahars, Nat. Hazards Earth Syst. Sci. 18:2111-2125.

Dawson RJ, Thomson D, Johns D, Wood R, Darch G, Chapman L, Hughes PN, Watson GVR, Paulson K, Bell S,
Gosling SN, Powrie W, Hall JW (2018). A systems framework for national assessment of climate risks to
ED

infrastructure. Phil. Trans. R. Soc. A 376: 20170298, http://dx.doi.org/10.1098/rsta.2017.0298.

Deng L, Wang W, Yu Y (2016). State-of-the-art review on the causes and mechanisms of bridge collapse. J. Perform.
Constr. Facil. 30(2): 04015005.
Doll C, Klug S, Köhler J, Papanikolaou A, Mitsakis V et al. (2012). Project summary and policy conclusions.
PT

Deliverable 7. EU FP7 research project No 233783: WEATHER, https://cordis.europa.eu/project/rcn/99129_en.html

Dong Y, Frangopol DM (2015). Risk and resilience assessment of bridges under mainshock and aftershocks
incorporating uncertainties. Eng. Struct. 83:198–208.
CE

Dong Y, Frangopol DM, Saydam D (2013). Time-variant sustainability assessment of seismically vulnerable bridges
subjected to multiple hazards. Earthquake Engineering and Structural Dynamics 42:1451–1467.
Dutta CH, Roy R (2002). A critical review on idealization and modelling for interaction among soil–foundation–structure
system. Computers and Structures 80(3):1579-1594.
AC

Eidsvig UMK, Kristensen K, Vangelsten BV (2017). Assessing the risk posed by natural hazards to infrastructures. Nat.
Hazards Earth Syst. Sci. 17:481-504.
EN 1998-3 (2005). Eurocode 8: Design of structures for earthquake resistance-Part 3: Assessment and retrofitting of
buildings. Brussels, CEN.
EN1991-1-7 (2006). Eurocode 1: Actions on structures - Part 1-7: General actions - Accidental actions. The European
Union.
Ertugay K, Argyroudis S, Düzgün S (2016). Accessibility modelling in earthquake case considering road closure
probabilities: A case study of health and shelter service accessibility in Thessaloniki, Greece. International Journal of
Disaster Risk Reduction 17:49–66.
Espinet X, Schweikert A, van den Heever N, Chinowsky P (2016). Planning resilient roads for the future environment
and climate change: Quantifying the vulnerability of the primary transport infrastructure system in Mexico. Transport
Policy 50:78-86. doi:10.1016/j.tranpol.2016.06.003.

35
 ACCEPTED MANUSCRIPT

Eusgeld I, Nan C, Dietz S (2011). “System-of-systems” approach for interdependent critical infrastructures. Reliab Eng
Syst Saf 96(6):679-686.
Fabozzi S, Bilotta E, Lanzano G (2017). A numerical study on seismic vulnerability of tunnel linings. In: Proceedings of
PBD III, Earthquake Geotechnical Engineering, Vancouver.
Fabozzi S, Bilotta E, Picozzi M, Zollo A (2018). Feasibility study of a loss-driven earthquake early warning and rapid
response systems for tunnels of the Italian high-speed railway network. Soil Dynamics and Earthquake Engineering
112:232–242
Faturechi R, Miller-Hooks E (2015). Measuring the performance of transportation infrastructure systems in disasters: a
comprehensive review. J. Infrastruct. Syst. 21(1).
FHWA (2012a). Evaluating scour at bridges, Hydraulic Engineering Circular No. 18, Fifth edition, Publication No.
FHWA-HIF-12-003, Washington, DC (Arneson A, Zevenbergen LW, Lagasse PF, Clopper PE).
FHWA (2012b). Climate Change & Extreme Weather Vulnerability Assessment Framework. U.S. Department of

T
Transportation (DOT). Retrieved from

IP
https://www.fhwa.dot.gov/environment/sustainability/resilience/publications/vulnerability_assessment_framework/fh
wahep13005.pdf
FHWA (2013). Risk-Based Transportation Asset Management: Building Resilience into Transportation Assets. Report 5:

