You are on page 1of 12

Climate Impact Risks and Climate Adaptation Engineering

for Built Infrastructure


Mark G. Stewart 1 and Xiaoli Deng 2
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: A changing climate may increase the frequency or intensity of natural hazards, resulting in increased infrastructure damage. The
paper will describe how risk-based approaches are well suited to optimising climate adaptation strategies related to the construction, design,
operation, and maintenance of built infrastructure. Climate adaptation engineering involves estimating the risks, costs, and benefits of climate
adaptation strategies and assessing at what point in time climate adaptation becomes economically viable. Stochastic methods are used to
model infrastructure performance, risk reduction, and effectiveness of adaptation strategies, exposure, and costs. These concepts will be
illustrated with recent research on risk-based life-cycle assessments of climate adaptation strategies for Australian housing subject to extreme
wind events. This will pave the way for more efficient and resilient infrastructure, and help future proof new and existing infrastructure to a
changing climate. DOI: 10.1061/AJRUA6.0000809. © 2014 American Society of Civil Engineers.
Author keywords: Risk; Climate change; Cost-benefit analysis; Infrastructure; Climate adaptation.

Introduction shows that business-as-usual CO2 emissions continue to track at


the high end of emission scenarios, with mean temperature
A changing climate may result in more intense tropical cyclones increases of 4–5°C more likely by the year 2100 (Peters et al.
and storms, more intense rain events and flooding, and other 2013). The impacts on people and infrastructure may be consider-
climate-related hazards. Moreover, increases in CO2 atmospheric able if there is no climate adaptation engineering to existing and
concentrations, as well as changes in temperature and humidity, new infrastructure. Some posit that climate change may even be
may reduce the durability of concrete, steel, and timber structures a threat to national security, but Stewart (2014a) suggests that cli-
(Bastidas-Arteaga et al. 2010, 2013; Wang et al. 2012; Stewart et al. mate change threats to U.S. national security are modest and man-
2011b, 2012a; Nguyen et al. 2013; Peng and Stewart 2014; Wang ageable. On the other hand, higher temperatures in higher-latitude
and Wang 2012). The impact of climate change on infrastructure regions such as Russia and Canada can be beneficial through higher
performance is a temporal and spatial process, but most existing agricultural yields, lower winter mortality, lower heating require-
models of infrastructure performance are based on a stationary cli- ments, and a potential boost to tourism (Stern 2007).
mate. Hence, there is a need to quantify the costs and benefits of There is seldom mention of probabilities, or quantitative mea-
adaptation strategies. Climate adaptation engineering involves es- sures of vulnerability, or the likelihood or extent of losses in risk
timating the risks, costs, and benefits of climate adaptation strate- and risk management reports on climate change and infrastructure.
gies and assessing at what point in time climate adaptation becomes While useful for initial risk screening, intuitive and judgement-
economically viable. Climate adaptation measures aim to reduce based risk assessments are of limited utility to complex decision-
the vulnerability or increase the resiliency of built infrastructure making since there are often a number of climate scenarios,
to a changing climate; this may include, for example, enhancement adaptation options, limited funds, and doubts about the cost-
of design standards, retrofitting or strengthening of existing struc- effectiveness of adaptation options. In this case, the decision-maker
tures, utilization of new materials, and changes to inspection and may still be uncertain about the best course of action, and so a de-
maintenance regimes. Engineers have a unique capability to model tailed risk analysis is required [e.g., AS 5334 (Standards Australia
infrastructure vulnerability, and these skills will be essential to 2013)]. For this reason, there is a need for sound system and prob-
modeling future climate impacts and measures to ameliorate these abilistic modeling that integrates the engineering performance of
losses. infrastructure with the latest developments in stochastic modeling,
The climate change literature places more emphasis on impact structural reliability, and decision theory.
modeling than climate adaptation engineering modeling. This is to The paper will describe how risk-based approaches are well
be expected when the current political and social environment is suited to optimizing climate adaptation strategies related to the de-
focused on mitigating (reducing) CO2 emissions. Latest research sign, construction, operation, and maintenance of built infrastruc-
ture. An important aspect is assessing when climate adaptation
1
Australian Professorial Fellow, Professor and Director, Centre for becomes economically viable, if adaptation can be deferred, and
Infrastructure Performance and Reliability, Univ. of Newcastle, NSW 2308, decision preferences for future costs and benefits (many of them
Australia (corresponding author). E-mail: mark.stewart@newcastle.edu.au intergenerational). Stochastic methods are used to model infrastruc-
2
Senior Lecturer, School of Engineering, Univ. of Newcastle, NSW ture vulnerability, effectiveness of adaptation strategies, exposure,
2308, Australia.
and costs. The concepts will be illustrated with a case study that
Note. This manuscript was submitted on April 8, 2014; approved on
July 29, 2014; published online on August 27, 2014. Discussion period
considers climate change and cost-effectiveness of designing
open until January 27, 2015; separate discussions must be submitted for new houses in Sydney, Australia to be less vulnerable to severe
individual papers. This paper is part of the ASCE-ASME Journal of Risk storms. To be sure, there are other case studies assessing the
and Uncertainty in Engineering Systems, Part A: Civil Engineering, efficiency and cost-effectiveness of climate adaptation strategies
© ASCE, 04014001(12)/$25.00. for built infrastructure; for example, floods and sea-level rises

© ASCE 04014001-1 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
(e.g., Hinkel et al. 2010; Hall et al. 2012; Botzen et al. 2013; The 2006 review by economist Nicholas Stern (Stern 2007) pre-
Kundzewicz et al. 2013; Holden et al. 2013; Val et al. 2013), cyclo- dicts that if no action is taken against climate change, the mean loss
nes and severe storms (Bjarnadottir et al. 2011a, b; Nishijima et al. of GDP would be 2.9 and 13.8% each year (“now and forever”) by
2012; Li and Stewart 2011; Stewart et al. 2012b), and corrosion of years 2100 and 2200, respectively. This is equivalent to worldwide
reinforced concrete (Stewart and Peng 2010; Bastidas-Arteaga and losses of up to $10 trillion each year by 2200. Not surprisingly,
Stewart 2013). This and other research will help future proof built some consider it to be highly pessimistic in its assumptions (Lomborg
infrastructure to a changing climate. 2006; Mendelsohn 2006). However, the Australian Garnaut Review
predicted that unmitigated climate change would reduce Australian
GDP by approximately 8% by 2100 (Garnaut 2008).
Key Issues These losses, however, do not reflect wealth creation, human
capital, and new improved technologies. Goklany (2008) states that
There are a number of issues and questions that need addressing. these “often reduce the extent of the human health and environmen-
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

These include tal ‘bads’ associated with climate change more than temperature
1. Cost neglect: increases exacerbate them.” Fatality rates and economic losses re-
• Who pays? lated to weather and climate are also 3–10 times higher in developing
• When? countries (IPCC 2012). Clearly then, if people are wealthier in the
• Who benefits? future, their well-being will be higher, and “the argument that we
2. Probability neglect—how confident are we about extreme should shift resources from dealing with the real and urgent problems
events in the current climate? confronting present generations to solving potential problems of
• How confident are we in changes to future climate? tomorrow’s wealthier and better positioned generations is unper-
• Changes in impact and loss? suasive at best and verging on immoral at worst.” (Goklany 2008).
3. Risk aversion:
• Is action needed now, is there no time to lose?
4. Acceptable risk: CO2 Mitigation versus Adaptation
• What risk from weather/climate is acceptable?
• Is risk reduction worth the cost? The cost to mitigate CO2 emissions is considerable. Stern (2007)
These are issues without easy answers, but the questions need to estimates that to stabilize CO2 levels to 550 ppm (by reducing total
be asked. They are also similar to other controversial and emotive emissions to three-quarters of today’s levels by the year 2050)
issues such as terrorism and homeland security (Mueller and would cost −1.0 to 3.5% of GDP, with a central estimate of approx-
Stewart 2011a, b). All too often, climate change studies assume imately 1%. The mean estimate would result in an annual mitiga-
tion cost of approximately $720 billion.
there is certainty about the future, and so suffer from probability
Lomborg (2009) assembled a group of experts who found that
neglect, as well as cost neglect by ignoring the large costs involved
climate change action ranked very low when compared with other
to mitigate CO2 emissions.
hazard and risk-reducing measures. In this case, the benefit-to-cost
ratio for CO2 mitigation was only 0.9 (not cost-effective), but
increased to 2.9 for a mix of mitigation and adaptation strategies.
Climate Change Impact Yohe et al. (2009) found that a global investment of $18 billion
per year in “R&D and mitigation” can halve business-as-usual
The 2014 Intergovernmental Panel for Climate Change (IPCC) fifth
CO2 emissions by the year 2100. Such action would reduce the
assessment report (AR5) concluded that the “warming of the cli-
impact of climate change by at least 60%.
mate system is unequivocal” (IPCC 2014). What is less certain is
Some of the more dire predictions of food and energy insecurity,
the impact that rising temperatures will have on rainfall, wind pat-
and mass migration can be ameliorated by funding climate-
terns, sea-level rise, and other phenomena. The latest IPCC AR5
adaptation measures in the developing world. Adaptation measures
report released in 2014 describes the following changes to climate
to reduce vulnerability of infrastructure, coastal zones, agriculture,
by the year 2100 (IPCC 2014):
forestry, fisheries, and human health to climate change hazards
1. Temperatures to increase from 1990 levels by anywhere from
would include: flood control dikes and levees, dams, cyclone shel-
1 to 6°C;
ters, storm-resistant and flood-resistant housing, improved commu-
2. Sea-level rise of 20–80 cm;
nication infrastructure, resettlement of populations to lower risk
3. More intense tropical cyclones and other severe wind events;
zones, and improved health care. The World Bank (2010) estimated
4. Enhanced monsoon precipitation.
that the cost to the developing world of adapting to an approxi-
The IPCC (2014) then suggests with a high or very high con-
mately 2°C warmer world by the year 2050 is approximately
fidence level that these changes to climate will increase drought-
$75 billion per year. This represents less than 0.2% of global
affected areas, with hundreds of millions of people to be affected
GDP. Clearly, investing in targeted adaptation measures has the
by coastal flooding and increases in the risk of fire, pests, and dis-
potential to dramatically reduce the impact of climate change.
ease outbreak. There will also be significant consequences for food Mitigation costs may be high, and the benefits of reduced CO2
and forestry production and food insecurity, and so on. These levels will take many decades to accrue. Modest and sustained in-
impacts will not be sudden, but gradual in their appearance. The vestments in R&D, CO2 mitigation and adaptation will lessen the
observed increase in weather-related losses in the United States worst impacts of climate change. Hence, a mix of mitigation and
and elsewhere is more a function of increased exposure with more adaptation is desirable to cope with a changing climate.
people moving to vulnerable coastal locations than climate-change
increases in wind speed or flood levels (Crompton and McAneney
2008; IPCC 2012). For instance, hurricane wind speeds are pre- Modelling Climate Hazards
dicted to increase at worst by 10% in 50 years due to climate
change, or a low 0.2% per year (Bjarnadottir et al. 2011a). This Atmosphere-ocean general circulation models (AOGCMs) are
suggests that there will be time to adapt to a changing climate. currently the main tool for climate change studies (IPCC 2012).

© ASCE 04014001-2 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
However, they are computationally demanding, which limits their where PrðCÞ = annual probability that a specific climate scenario
spatial resolution. Downscaling is possible to grids of 25–50 km, will occur; PrðHjCÞ = annual probability of a climate hazard
but the resulting models are still too large to capture extreme events (wind, heat, etc.) conditional on the climate; PrðDjHÞ = probability
such as tornadoes or extreme rainfall. Uncertainties associated with of infrastructure damage or other undesired effects conditional on
future emission scenarios are usually not quantified, and future cli- the hazard (also known as vulnerability or fragility) for the baseline
mate projections are produced separately for individual scenarios case of no extra protection (i.e., business-as-usual); PrðLjDÞ =
(Stewart et al. 2014). Not surprisingly, uncertainties in climate projec- conditional probability of a loss (economic loss, loss of life,
tions are considerable. This makes impact and adaptation modeling etc.) given occurrence of the damage; and L = loss or consequence
even more challenging. For more details, see Stewart et al. (2014). if full damage occurs. In some cases, ‘damage’ may equate to ‘loss’
A major climate hazard is coastal flooding induced by extreme and so PrðLjHÞ ¼ PrðDjHÞPrðLjDÞ. The summation sign in
water level events along low-lying, highly populated coastlines due Eq. (1) refers to the number of possible climate scenarios, hazards,
to presently and continuously rising sea levels. Statistical analysis damage levels, and losses.
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