CR
managing external threats through risk-based asset management. US Department of Transportation, Federal Highway
Administration, March, Available at: https://www.fhwa.dot.gov/asset/pubs/hif13018.pdf
FHWA (2014). Highways in the Coastal Environment: Assessing Extreme Events. Hydraulic Engineering Circular No.
25 (vol. 2), Publication No. FHWA- NHI-14-006, Washington, DC (Douglass SL, Webb BM, Kilgore R).
FHWA (2015). Climate change adaptation
US for
https://www.fhwa.dot.gov/pavement/sustainability/hif15015.pdf
pavements, Tech Brief, FHWA-HIF-15-015,

Franchin P (2018). Research needs towards a resilient community. Theme lecture at the 16th European Conference of
AN
Earthquake Engineering, Thessaloniki, Greece, June 18-21.
Franchin P, Pinto PE (2009). Allowing traffic over mainshock damaged bridges. Journal of Earthquake Eng, 13(5):585–
599.
Frangopol DM, Lin K-Y, Estes AC (1997). Reliability of reinforced concrete girders under corrosion attack. Journal of
M

Structural Engineering 123(3):286-297.

Gardoni P, Der Kiureghian A, Mosalam KM (2002). Probabilistic capacity models and fragility estimates for reinforced
concrete columns based on experimental observations. ASCE J. Eng. Mech. 128:1024–1038.
ED

Gehl P, D’Ayala D (2016). Development of Bayesian Networks for the multi-hazard fragility assessment of bridge
systems. Structural Safety 60:37-46.
Ghosh J, Padgett JE (2011). Probabilistic seismic loss assessment of aging bridges using a component‐ level cost
estimation approach. Earthquake Engineering & Structural Dynamics 40(15):1743-1761.
PT

Ghosh J, Padgett, JE, Sánchez-Silva M (2015). Seismic damage accumulation in highway bridges in earthquake-prone
regions. Earthquake Spectra 31(1):115–135.
Ghosh J, Sood P (2016). Consideration of time-evolving capacity distributions and improved degradation models for
CE

seismic fragility assessment of aging highway bridges. Reliab Eng Syst Saf 154:197-218.
Gidaris I, Padgett JE, Barbosa AR, Chen S, Cox D, Webb B, Cerato A (2017). Multiple-hazard fragility and restoration
models of highway bridges for regional risk and resilience assessment in the United States: state-of-the-art review.
Journal of Structural Engineering 143(3).
AC

Gill JC, Malamud BD (2014). Reviewing and visualizing the interactions of natural hazards. Reviews of Geophysics,
52(4):680-722.
Goretti A, Sarli V (2006). Road network and damaged buildings in urban areas: short and long-term interaction, Bulletin
of Earthquake Engineering 4:159-75.
Griffiths JS, Radford T (2012). An introduction to earthworks in Europe. Geological Society of London, Engineering
Geology Special Publications 26(1):1-4, doi: 10.1144/EGSP26.1
Guo X, Wu Y, Guo Y (2016). Time-dependent seismic fragility analysis of bridge systems under scour hazard and
earthquake loads. Engineering Structures 121:52–60.
Hackl J, Lam JC, Heitzler M, Adey BT, Hurni L (2018). Estimating network related risks: A methodology and an
application in the transport sector. Nat. Hazards Earth Syst. Sci. 18:2273–2293.

36
 ACCEPTED MANUSCRIPT

Hancilar U, Taucer F (eds) (2013). Guidelines for typology definition of European physical assets for earthquake risk
assessment. SYNER-G Reference Report 2. Publications Office of the European Union, Luxembourg, doi:
10.2788/68751.
Highways England (2016). Resilience of geotechnical assets on the Strategic Road Network to severe weather events.
Phase 2 – Final report, Issue 3.
Hughes JF, Healy K (2014). Measuring the resilience of transport infrastructure. NZ Transport Agency research report
546. 82pp.
Huh J, Tran QH, Haldar A, Park I, Ahn J-H (2017). Seismic vulnerability assessment of a shallow two-story underground
RC box structure. Appl. Sci. 7(735), doi: 10.3390/app7070735.
Hung C, Yau W (2017). Vulnerability evaluation of scoured bridges under floods. Engineering Structures 132: 288-299.
Iervolino I, Galasso C, Paolucci R, Pacor F (2011). Engineering ground motion record selection in the Italian
Accelerometric Archive. Bulletin of Earthquake Engineering 9:1761–78.