based on extreme value theory has been employed to estimate prob- If we modify Eq. (1) where ΔR is the reduction in risk caused by
abilities of extreme water levels and assess suitable design levels. climate adaptation measures, then expected loss after climate adap-
Estimating the average recurrence interval (ARI) and the annual tation is
exceedance probability (AEP) needs sea-level measurements over X
a long period greater than 30 years, traditionally observed at tide Eadapt ðLÞ ¼ ð1 − ΔRÞEðLÞ − ΔB ð2Þ
gauges. Over the last five decades, several statistical analysis meth-
ods for estimating the ARI and AEP have been developed (cf. where ΔR = reduction in risk caused by climate adaptation (or
Haigh et al. 2010). There is, however, no universally accepted other protective) measures; EðLÞ = business-as-usual risk given
method available at transnational or even national scales. There- by Eq. (1); and ΔB = cobenefit of adaptation such as reduced losses
fore, the most applicable method has to be chosen based on differ- to other hazards, increased energy efficiency of new materials, etc.
ent stretches of coastline and length of sea-level records. Costs of adaptation, timing of adaptation, discount rates, future
Over regions where long periods of tide-gauge data are not growth in infrastructure, and spatial and time-dependent increases
available or the spatial distribution of tide-gauge sites is sparse in climate hazards need to be included in any risk analysis. Fig. 1
along coastline, modeling of extreme sea levels can be made from summarizes the major steps in developing risk-based decision
sea level fields produced by ocean circulation models. A 2D depth- support for assessing the risks, costs, and benefits of climate-
averaged barometric ocean circulation model is usually configured adaptation measures.
over coastal oceans. The mode is driven by tidal and atmospheric Future climate is projected by defining carbon emission sce-
forcing. The tidal component including the Earth geocentric tide narios in relation to changes in population, economy, technol-
and other nonlinear tidal constituents are simulated using existing ogy, energy, land use and agriculture—a total of four scenario
global tidal modes and the harmonic analysis algorithm. The storm families, i.e., A1, A2, B1, and B2 are defined (IPCC 2000) and
surge component is simulated through the model driven by wind used in the IPCC’s Fourth Assessment Report in 2007 (IPCC
stress and the atmospheric pressure at the sea level, for which 2007). The A1 scenarios indicate very rapid economic growth,
the wind stress calculated from the wind velocity at 10 m above a global population that peaks in midcentury and declines there-
the sea surface and the sea level pressure can be taken from the after, and the rapid introduction of new and more efficient tech-
climate forecast system reanalysis fields. To achieve the fine struc- nologies, as well as substantial reduction in regional differences
ture of the wind and pressure associated with tropical storms, the in per capita income. The B1 scenarios are of a world more in-
reanalysis field is enhanced by adding the idealized wind and pres- tegrated and more ecologically friendly with reductions in
sure profiles during cyclones (e.g., Zhang and Sheng 2013). The material intensity and the introduction of clean and resource ef-
total sea levels due to the combination of tides and storm surges ficient technologies. The IPCC AR5 (2014) uses Representative
use Monte Carlo simulation methods. Concentration Pathways (RCPs) where RCP8.5, RCP6.0, and
In recent years, satellite data with high resolution and homo- RCP4.5 are roughly equivalent to A1FI, A1B, and A1B to
geneous global coverage have already played an important role B1 CO2 emissions, respectively. For more details see Stewart
in monitoring extreme sea levels, especially storm surges. How- et al. (2014)
ever, advanced satellite products such as wind speed, wind direc- The estimation of PrðCÞ may be based on expert opinion about
tion from high resolution scatterometry, and sea state information the likelihood of each emission scenario, and multiple AOGCMs
from coastal altimetry are not yet widely used in modeling extreme may be used to infer the probabilistic characterisation of PrðHjCÞ
sea levels. Since these satellite data have been collected for more for future climate projections. Fig. 2 describes the projection of
than 20 years, assimilating them into numerical models will have CO2 concentrations based on the Model for Assessment of Green-
the potential to improve estimation of extreme sea level. (e.g., Cheng house-Gas Induced Climate Change, known as MAGICC (Wigley
et al. 2012; Deng et al. 2011, 2013; Idris et al. 2014). Nonetheless, et al. 1996), specifically related to A1FI, A1B, and 550 ppm CO2
uncertainties are considerable, and remote sensing technologies are stabilization scenarios. The variability increases for projection of
increasingly relied upon for measuring and predicting sea-level rise temperatures (Fig. 3).
and other phenomena. Wind speed projections also face distinct The stochastic modeling of infrastructure vulnerability (or
measuring and prediction challenges (e.g., Harper 2013). fragility) is PrðDjHÞ and is the probability of damage conditional
on the occurrence of a specific hazard
Risk-Based Decision Support PrðDjHÞ ¼ PrðRðXÞ − H < 0Þ ð3Þ

Definition of Risk where RðXÞ = function for resistance or capacity; X = vector of all
relevant variables that affect resistance; and H = known hazard
Risk for a system exposed to a climate hazard is level. The performance functions can be expressed in terms of
X structural damage or other losses and are derived from engineer-
EðLÞ ¼ PrðCÞ PrðHjCÞ PrðDjHÞ PrðLjDÞL ð1Þ ing models (e.g., Stewart and Melchers 1997). The reliability

© ASCE 04014001-3 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Flowchart of decision-support framework for assessing cost-effectiveness of adaptation measures

modeling of structural systems is well developed for engineered Exposure and loss data relates to direct and indirect loss or con-
constructions such as commercial buildings, bridges, towers, etc. sequences due to location and extent of infrastructure damage, for
where materials are uniform, and workmanship is subject to quality existing exposure and future projections. Most existing studies con-
control measures. However, nonengineered infrastructures, particu- sider direct losses related to building damage and contents losses.
larly houses, are very complex systems comprising hundreds to Indirect losses caused by business interruption, clean-up, loss dur-
thousands of components and connections between differing ma- ing reconstruction, extra demands on social services, and changes
terials. Poor detailing and workmanship issues contribute to most to demand and supply of intermediate consumption goods, post-
damage—so the engineering and stochastic models need to con- disaster inflation, etc. can also be significant (e.g., NAS 1999;
sider these variables—such as screw fasteners being spaced too Hallegatte 2008; Walker 2011). Indirect losses were estimated
far apart, or some not connected to purlins and battens, etc. Insur- for Hurricane Katrina as $42 billion or 39% of direct losses,
ance or building performance data may also be used to derive vul- and could have exceeded 100% of direct losses for a damaging
nerability models. For example, Fig. 4 shows a vulnerability model event twice as bad as Hurricane Katrina (Hallegatte 2008). An
for Australian houses subject to floods. Australian assessment of direct and indirect costs shows indirect

© ASCE 04014001-4 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
1100 200%
low
CO2 Concentration (ppm) 1000 mid
high
900 Direct + Indirect Loss
A1FI
800

Pr(L|D)
700 100% Direct Loss
A1B
600
550 ppm
500

400
Year 2000 CO2 level
300
0
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

2000 2020 2040 2060 2080 2100


0 10 20 30 40 50 60 70 80 90 100
Year Vulnerability Pr(D|H) (%)
Fig. 2. Projected low, mid, and high estimates of CO2 concentrations
Fig. 5. Direct and indirect costs as function of wind vulnerability (data
from Stewart et al. 2013)

24

23 A1FI Risk reduction (ΔR) may result from reduced vulnerability


A1B
550 ppm
PrðDjHÞ, PrðLjDÞ, or exposure (L). For instance, changes to plan-
22 ning may reduce the number of new properties built in a flood plain
Temperature (°C)

Year 2000
which will reduce L, or more stringent design codes may reduce the
21
vulnerability of new infrastructure. Systems and reliability model-
20 ing are essential tools to quantify the level of risk reduction, and the
extent of risk reduction due to adaptation measures will depend on
19 the hazard, the location, and the timing of adaptation.
The cobenefits of adaptation (ΔB) may include reduced embod-
18
ied energy and reduced carbon footprint over the life cycle of the
17 facility. This might consider the initial embodied energy associ-
2000 2020 2040 2060 2080 2100 ated with the dwelling including footings, structure and fit-out
Year together with the recurrent embodied energy associated with refur-
Fig. 3. Projected median temperatures for the lowest and highest GCM bishment over the life cycle and the operational energy needed
predictions for the A1FI, A1B, 550 ppm, and year 2000 emission sce- to operate a building.
narios for Sydney (Australia)
Cost-Effectiveness of Adaptation Strategies
Two criteria may be used to assess the cost-effectiveness of adap-
100
tation strategies:
90 1. Net present value (NPV); and
80 2. Probability of cost-effectiveness or PrðNPV > 0Þ.
Vulnerability Pr(D|H) (%)

70 The benefit of an adaptation measure is the reduction in dam-


ages associated with the adaptation strategy, and the ‘cost’ is the
60
cost of the adaptation strategy. The net benefit or net present value
50 (NPV) is equal to benefit minus the cost, which is also equivalent to
Single Storey
40 the present value or life-cycle cost of an adaptation strategy (sum of
30 Two Storey damage and adaptation costs) minus the business-as-usual or do-
(lower floor partial use as garage) nothing present value. The decision problem is to maximize NPV
20
10 X
NPV ¼ EðLÞΔR þ ΔB − Cadapt ð4Þ
0
-1 0 1 2 3 4 5 6
Water Depth Above Floor Level (m) where Cadapt = cost of adaptation measures including opportunity
costs that reduces risk by ΔR; ΔB = cobenefit; and EðLÞ =
Fig. 4. Flood vulnerability curves for residential construction in business-as-usual risk given by Eq. (1). Fig. 6 shows how adapta-
Brisbane (data from Mason et al. 2012) tion costs increase with risk reduction, while benefits increase. The
optimal adaptation occurs when NPV is a maximum, leading to an
optimal risk reduction. Other notations and formulae can be used to
provide optimal adaptation (e.g., Hall et al. 2012), but ultimately
costs of 9–40% of direct losses for bushfires, cyclones, and floods these also mostly rely on maximizing NPV.
(BTE 2001). Fig. 5 shows a typical loss function for wind vulner- Confidence bounds of NPV can then be calculated if input
ability, where indirect losses start to accumulate for damages that parameters are random variables. The probability that an adaptation
exceed 20%, and total losses are twice the direct losses for a cata- measure is cost-effective, denoted herein as PrðNPV > 0Þ, may
strophic event where PrðDjHÞ ¼ 100%. also be inferred. The previously given equations can be generalized

© ASCE 04014001-5 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
$

Max. NPV
Benefit
Optimal
Adaptation Adaptation
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

Cost

0% 100%
Risk Reduction (%)

Fig. 6. Schematic of net present value (NPV) showing optimal


adaptation

for any time period, discounting of future costs and more detailed
spatial and time-dependent cost and damage consequences. Fig. 7. Typical Australian truss roof and timber frame houses (images
The vulnerability, loss, and adaptation costs are subject to con- by Mark G. Stewart)
siderable uncertainty due to lack of available data and models. For
this reason, calculations of risks, costs, and benefits will be impre-
cise. Hence, a break-even analysis may be useful where minimum
risk reduction or maximum cost of adaptation necessary for adap-
tation to be cost-effective is selected such that there is 50% prob- Wind Loads for Houses AS4055 (Standards Australia 2012) is based
ability that benefits equal cost—i.e., meanðNPVÞ ¼ 0. In other on AS/NZS1170.2 (Standards Australia 2011) and is used to deter-
words, if the actual cost of adaptation exceeds the predicted mine the appropriate wind classification for design of residential
break-even value, then adaptation is not cost-effective. Decision- (domestic) housing. In this case, residential housing is designed
makers can then judge whether an adaptation strategy meets these to resist wind speeds with annual probability of exceedance of 1
break-even values. in 500. The standard AS4055 (Standards Australia 2012) classifies
Governments and their regulatory agencies normally exhibit design loads on houses into categories N1–N6 for noncyclonic
risk-neutral attitudes in their decision-making, as described by regions. Each increase in noncyclonic wind classification (e.g., N1
Eq. (4). This is confirmed by the U.S. Office of Management to N2) raises the design wind speed that is equivalent to at least a
and Budget (OMB) (OMB 1992), and also by many practitioners 50% increase in design wind pressure. These wind classifications are
and researchers (e.g., Sunstein 2002; Faber and Stewart 2003; then used by building codes to determine appropriate deemed-to-
Ellingwood 2006). This entails using mean or average estimates comply sizing and detailing requirements for residential construction.
for risk and cost-benefit calculations, and not worst-case or pessi- To reduce housing damage in the future, one option may be to
mistic estimates. Probability neglect is a form of risk aversion as strengthen or retrofit existing construction. However, Stewart and
decision-makers are clearly averse to events of large magnitude Wang (2011) found such strategies often failed to be cost-effective,
irrespective of the probability of it actually occurring. Utility theory and if cost efficient, then only marginally so. Other adaptation strat-
can be used if the decision maker wishes to explicitly factor risk egies may restrict construction of new housing in vulnerable
aversion or proneness into the decision process (e.g., Stewart (exposed) locations. A more feasible adaptation strategy may be
et al. 2011a). one that increases design wind loads for new houses leading to long-
term reduction of vulnerability (and damages) of houses (Stewart
et al. 2013; Stewart 2014b). It is important to note that AS4055
Case Study: Strengthening New Houses in Sydney
(Standards Australia 2012) (as well as many other Australian
against Extreme Wind
and international building standards) is based on limited experi-
Severe storms have caused annual insured losses (1967–2005) of mental and field data and expert judgement by committee mem-
nearly $300 million in the Australian states of Queensland, New bers, and has not been subject to risk or cost-benefit analyses
South Wales, and Victoria (BITRE 2008). These losses account (Walker and Musulin 2012). Hence, existing design requirements
for nearly 25% of all losses from natural disasters in Australia. may be suboptimal even for the current climate.
Southern Australian contemporary housing generally comprises The case study herein applies break-even (risk neutral) and
a detached dwelling on a 600–800 m2 block, one or two storeys risk-averse analyses to compare the risks, costs, and benefits of
high, timber construction with brick veneer cladding, and a tiled climate adaptation strategies for new housing in the Australian
or metal sheeting roof (Fig. 7). The wind vulnerability of this hous- coastal city of Sydney. Sydney is the largest city in Australia with
ing type will not be too dissimilar to that for housing in the United a population of nearly 5 million (about 20% of the total population
States, Canada, and New Zealand. of Australia). Sydney is located in South-East Australia where
The Australian Standard for wind loads [AS/NZS1170.2 wind hazard is dominated by noncyclonic winds (thunderstorms
(Standards Australia 2011)] is the reference standard for design and east-coast lows). For more details of this case study, see
of all structures, including housing. The Australian Standard Stewart (2013).

© ASCE 04014001-6 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
Wind Hazard Modelling
The annual probability of winds PrðHjCÞ is derived from the Gum-
bel distribution to model the annual probability of exceedance of
noncyclonic winds (winds not associated with tropical cyclones
that occur in northern Australia) (Wang et al. 2013). Unless noted
otherwise, wind speed is the gust wind speed at 10 m in terrain
category 2. Climate projections by the Australian Commonwealth
Scientific and Industrial Research Organisation (CSIRO) suggest
that for noncyclonic winds, the mean wind speed may increase
by up to 20% by the year 2070 along the east coast of Australia
(CSIRO 2007). If the relationship between mean wind speed
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

and peak gust wind speed is constant, then proportional increases


in gust wind and mean wind speeds are identical. Since there are
still many uncertainties to properly define the future trend of
extreme winds in Australia, three possible climate scenarios (C) Fig. 8. Probability distribution of change in wind speeds to year 2070
are considered: (1) no change, (2) B1, and (3) A1FI emission
scenarios. The variability of current peak wind loads is significant
with COV of up to 50%.
CSIRO (2007) suggest the average annual change in mean wind
speed is projected to decrease by 1% in Sydney by the year 2070,
100
with 10 and 90% of −15 and þ12%, respectively, for the A1FI
(high) emission scenario, and 10th and 90th percentiles of −8 to 90

þ6% for the B1 (medium) emission scenario (Table 1). Projected 80

Vulnerability Pr(D|H) (%)


N1 N2 N3 N4
changes in wind speeds for Brisbane are higher than for Sydney. 70
Note that climate projections are relative to levels in the year 1990. 60
Truncated normal distributions are used to represent uncertainty of
50
changes in wind speeds where 10 and 90% bounds provided by
CSIRO (2007) allow the standard deviation of the two truncated 40

normal distributions each with cumulative probabilities of 50% 30


to be calculated, Fig. 8. A time-dependent linear change in wind 20
speed for all emission scenarios is assumed. 10
0
20 30 40 50 60 70 80 90 100
Wind Vulnerability and Loss Function Peak Gust Wind Speed (m/s)