T
ITA (1991). Report on the damaging effects of water on tunnels during their working life. Tunnelling and Underground

IP
Space Technology 6(1):11-76.
Japan Road Association (JRA) (2007). Guideline for restoration work of road after earthquakes (in Japanese).
Jasim FH, Vahedifard F (2017). Fragility curves of earthen levees under extreme precipitation. Geotechnical Frontiers

CR
2017, GSP 278:353-362.
JBA Trust (2014). Flood and scour related failure incidents at railway assets between 1846 and 2013, Project W13-4224.
Jenelius E, Mattsson L-G (2015). Road network vulnerability analysis. Computers, Environment and Urban Systems
49:136-147.

US
Kallias AN, Imam B, Chryssanthopoulos MK (2017). Performance profiles of metallic bridges subject to coating
degradation and atmospheric corrosion. Structure and Infrastructure Engineering 13(4):440-453.
Kameshwar S, Padgett JE (2014). Multi‐ hazard risk assessment of highway bridges subjected to earthquake and
AN
hurricane hazards, Engineering Structures 78:154‐ 166.
Kappes MS, Papathmoa-Kohle M, Keiler M (2012). Assessing physical vulnerability for multi-hazards using an
indicator-based methodology. Appl. Geogr. 32(2): 577–590.
Kappos AJ, Saiidi MS, Aydınoğlu MN, Isaković T (Eds.) (2012). Seismic design and assessment of bridges: inelastic
M

methods of analysis and case studies (Vol. 21). Springer Science & Business Media.
Karamlou A, Bocchini P (2017). Functionality-fragility surfaces. Earthquake Engineering Structural Dynamics
46(10):1687–1709, doi: 10.1002/eqe.2878.
ED

Katsanos EI, Sextos AG (2013). ISSARS: An integrated software environment for structure specific earthquake ground
motion selection. Advances in Engineering Software 58:70–85.
Kawashima K, Unjoh S, Jun-Ichi Hoshikuma J-I, Kosa K (2011). Damage of bridges due to the 2010 Maule, Chile,
earthquake. Journal of Earthquake Engineering 15(7):1036-1068.
PT

Khademi N, Balaei B, Shahri M, Mirzaei M, Sarrafi B, Zahabiun M, Mohaymany AS (2015). Transportation network
vulnerability analysis for the case of a catastrophic earthquake. Int J Disaster Risk Reduction 12:234-254.
Kiani M, Ghalandarzadeh A, Akhlaghi T, Ahmadi M (2016). Experimental evaluation of vulnerability for urban
CE

segmental tunnels subjected to normal surface faulting. Soil Dynamics and Earthquake Engineering 89:28-37.
Kiel, J et al. (2016). A decision support system for the resilience of critical transport infrastructure to extreme weather
events. Transportation Research Procedia 14:68-77.
Kim H, Sim S-H, Lee J, Lee Y-J, Kim J-M (2017). Flood fragility analysis for bridges with multiple failure modes.
AC

Advances in Mechanical Engineering 9(3):1–11.

Kirby AM, Roca M, Kitchen A, Escarameia M, Chesterton OJ (2015). Manual on scour at bridges and other hydraulic
structures, 2nd edn., CIRIA Report C742, CIRIA, London.
Kongar I, Esposito S, Giovinazzi S (2017). Post-earthquake assessment and management for infrastructure systems:
learning from the Canterbury (New Zealand) and L’Aquila (Italy) earthquakes. Bulletin of Earthquake Engineering
15(2):589-620.
Kumar R, Gardoni P (2014). Effect of seismic degradation on the fragility of reinforced concrete bridges. Engineering
Structures 79:267-275.
Kwon OS, Elnashai AS (2010). Fragility analysis of a highway over-crossing bridge with consideration of soil–structure
interactions. Structure and Infrastructure Engineering 6:159-178.
Lagaros N, Tsompanakis Y, Psarropoulos P, Georgopoulos E (2009). Computationally efficient seismic fragility analysis
of geostructures. Comp. Struct., 87, 1195-1203.