Australian contemporary housing in South-East Australia is typical Fig. 9. Wind vulnerability curves for Sydney, for contemporary resi-
of that shown in Fig. 7. The housing is weaker than those designed dential housing in South-East Australia
for cyclonic (northern) regions of Australia where Australian
Standards dictate more stringent design requirements. A wind vul-
nerability function expresses building damage or loss as a function
of wind speed. Vulnerability curves developed by Geoscience Adaptation Strategy: Strengthen New Housing
Australia and James Cook University (Wehner et al. 2010) The adaptation strategy considered herein is to design new houses
representative of contemporary housing in Sydney are shown in by enhanced design codes, in this case, increasing the current
Fig. 9. These vulnerability curves are quite uncertain, but provide Australian Standard AS4055 (Standards Australia 2012) wind clas-
a useful starting point. sification by one category (Table 2). For example, for Sydney this
The loss function includes direct and indirect losses, and is means that new construction would be designed for wind classifi-
shown in Fig. 5. House replacement value is clearly variable cation N2 rather than the current requirement of N1 for nonfore-
and depends on the location, type, size, age, and condition shore locations. For example, this would mean that the number of
of the house. The average insured value of a house and its nails for a roof batten to roof truss connection should increase from
contents is approximately 25% higher than house replace-
ment value. The loss L is equal to insured value of the house
and its contents, normalised to L ¼ 1.25 of house replacement
Table 2. Adaptation Measure Showing Proposed Increase in Wind
value.
Classification
Adaptation measure:
Existing specifications proposed increase
Table 1. Change in Wind Speeds to Year 2070
(AS4055-2012) in wind classifications
B1 emission scenario A1FI emission scenario
Design Design
10th Mean 90th 10th Mean 90th Wind gust wind Wind gust wind
Location (%) (%) (%) (%) (%) (%) Location classification speed (m=s) classification speed (m=s)
Brisbane −1 þ3 þ10 −2 þ6 þ19 Sydney — — — —
Sydney −8 0 þ6 −15 −1 þ12 Foreshore N2 40 N3 50
Melbourne −9 −1 þ6 −18 −1 þ12 Nonforeshore N1 34 N2 40

© ASCE 04014001-7 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
100
percentage change in risk caused by the adaptation strategy
Reduction in Vulnerability (%) 90 is ΔR ¼ 50 − 65%.
80 The NPV for a single house built to enhanced standards at time
70
tadapt for a given climate change scenario Cs (no change, B1, or
N1 to N2
A1FI) is
60
50
N2 to N3
XT
ΔR × EðtÞ Cadapt
NPVðTjCs Þ ¼ t−2018
− ð5Þ
40
t¼tadapt ð1 þ rÞ ð1 þ rÞtadapt −2018
30
20 where NPV is expressed as percentage of replacement value of the
10 house; Cadapt = cost of the adaptation strategy expressed as percent-
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

age of house replacement value; EðtÞ = damage risk per house


0
0 10 20 30 40 50 60 70 80 90 100 110 120
associated with current wind classification (business as usual risk);
Peak Gust Wind Speed (m/s) ΔR = reduction in risk associated with increasing the current wind
classification by one category (e.g., from N2 to N3); and r =
Fig. 10. Reduction in vulnerability for new housing with increase of discount rate. Cobenefits are assumed as ΔB ¼ 0.
wind classification, for Sydney

Results: Scenario-Based Analysis


one plain shank nail to two plain shank nails, or a single deformed The models described herein are the best available models, but as
shank nail should replace a single plain shank nail [AS 1684.4 described previously, have their limitations and uncertainties. For
(Standards Australia 2010)]. This means that new construction this reason, a break-even analysis is conducted.
has increased strengths of structural components and connections, Any proposal to change building regulation within the Building
leading houses to have significantly reduced wind vulnerability. Code of Australia would take some time. Hence, we assume the
These enhanced building requirements will result in additional earliest time of adaptation is by the year 2018. Results are calcu-
costs of new construction (Cadapt ) of 1–2% of the value of a house, lated using Monte-Carlo event-based simulation methods. Costs
or approximately $2,500 to $5,000 (AGO 2007). and benefits are calculated for the 52 year period 2018–2070 as
Fig. 10 shows percentage reduction in vulnerability as a function 2070 is the limit of projections of wind hazard provided by CSIRO
of wind speed for Sydney. It is evident that designing new houses to (2007). Costs are in 2012 Australian dollars and the discount rate is
enhanced wind classification will reduce vulnerability often by 4%. The stochastic variability of wind speed means that NPV is
more than 50% for Sydney. The reduction in vulnerability reduces variable. The probability distribution of NPV is highly nonGaussian
only for very high wind speeds where damages asymptote to 100% and so Monte-Carlo methods are well suited to this type of analysis.
resulting in reduced relative reductions in vulnerability. Vulnerabil- In this scenario-based approach, PrðCs Þ ¼ 100%.
ity reduction is higher for nonforeshore locations due to lower wind Figs. 11 and 12 show the maximum adaptation cost Cadapt for the
speeds than foreshore locations. Similarly, vulnerability reduction adaptation measure (per new house) to be cost-effective for risk
will decrease as climate change becomes more severe and wind reductions of 10–100% for foreshore and nonforeshore locations
speeds increase. The overall reduction in risk calculated as in Sydney. These figures show

20% 20%

No change No change
B1 B1
15% A1FI 15% A1FI
Break-Even Adaptation Cost

Break-Even Adaptation Cost

10% 10%

5% 5%

0% 0%
10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(a) Risk Reduction R (b) Risk Reduction R

Fig. 11. (a) Break-even adaptation costs for foreshore locations in Sydney; (b) break-even adaptation costs for nonforeshore locations in Sydney

© ASCE 04014001-8 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
20% 20%

No change No change
B1 B1
15% A1FI 15% A1FI

Maximum Adaptation Cost


Maximum Adaptation Cost

for Pr(NPV>0)=90%
for Pr(NPV>0)=90%

10% 10%
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

5% 5%

0% 0%
10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(a) Risk Reduction R (b) Risk Reduction R

Fig. 12. (a) Maximum adaptation costs to ensure PrðNPV > 0Þ ¼ 90% for foreshore locations in Sydney; (b) maximum adaptation costs to ensure
PrðNPV > 0Þ ¼ 90% for nonforeshore locations in Sydney