37
 ACCEPTED MANUSCRIPT

Lagasse PF, Schall JD, Johnson F, Richardson EV, Chang F (1995). Stream stability at highway structures. Washington
DC.
Lam JC, Adey BT, Heitzler M, Hackl J, Gehl P, van Erp N, D’Ayala D, van Gelder P, Hurni L (2018). Stress tests for a
road network using fragility functions and functional capacity loss functions. Reliab Eng Syst Saf 173:78-93.
Lamb R, Aspinall W, Odbert H, Wagener T (2017). Vulnerability of bridges to scour: insights from an international
expert elicitation workshop. Nat. Hazards Earth Syst. Sci. 17:1393–1409.
Lee T-H, Park D, Nguyen DD, Park J-S (2016). Damage analysis of cut-and-cover tunnel structures under seismic
loading. Bull Earthquake Eng 14:413-431.
Linkov I, Bridges T, Creutzig F, Decker J, Fox-Lent C, Kröger W, et al. (2014). Changing the resilience paradigm.
Nature Climate Change 4(6):407.
Liu B, Siu YL, Mitchell G (2016). Hazard interaction analysis for multi-hazard risk assessment: a systematic
classification based on hazard-forming environment. Natural Hazards and Earth System Sciences, 16(2), 629-642.

T
Lozano-Valcárcel JM, Obregón N (2017). Generation and simple sensitivity analysis of fragility surfaces generated for

IP
two failure mechanisms in river levees for flood risk analysis using Monte Carlo simulations, World Environmental
and Water Resources Congress.
Mackie K, Stojadinovic B (2005). Fragility basis for California highway overpass bridge seismic decision making, PEER

CR
Report 2005/12, Pacific Earthquake Engineering Research Center, University of California, Berkeley CA.
Mackie K, and Stojadinovic B (2006). Post‐ earthquake functionality of highway overpass bridges, Earthquake
Engineering and Structural Dynamics 35(1):77‐ 93.
Mangalathu S, Padgett JE, DesRoches R (2018). Development of a bridge-specific fragility methodology to improve the

US
seismic resilience of bridges. Earthquakes and Structures 15(3):253-261.
Markolf S, Hoehne C, Fraser A, Underwood S, Chester M (2019). Transportation resilience to climate change and
extreme weather events – beyond risk and robustness. Transport Policy. DOI: 10.1016/j.tranpol.2018.11.003.
AN
Martinovic K, Gavin K, Reale C (2016). Assessing the vulnerability of Irish rail network earthworks, Transportation
Research Procedia 14:1904-1913.
Maruyama Y, Yamazaki F, Mizuno K, Tsuchiya Y, Yogai H (2010). Fragility curves for expressway embankments based
on damage datasets after recent earthquakes in Japan. Soil Dynamics and Earthquake Engineering 30:1158-1167.
M

Marzocchi W, Mastellone M, Di Ruocco A, Novelli P, Romeo E, Gasparini P (2009). Principles of multi-risk

assessment: interactions amongst natural and man-induced risks. European Commission, Directorate-General for
Research, Environment Directorate.
ED

Mattsson L-G, Jenelius E (2015). Vulnerability and resilience of transport systems - a discussion of recent research.
Transportation Research Part A: Policy and Practice 81:16-34.
Mayoral JM, Argyroudis S, Castañon E (2016). Vulnerability of floating tunnel shafts for increasing earthquake loading.
Soil Dynamics and Earthquake Engineering 80:1-10
PT

Melchers RE, Frangopol DM (2008). Probabilistic modelling of structural degradation. Reliab Eng Syst Saf 93(3).
Melville BW, Coleman SE (2000). Bridge scour. Water Resources Publications.
Meyer MD, Weigel B (2011). Climate change and transportation engineering: preparing for a sustainable future. Journal
CE

of Transportation Engineering, 137(6). doi:10.1061/(ASCE)TE.1943-5436.0000108.