1. The break-even cost of adaptation [i.e., maximum cost for (B1) emission scenario, minimum risk reduction must exceed 11
meanðNPVÞ ¼ 0, risk neutral]; and and 17% for foreshore and nonforeshore locations, respectively.
2. The maximum cost of adaptation to ensure that there is 90% Given that the previous section shows that risk reductions of
surety that benefits exceed the cost (i.e., a risk-averse decision 50–65% can be achieved for Sydney based on the vulnerability
maker may prefer a small likelihood of a net loss). models described herein, then it is likely that designing new hous-
Fig. 11 shows that if risk reduction is over 50% and there is no ing to enhance wind classifications is a cost-effective adaptation
change of climate, the break-even analysis shows that adaptation is strategy for Sydney.
cost-effective if the adaptation cost is less than 9.3 and 5.5% of Clearly, more research is needed to improve the confidence of
house replacement cost for foreshore and nonforeshore locations, wind hazard and vulnerability modeling. Nonetheless, preliminary
respectively. The effect of a changing climate on break-even adap- results show that vulnerability reduced at a modest cost can lead to
tation costs is negligible, as this will increase the break-even adap- a cost-effective adaptation measure.
tation cost to 9.7 and 5.7% for foreshore and nonforeshore
locations, for the A1FI emission scenario. Hence, even if climate
projections are wrong, adaptation measures still satisfies a no re- Discount Rates
grets or win-win policy (Susskind 2010). There is some uncertainty about the level of discount rate, particu-
On the other hand, the maximum cost of adaptation to ensure larly for climate change economic assessments [e.g., Dasgupta
that there is 90% surety that benefits exceed the cost will be less (2008), for more details see Stewart (2013)]. A high discount rate
than the break-even costs (Fig. 12). In this case, the adaptation is reduces the cost-effectiveness of adaptation strategies because the
preferable if the risk reduction exceeds 50% and the adaptation cost benefit of reduction in damages into the future is reduced, thus
is less than 6.8 and 4.0% of house replacement cost for foreshore reducing NPV and lowering the break-even adaptation costs. None-
and nonforeshore locations, respectively, and assuming no change theless, a 7 or 10% discount rate will still produce break-even adap-
in climate. As there is significant uncertainty associated with wind tation costs for Sydney of 3–6%, which are likely to be higher than
hazard projections for B1 and A1FI emission scenarios, the vari- actual adaptation costs. Discount rates of 1.35 and 2.65% as used
ability of NPV is higher for these scenarios. Hence, the maximum by Garnaut (2008) result in higher break-even values (7–15%),
cost of adaptation to ensure that there is 90% surety that benefits which increase the likelihood of adaptation being cost-effective.
exceed the cost will be up to 3% lower than for no climate change. These relatively low discount rates were selected so as to not under-
This suggests that even a risk averse decision-maker would adopt estimate climate impacts on future generations. However, others
the adaptation measure since the anticipated cost of adaptation is suggest higher discount rates when assessing economic impacts
very low (1–2%) and risk reduction exceeds 50%. of climate change (e.g., Nordhaus 2007).
The cost of adaptation for Sydney is likely to be 1.1% for
foreshore locations (N2 to N3), and less for nonforeshore locations
(N1 to N2). If we adopt a cost of adaptation of 1.1 and 1.0% for Time of Adaptation
these locations, and no change of climate, then the break-even risk The effects of a changing climate tend to worsen into the future so
reduction must exceed 6 and 9% for foreshore and nonforeshore benefits of adaptation in the next decade or so are lower compared
locations, respectively, to be 50% certain that NPV > 0 for all to those later in the century. Hence, deferring time of adaptation is
climate projections. The minimum risk reductions increase to an option worth considering. There is also the benefit of reduced
ensure 90% certainty that NPV > 0. For example, for the medium present (discounted) value of adaptation cost if this cost is deferred.

© ASCE 04014001-9 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
Stewart et al. (2013) and Stewart (2014b) have shown that while Nonetheless, it illustrates how this may affect uncertainty and vari-
adaptation that is implemented as early as possible has the highest ability of NPV, and its effect on decision-making.
NPV, deferred adaptation of 5–20 years also yields a high NPV and
high likelihood that adaptation is cost-effective.
Further Work
Results: Likelihood of Climate Scenarios This paper highlights that a risk-based approach to optimizing
adaptation requires the following information:
The previous section described results for a scenario-based analysis 1. Effect of climate scenarios on frequency and intensity of
by assuming that the probability of climate scenario PrðCs Þ is hazards;
100%. However, there is unlikely to be such certainty about any 2. Vulnerability of infrastructure to hazards;
climate scenario. The IPCC Special Report on Emissions Scenarios
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

3. Loss functions;
(SRES) states that probabilities or likelihood are not assigned to 4. Risk reduction for adaptation measures; and
individual SRES scenarios (IPCC 2000). One approach is to esti- 5. Cost of adaptation measures.
mate the relative likelihood of the occurrence of each climate sce- The break-even approach to economic assessment of the costs
nario, such as by expert opinion, Bayesian probability updating and benefits of adaptation is used due to considerable modeling and
when new data/models are available, etc. The overall NPV at time parameter uncertainty of these variables.
T is calculated as More accurate predictions of hazard, vulnerability, risk reduc-
tion, and economic assessment are challenging. A key challenge,
X
S at least for engineers, is the development of vulnerability models
NPVðTÞ ¼ PrðCs ÞNPVðTjCs Þ ð6Þ for damage prediction. This will utilize reliability and probabilistic
s¼1 modeling in time and space to model damage initiation and pro-
gression. There is much work on predicting reliabilities for the ul-
where S = total number of climate scenarios considered timate limit state where life-safety is the major criterion. However,
(S ¼ 3); Cs ¼ sth climate scenario; and NPVðTjCs Þ is given modeling of damage and serviceability limit states is a less tractable
by Eq. (5). problem as this requires information about damage progression,
IPCC (2007) states that anthropogenic forcing is likely to have load sharing, and redistribution of loads of failed components or
contributed to changes in wind patterns. The term likely is associ- connections, and other spatial and time-dependent processes. There
ated with a probability that exceeds 67%. It follows that the is also difficulty in modeling the vulnerability of housing and other
likelihood of no change in wind patterns is 33%, leading to nonengineered forms of construction where there is high variability
Prðno changeÞ ¼ 33%. If emissions scenarios B1 and A1FI are of construction materials, processes, and structural systems, high
considered equally likely, then PrðB1Þ ¼ PrðA1FIÞ ¼ 33%. variability of connection capacities (e.g., nailed connections, roof
For sake of illustration, the effect of climate scenario likelihood sheeting fasteners), less code compliance, and other variabilities
is assessed for nonforeshore locations in Sydney and 50% risk and uncertainties that are less evident in reinforced concrete, steel
reduction, as shown in Table 3. As expected, climate scenario like- and other forms of engineered construction.
lihood has little effect on the break-even adaptation cost as this
metric is less sensitive to the climate scenario. However, the maxi-
mum adaptation cost to be 90% certain that adaptation is cost-
effective is sensitive to the climate scenario. In this case, a weighted
Conclusions
average of NPV given by Eq. (6) reduces variability of NPV, The performance of new and existing infrastructure will degrade if
resulting in higher 10th percentiles of NPV and so higher maxi- subject to more extreme climate-related hazards or accelerated cli-
mum adaptation costs. For example, if there is 100% certainty mate-change induced degradation of material properties. Climate
of no change, the maximum adaptation cost is 4.0% for adapta- adaptation engineering involves estimating the risks, costs, and
tion to be preferred, however, this falls to 3.5% if there is equal benefits of climate adaptation strategies and assessing at what point
likelihood of all scenarios, and reduces further to 2.6% if PrðB1Þ ¼ in time climate adaptation becomes economically viable. This
PrðA1FIÞ ¼ 50%. paper has described how risk-based approaches are well suited
Eq. (6) and Table 3 are clearly a simplification of what is a chal- to optimizing climate adaptation strategies for built infrastructure.
lenging issue of degree of belief in climate scenarios and how it The concepts were illustrated with a state-of-the-art application of
might influence the cost-effectiveness of adaptation policy options.
risk-based assessment of climate adaptation strategies for Austral-
ian housing subject to extreme wind events. It was found that wind
vulnerability of new housing in Sydney can be reduced by 50–65%
Table 3. Effect of Climate Scenario Likelihood on Maximum Cost of at modest cost, and can be shown to be a cost-effective adaptation
Adaptation, for 50% Risk Reduction and Nonforeshore Exposure in measure.
Sydney
Probability of Maximum adaptation costs for
climate scenario adaptation to be preferred option
Acknowledgments
Pr (no Pr Pr Mean PrðNPV > 0Þ
change) (%) (B1) (%) (A1FI) (%) (NPV > 0) (%) ¼ 90% (%) The authors appreciate the financial support of the Common-
100 0 0 5.5 4.0 wealth Scientific and Industrial Research Organisation (CSIRO)
0 100 0 5.5 3.0 Flagship Cluster Fund through the project Climate Adaptation
0 0 100 5.7 1.9 Engineering for Extreme Events in collaboration with the Sustain-
33.3 33.3 33.3 5.6 3.5 able Cities and Coasts Theme, the CSIRO Climate Adaptation
0 50 50 5.6 2.6
Flagship.