Miller M, Baker JW (2015). Ground-motion intensity and damage map selection for probabilistic infrastructure network
risk assessment using optimization. Earthquake Engineering & Structural Dynamics 44(7): 1139-1156.
Mitoulis SA (2012). Seismic design of bridges with the participation of seat-type abutments, Eng Struct 44: 222-233.
AC

Mitoulis S, Palaiochorinou A, Georgiadis I, Argyroudis S (2016). Extending the application of integral abutment bridges
in earthquake prone areas by using novel isolators of recycled materials. Earthquake Engineering and Structural
Dynamics 45(14):2283–2301, doi: 10.1002/eqe.2760.
MunichRe (2013). Press release: Floods dominate natural catastrophe statistics in first half of 2013. Available on:
https://www.munichre.com/en/media-relations/publications/press-releases/2013/2013-07-09-press-release/index.html.
Muriel-Villegas JE, Alvarez-Uribe KC, Patiño-Rodrígue CE, Villegas JG (2016). Analysis of transportation networks
subject to natural hazards-Insights from a Colombian case (2016). Reliab Eng Syst Saf 152:151-165.
Murray AT, Matisziw TC, Grubesic TH (2008). A methodological overview of network vulnerability analysis, Growth
and Change 39(4):573–592.
Mylonakis GE, Gazetas G (2000). Seismic soil-structure interaction: beneficial or detrimental? Journal of Earthquake
Engineering 4:277–301.

38
 ACCEPTED MANUSCRIPT

Nakamura S (2015). Fragility characteristics about a damage of expressway embankment caused by the 2011 off the
Pacific Coast of Tohoku earthquake. In: Life-Cycle of Structural Systems Design, Assessment, Maintenance and
Management (Eds. Hitoshi Furuta, Dan M. Frangopol and Mitsuyoshi Akiyama), CRC Press 2014, Pages 1355-1361,
Print ISBN: 978-1-138-00120-6.
National Cooperative Highway Research Program (2010a). Effects of debris on bridge pier scour, NCHRP Report 653,
Transportation Research Board, National Academy of Science, Washington, D.C., (Lagasse PF, Clopper PE,
Zevenbergen LW, Spitz WJ, Girard LG).
National Cooperative Highway Research Program (NCHRP) (2010b). Estimation of scour depth at bridge abutments,
NCHRP Project 24-20, Draft Final Report, Transportation Research Board, National Academy of Science,
Washington, D.C., (Ettema R, Nakato T, Muste M).
National Cooperative Highway Research Program (NCHRP) (2011). Evaluation of bridge pier scour research: scour
processes and prediction, NCHRP Project 24-27(01), Transportation Research Board, National Academy of Science,

T
Washington, D.C., (Ettema R, Constantinescu G, Melville BW).

IP
National Cooperative Highway Research Program (NCHRP) (2014). Strategic Issues Facing Transportation, Volume 2:
Climate Change, Extreme Weather Events, and the Highway System: Practitioner s Guide and Research Report.
Washington, DC: The National Academies Press. https://doi.org/10.17226/22473.

CR
National Infrastructure Unit (NIU) (2011). National infrastructure plan 2011. Wellington: National Infrastructure Unit,
The Treasury.
National Institute of Building Sciences (NIBS) (2004). HAZUS-MH: Technical manual. Report developed by FEMA,
Washington, DC, www.fema.gov/plan/prevent/hazus

EUR 25553 EN. US

Nemry F, Demirel H (2012). Transport and climate change: a focus on road and rail transport infrastructures. JRC72217,

Nguyen DD, Park D, Shamsher S, Nguyen VQ, Lee TH (2019). Seismic vulnerability assessment of rectangular cut-and-
AN
cover subway tunnels. Tunnelling and Underground Space Technology 86:247-261.
Nielson B, DesRoches R (2007). Analytical seismic fragility curves for typical highway bridge classes in the central and
Southeastern United States. Earthquake Spectra 23(3):615–633.
O'Brien EJ, Hajializadeh D, Richard RT (2015). Quantifying the impact of critical infrastructure failure due to extreme
M

weather events. In Proc of the 12th International Conference on Applications of Statistics and Probability in Civil
Engineering (ICASP12), Vancouver, Canada, 12 -15 July.
Ouyang M (2014). Review on modeling and simulation of interdependent critical infrastructure systems. Reliab Eng Syst
ED

Saf 121:43-60.
Padgett JE, DesRoches R (2007). Sensitivity of seismic response and fragility to parameter uncertainty. Journal of
Structural Engineering 133(12):1710-1718.
Padgett JE, DesRoches R (2009). Retrofitted bridge fragility analysis for typical classes of multispan bridges. Earthquake
PT

Spectra 25(1):117–141.
Padgett JE, Ghosh J, Duenas-Osorio L (2013). Effects of liquefiable soil and bridge modelling parameters on the seismic
reliability of critical structural components. Struct Infrastruct Eng 9(1):59–77
CE

Pant R, Thacker S, Hall JW, Alderson D, Barr S (2018). Critical infrastructure impact assessment due to flood exposure.
J Flood Risk Management 11:22-33.
Peduto D, Elia F, Rosario Montuori R (2018). Probabilistic analysis of settlement-induced damage to bridges in the city
of Amsterdam (The Netherlands). Transportation Geotechnics 14:169–182.
AC

Petrucci O, Gulla G (2010). A simplified method for assessing landslide damage indices. Nat Hazards 52:539–560.
Pitilakis, et al. (2010). Physical vulnerability of elements at risk to landslides: Methodology for evaluation, fragility
curves and damage states for buildings and lifelines. Deliverable 2.5, EU FP7 research project No 226479: SafeLand.
Pitilakis K et al. (eds) (2014) SYNER-G: typology definition and fragility functions for physical elements at seismic risk.
GGEE 27, Springer.
Porter K (2015) A Beginner’s Guide to Fragility, Vulnerability, and Risk. In: Beer M, Kougioumtzoglou I, Patelli E, Au
IK (eds) Encyclopedia of Earthquake Engineering. Springer, Berlin, Heidelberg.
Power M, Fishman K, Richards R, Makdisi F, Musser S, Leslie Youd TL (2004). Seismic Retrofitting Manual for
Highway Structures: Part 2-Retaining Structures, Slopes, Tunnels, Culverts, and Roadways, Publication No. FHWA-
HRT-05-067, Federal Highway Administration, US Department of Transportation.
Prasad GG, Banerjee S (2013). The impact of flood-induced scour on seismic fragility characteristics of bridges, J. Earth.
Eng. 17:803-828.

39
 ACCEPTED MANUSCRIPT

Pregnolato M, Ford A, Wilkinson SM, Dawson RJ (2017). The impact of flooding on road transport: A depth-disruption
function. Transportation Research Part D: Transport and Environment 55:67-81.
Prendergast L J, Gavin K (2014). A review of bridge scour monitoring techniques. Journal of Rock Mechanics and
Geotechnical Engineering 6(2):138-149.
Qiu W, Huang G, Zhou H, Xu W (2018). Seismic vulnerability analysis of rock mountain tunnel. International Journal of
Geomechanics 18(3).
Rattanachot W, Wang Y, Chong D, Suwansawas S (2015). Adaptation strategies οf transport infrastructures τo global
climate change. Transport Policy 41:159-166. doi:10.1016/j.tranpol.2015.03.001.
Reder Α, Iturbide Μ, Herrera S, Rianna G, Mercogliano P, Gutiérrez JM (2018). Assessing variations of extreme indices
inducing weather-hazards on critical infrastructures over Europe—the INTACT framework. Climatic Change 148(1-
2):123-138.
Roca M, Whitehouse R (2012). Scour risk assessment at river crossings. In Proc. ICSE6 Paris, August 27-31.