© ASCE 04014001-10 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
References 16th Australasian Wind Engineering Society Workshop, Brisbane,
Australia, 1–10.
Australian Greenhouse Office (AGO). (2007). “An assessment of the need Hinkel, J., Nicholls, R. J., Vafeidis, A. T., Tol, R. S. J., and Avagianou, T.
to adapt buildings for the unavoidable consequences of climate change.” (2010). “Assessing risk of and adaptation to sea-level rise in the
Final Rep., Commonwealth of Australia, Canberra, Australia. European Union: An application of DIVA.” Mitigation Adaptation
Bastidas-Arteaga, E., Chateauneuf, A., Sánchez-Silva, M., Bressolette, P. H., Strategies Global Change, 15(7), 703–719.
and Schoefs, F. (2010). “Influence of weather and global warming in Holden, R., Val, D. V., Burkhard, R., and Nodwell, S. (2013). “A network
chloride ingress into concrete: A stochastic approach.” Struct. Saf., flow model for interdependent infrastructures at the local scale.” Saf.
32(4), 238–249. Sci., 53(3), 51–60.
Bastidas-Arteaga, E., Schoefs, F., Stewart, M. G., and Wang, X. (2013). Idris, N. H., Deng, X., and Andersen, O. B. (2014). “The importance of
“Influence of global warming on durability of corroding RC structures: coastal altimetry retracking and detiding: A case study around the Great
A probabilistic approach.” Eng. Struct., 51, 259–266. Barrier Reef, Australia.” Int. J. Remote Sens., 35(5), 1729–1740.
Bastidas-Arteaga, E., and Stewart, M. G. (2013). “Probabilistic cost-benefit Intergovernmental Panel on Climate Change (IPCC). (2000). “Emission
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

analysis of climate change adaptation strategies for new RC structures scenarios.” Special Rep. of the Intergovernmental Panel on Climate
exposed to chloride ingress.” 11th Int. Conf. on Structural Safety and Change, Cambridge University Press, Cambridge, U.K.
Reliability, CRC Press, 1503–1510. Intergovernmental Panel on Climate Change (IPCC). (2007). “Climate
Bjarnadottir, S., Li, Y., and Stewart, M. G. (2011a). “A probabilistic-based change 2007: Synthesis report.” Contribution of Working Groups I,
framework for impact and adaptation assessment of climate change on II and III to the Fourth Assessment Rep. on Intergovernmental
hurricane damage risks and costs.” Struct. Saf., 33(3), 173–185. Panel on Climate Change, R. K. Pachauari and A. Reisinger, eds., Core
Bjarnadottir, S., Li, Y., and Stewart, M. G. (2011b). “Social vulnerability writing team, Geneva, Switzerland.
index for coastal communities at risk to hurricane hazard and a changing Intergovernmental Panel on Climate Change (IPCC). (2012). “Managing
climate.” Nat. Hazards, 59(2), 1055–1075. the risks of extreme events and disasters to advance climate change
Botzen, W. J. W., Alerts, J. C. J. H., and van den Bergh, J. C. J. M. (2013). adaptation.” A Special Rep. of Working Groups I and II of the Intergov-
“Individual preferences for reducing flood risk to near zero through ernmental Panel on Climate Change, C. B. Field, et al., eds.,
elevation.” Mitigation Adaptation Strategies Global Change, 18(2), Cambridge University Press, Cambridge, U.K.
229–244. Intergovernmental Panel on Climate Change (IPCC). (2014). “Climate
Bureau of Infrastructure, Transport, and Regional Economics (BITRE). change 2014: Impacts, adaptation, and vulnerability.” Technical Sum-
(2008). About Australia’s regions, Canberra, Australia. mary, Draft.
Bureau of Transport Economics (BTE). (2001). “Economic costs of natural Kundzewicz, Z. W., et al. (2013). “Assessing river flood risk and adaptation
disasters in Australia.” Bureau of Transport Economics Rep. 103, in Europe—Review of projections for the future.” Mitigation Adapta-
Canberra, Australia. tion Strategies Global Change, 15(7), 641–656.
Cheng, Y., Andersen, O. B., and Knudsen, P. (2012). “Integrating non-tidal Li, Y., and Stewart, M. G. (2011). “Cyclone damage risks caused by
sea level data from altimetry and tide gauges for coastal sea level
enhanced greenhouse conditions and economic viability of strength-
prediction.” Adv. Space Res., 50(8), 1099–1106.
ened residential construction.” Nat. Hazards Rev., 10.1061/(ASCE)
Commonwealth Scientific and Industrial Research Organisation (CSIRO).
NH.1527-6996.0000024, 9–18.
(2007). Climate change in Australia: Technical Rep. 2007, Marine and
Lomborg, B. (2006). “Stern review: The dodgy numbers behind the latest
Atmospheric Research Division, Canberra, Australia.
warming scare.” Wall Street J., 2, in press.
Crompton, R. P., and McAneney, J. K. (2008). “Normalised australian
Lomborg, B. (2009). Global crises, global solutions, Cambridge University
insured losses from meteorological hazards: 1967–2006.” Environ.
Press, Cambridge, U.K.
Sci. Policy, 11(5), 371–378.
Mason, M., Phillips, E., Okada, T., and O’Brien, J. (2012). Analysis of
Dasgupta, P. (2008). “Discounting climate change.” J. Risk Uncertainty,
damage to buildings following the 2010/2011 east Australian floods,
37(2–3), 141–169.
NCCARF, Griffith Univ., Brisbane, Australia.
Deng, X., et al. (2011). “Satellite altimetry for geodetic, oceanographic, and
climate studies in the Australian region.” Coastal altimetry, Springer, Mendelsohn, R. O. (2006). “A critique of the stern report.” Regulation,
Heidelberg, Germany, 473–508. 29(4), 42–46.
Deng, X., Andersen, O. B., Gharineiat, Z., and Stewart, M. G. (2013). Mueller, J., and Stewart, M. G. (2011a). Terror, security, and money:
“Observing and modelling the high sea level from satellite radar altim- Balancing the risks, benefits, and costs of homeland security, Oxford
etry during tropical cyclones.” Int. Association of Geodesy Symp., University Press, Oxford, U.K.
Potsdam, Germany. Mueller, J., and Stewart, M. G. (2011b). “The price is not right: The U.S.
Ellingwood, B. R. (2006). “Mitigating risk from abnormal loads and spends too much money to fight terrorism.” Playboy, 58(10), 149–150.
progressive collapse.” J. Perform. Constr. Facil., 10.1061/(ASCE) National Academy of Sciences (NAS). (1999). The impact of natural
0887-3828(2006)20:4(315), 315–323. disasters: A framework for loss estimation, Washington, DC.
Faber, M. H., and Stewart, M. G. (2003). “Risk assessment for civil engi- Nguyen, M. N., Wang, X., and Leicester, R. H. (2013). “An assessment
neering facilities: Critical overview and discussion.” Reliab. Eng. Syst. of climate change effects on atmospheric corrosion rates of steel
Saf., 80(2), 173–184. structures.” Corros. Eng. Sci. Technol., 48(5), 359–369.
Garnaut, R. (2008). “The Garnaut climate change review.” Final Rep., Nishijima, K., Maruyama, T., and Graf, M. (2012). “A preliminary impact
Commonwealth of Australia, Cambridge University Press, Cambridge, assessment of typhoon wind risk of residential buildings in Japan under
U.K. future climate change.” Hydrol. Res. Lett., 6(1), 23–28.
Goklany, I. M. (2008). “What to do about climate change.” Policy Analysis, Nordhaus, W. D. (2007). “A review of the stern review on the economics of
No. 609, Cato Institute, Washington, DC. climate change.” J. Econ. Lit., 45(3), 686–702.
Haigh, I. D., Nicholls, R., and Wells, N. (2010). “A comparison of the main Office of Management, and Budget (OMB). (1992). “Guidelines and
methods for estimating probabilities of extreme still water levels.” discount rates for benefit-cost analysis of federal programs (revised).”
Coastal Eng., 57(9), 838–849. Circular No. A-94, Washington, DC.
Hall, J. W., Brown, S., Nicholls, R. J., Pidgeon, N. F., and Watson, R. T. Peng, L., and Stewart, M. G. (2014). “Climate change and corrosion
(2012). “Proportionate adaptation.” Nat. Clim. Change, 2(12), 833–834. damage risks for reinforced concrete infrastructure in China.” Struct.
Hallegatte, S. (2008). “An adaptive regional input-output model and its Infrastruct. Eng.
application to the assessment of the economic cost of Katrina.” Risk Peters, G. P., et al. (2013). “The challenge to keep global warming below
Anal., 28(3), 779–799. 2°C.” Nat. Clim. Change, 3(1), 4–6.
Harper, B. A. (2013). “Best practice in tropical cyclone wind hazard Standards Australia. (2010). “Residential timber-framed construction—
modelling: In search of data and emptying the skeleton cupboard.” Part 4: Simplified—Non-cyclonic areas.” AS1684.4, Sydney, Australia.