T
Rossetto T, D’Ayala D, Ioannou I, Meslem A (2014). Evaluation of Existing Fragility Curves. In: Pitilakis K, Crowley

IP
H, Kaynia AM (eds) SYNER-G: Typology definition and fragility functions for physical elements at seismic risk.
Geotechnical, Geological and Earthquake Engineering 27, Springer Netherlands.
Schultz MT, Gouldby BP, Simm JD, Wibowo JL (2010). Beyond the factor of safety: Developing fragility curves to

CR
characterize system reliability. Water Resources Infrastructure Program ERDC-SR-10-1. U.S. Army Corps of
Engineers, Engineering Research and Development Center, Washington, DC
Schweikert A, Chinowsky P, Kwiatkowski K, Espinet X (2014). The infrastructure planning support system: Analyzing
the impact of climate change on road infrastructure and development. Transport Policy 35:146-153.

analysis. Nat Hazards 67:723-746. US

Selva J, Argyroudis S, Pitilakis K (2013). Impact on loss/risk assessments of inter-model variability in vulnerability

Sextos AG, Kappos A (2008). Seismic response of bridges under asynchronous excitation and comparison with EC8
AN
design rules, Bulletin of Earthquake Engineering 7:519-545.
Shinozuka M, Feng MQ, Kim H, Uzawa T, Ueda T (2001). Statistical analysis of fragility curves. Technical Rep.
MCEER, Buffalo, NY.
Silva V, Akkar S, Baker J, Bazzurro P, Castro JM, Crowley H, Dolsek M, Galasso C, Lagomarsino S, Monteiro R,
M

Perrone D, Pitilakis K, Vamvatsikos D (2019). Current Challenges and Future Trends in Analytical Fragility and
Vulnerability Modelling. Earthquake Spectra (in press).
Stefanidou S, Sextos AG, Kotsoglou AN, Lesgidis N, Kappos AJ (2017). Soil-structure interaction effects in analysis of
ED

seismic fragility of bridges using an intensity-based ground motion selection procedure. Eng. Struct. 151:366-380.
Stefanidou SP, Kappos AJ (2017). Methodology for the development of bridge- specific fragility curves. Earthq. Eng.
Struct. Dyn. 46(1):73-93.
Stefanidou S, Kappos A (2018). Bridge‑ specific fragility analysis: when is it really necessary? Bulletin of Earthquake
PT

Engineering. https://doi.org/10.1007/s10518-018-00525-9.
Stern PC, Ebi KL, Leichenko R, Olson RS, Steinbruner JD, Lempert R (2013). Managing risk with climate vulnerability
science. Nature Climate Change 3(7):607.
CE

Suh J, Choi Y, Roh T-D, Lee H-J, Park H-D (2011). National-scale assessment of landslide susceptibility to rank the
vulnerability to failure of rock-cut slopes along expressways in Korea, Environmental Earth Sciences 63:619-632.
Tanasic N, Hajdin R (2017). Management of bridges with shallow foundations exposed to local scour. Structure and
Infrastructure Engineering, doi.org/10.1080/15732479.2017.1406960
AC

Tanasic N, Ilic V, Hajdin R (2013). Vulnerability assessment of bridges exposed to scour, Transportation Research
Record: Journal of the Transportation Research Board 2360:36‐ 44.
The Parliamentary Office of Science and Technology (2010), Post Note Number 362, October.
Torbol M, Gomez H, Feng M (2013). Fragility analysis of highway bridges based on long-term monitoring data.
Computer-Aided Civil and Infrastructure Engineering 28:178-192.
Transportation Research Board (TRB) (2008). TRB Special Report 290 - Potential Impacts of Climate Change on U.S.
Transportation. Washington, D.C.: The National Academy of Sciences.
Tsionis G, Fardis MN (2014). Fragility functions of road and railway bridges. In: Pitilakis K, Crowley H, Kaynia AM
(eds) SYNER-G: Typology definition and fragility functions for physical elements at seismic risk. Geotechnical,
Geological and Earthquake Engineering 27, Springer Netherlands.
Tsubaki R, Bricker JD, Ichii K, Kawahara Y (2016). Development of fragility curves for railway embankment and ballast
scour due to overtopping flood flow. Nat. Hazards Earth Syst. Sci., 16:2455–2472.

40
 ACCEPTED MANUSCRIPT

Tubaldi E, Macorini L, Izzuddin BA, Manes C, Laio F (2017). A framework for probabilistic assessment of clear-water
scour around bridge piers. Structural Safety 69:11–22.
van Leeuwen Z, Lamb R (2014). Flood and scour related failure incidents at railway assets between 1846 and 2013.
Skipton, UK, JBA Trust, Prj W13-4224.
Vespignani A (2010). Complex networks: The fragility of interdependency. Nature, 464(7291): 984
Wang B, Reul M, Couture R, Bobrowsky PT, Blais-Stevens A (2012). Review of existing landslide guidelines - national
technical guidelines and best practices on landslides; Geological Survey of Canada, Open File 7058, 13 p.
Washington State Department of Transportation (WSDOT) (2011). Climate impacts vulnerability assessment, Nov. 2011.
Werner SD, Taylor CE, Cho S, Lavoie J-P, Huyck C, Eitzel C, Chung H, Eguchi RT (2006). REDARS 2: Methodology
and Software for Seismic Risk Analysis of Highway Systems MCEER-06-SP08.
White DJ, Mekkawy MM, Sritharan S, Suleiman MT (2007). Underlying causes for settlement of bridge approach
pavement systems. Journal of Performance of Constructed Facilities, 21(4).

T
Willway T, Baldachin L, Reeves S, Harding M, McHale M, Nunn M (2008). The effects of climate change on highway

IP
pavements and how to minimize them. Published Project Report 184. Transport Research Laboratory, Wokingham.
Winter M G (2014). A strategic approach to landslide risk reduction. International Journal of Landslide and Environment
2(1):14-23.

CR
Winter M G (2018). The quantitative assessment of debris flow risk to road users on the Scottish trunk road network:
A85 Glen Ogle. Published Project Report PPR 799. Transport Research Laboratory, Wokingham.
Winter MG, Smith JT, Fotopoulou S, Pitilakis K, Mavrouli O, Corominas J, Argyroudis S (2014). An expert judgement
approach to determining the physical vulnerability of roads to debris flow, Bull. Eng. Geol. Envir. 73(2):291‐ 305.

US
Wong J F C, Winter M G (2018). The quantitative assessment of debris flow risk to road users on the Scottish trunk road
network: A83 Rest and be Thankful. Published Project Report PPR 798. Transport Research Laboratory,
Wokingham.
AN
Wright KM, Hogan C (2008). The potential impacts of global sea level rise on transportation infrastructure. Washington,
D.C.: ICF International.
Wu X.Z (2015). Development of fragility functions for slope instability analysis. Landslides 12:165-175.
Yang J, Sun H, Wang L, Li L, Wu B (2013). Vulnerability evaluatiοn οf the highway transportation system against
M

meteorological disasters. Procedia-social and Behavioral Sciences 96:280-293.

Yilmaz T, Banerjee S, Johnson PA (2016). Performance of two real-life California bridges under regional natural
hazards. Journal of Bridge Engineering 21(3).
ED

Yilmaz T, Banerjee S, Johnson PA (2017). Uncertainty in risk of highway bridges assessed for integrated seismic and
flood hazards Structure and Infrastructure Engineering 14(9).
Yin C, Shi J, Liu F-F, Tian W-P, Tian W (2017). Embankment seismic fragility assessment based on IDA-PSDA. China
J. Highway Transp., 30(5):28-37.
PT

Zaghi AE, Padgett JE, Bruneau M, Barbato M, Li Y, Mitrani-Reiser J, McBride A (2016). Establishing common
nomenclature, characterizing the problem, and identifying future opportunities in multihazard design. J. Struct. Eng.,
142(12):H2516001.
CE

Zanini MA, Faleschini F, Zampieri P, Pellegrino C, Gecchele G, Gastaldi M, Rossi R (2017). Post-quake urban road
network functionality assessment for seismic emergency management in historical centres. Structure and
Infrastructure Engineering 13(9):1117-1129.
Zhong J, Gardoni P, Rosowsky D (2012). Seismic fragility estimates for corroding reinforced concrete bridges. Structure
AC

and Infrastructure Engineering 8(1):55-69.

Zurich Insurance Group and JBA Trust (2016). Flooding after Storm Desmond, http://www.jbatrust.org/wp-
content/uploads/2016/08/flooding-after-storm-desmond-PUBLISHED-24-August-2016.pdf.

41