© ASCE 04014001-11 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001
Standards Australia. (2011). “Structural design actions—Part 2: Wind Stewart, M. G., Wang, X., and Willgoose, G. R. (2013). “Direct and indirect
actions.” AS/NZS 1170.2, Sydney, Australia. cost and benefit assessment of climate adaptation strategies for housing
Standards Australia. (2012). “Wind loads for houses.” AS4055, Sydney, for extreme wind events in Queensland.” Nat. Hazards Rev., 10.1061/
Australia. (ASCE)NH.1527-6996.0000136, 04014008.
Standards Australia. (2013). “Climate change adaptation for settlements and Sunstein, C. R. (2002). The cost-benefit state: The future of regulatory
infrastructure—A risk based approach.” AS5334, Sydney, Australia. protection, American Bar Association Publishing, Chicago.
Stern, N. (2007). The economics of climate change: The Stern review, Susskind, L. (2010). “Responding to the risks posed by climate change:
Cambridge University Press, Cambridge, U.K. Cities have no choice but to adapt.” Town Plann. Rev., 81(3), 217–235.
Stewart, M. G. (2013). “Risk and economic analysis of residential housing Val, D. V., Holden, R., and Nodwell, S. (2013). “Probabilistic assessment of
climate adaptation strategies for wind hazards in south-east Australia.” failures of interdependent infrastructures due to weather related
CAEx Rep. 1/2013, Climate Adaptation Engineering for Extreme Events hazards.” Safety, reliability, risk and life-cycle performance of struc-
Cluster, CSIRO Climate Adaptation Flagship, Canberra, Australia. tures and infrastructure, G. Deodatis, B. R. Ellingwood, and D. M.
Stewart, M. G. (2014a). “Climate change and national security: Balancing Frangopol, eds., Taylor and Francis, London, 1551–1557.
Downloaded from ascelibrary.org by Universidad Militar Nueva Granada on 02/02/20. Copyright ASCE. For personal use only; all rights reserved.

the costs and benefits.” Dangerous world? Threat perception and U.S. Walker, G. R. (2011). “Comparison of the impacts of cyclone tracy and the
national security, C. Preble and J. Mueller, eds., Cato Institute, newcastle earthquake on the Australian building and insurance indus-
Washington, DC. tries.” Aust. J. Struct. Eng., 11(3), 283–293.
Stewart, M. G. (2014b). “Risk and economic viability of housing climate Walker, G. R., and Musulin, R. (2012). “Utilising catastrophic risk mod-
adaptation strategies for wind hazards in southeast Australia.” Mitiga- elling for cost benefit analysis of structural engineering code changes.”
tion adaptation strategies global change, in press. Australasian Structural Engineering Conf., Engineers Australia, 92–99.
Stewart, M. G., Ellingwood, B. R., and Mueller, J. (2011a). “Homeland
Wang, C.-H., and Wang, X. (2012). “Vulnerability of timber in ground con-
security: A case study in risk aversion for public decision-making.”
tact to fungal decay under climate change.” Clim. Change, 115(3–4),
Int. J. Risk Assess. Manage., 15(5–6), 367–386.
777–794.
Stewart, M. G., and Melchers, R. E. (1997). Probabilistic risk assessment of
Wang, C.-H., Wang, X., and Khoo, Y. B. (2013). “Extreme wind gust
engineering systems, Chapman and Hall, London.
hazard in Australia and its sensitivity to climate change.” Nat. Hazards,
Stewart, M. G., and Peng, J. (2010). “Life cycle cost assessment of climate
67(2), 549–567.
change adaptation measures to minimise carbonation-induced corrosion
risks.” Int. J. Eng. Uncertainty: Hazards, Assess. Mitigation, 2(1–2), Wang, X., Stewart, M. G., and Nguyen, M. (2012). “Impact of climate
35–46. change on corrosion and damage to concrete infrastructure in
Stewart, M. G., Val, D., Bastidas-Arteaga, E., O’Connor, A., and Wang, X. Australia.” J. Clim. Change, 110(3–4), 941–957.
(2014). Climate adaptation engineering and risk-based design and Wehner, M., Ginger, J., Holmes, J., Sandland, C., and Edwards, M. (2010).
management of infrastructure, maintenance and safety of aging “Development of methods for assessing the vulnerability of Australian
infrastructure, D. M. Frangopol, and Y. Tsompanakis, eds., Taylor residential building stock to severe wind.” IOP Conf. Ser. Earth Envi-
and Francis, London. ron. Sci., 11(1), 12–17.
Stewart, M. G., and Wang, X. (2011). Risk assessment of climate adapta- Wigley, T. M. L., Richels, R., and Edmonds, J. A. (1996). “Economic
tion strategies for extreme wind events in Queensland, CSIRO Climate and environmental choices in the stabilization of atmospheric CO2
Adaptation Flagship, Canberra, Australia. concentrations.” Nature, 379(6562), 240–243.
Stewart, M. G., Wang, X., and Nguyen, M. (2011b). “Climate change World Bank. (2010). “Economics of adaptation to climate change: Synthe-
impact and risks of concrete infrastructure deterioration.” Eng. Struct., sis report.” International Bank for Reconstruction and Development/
33(4), 1326–1337. World Bank, Washington, DC.
Stewart, M. G., Wang, X., and Nguyen, M. (2012a). “Climate change Yohe, G. W., Tol, R. S. J., Richels, R. G., and Blanford, G. F. (2009).
adaptation for corrosion control of concrete infrastructure.” Struct. “Climate change.” Global crises, global solutions, B. Lomborg, ed.,
Saf., 35(3), 29–39. Cambridge University Press, Cambridge, U.K.
Stewart, M. G., Wang, X., and Willgoose, G. R. (2012b). Indirect cost and Zhang, H., and Sheng, J. (2013). “Estimation of extreme sea levels over the
benefit assessment of climate adaptation strategies for extreme wind eastern continental shelf of North America.” J. Geophys. Res. Oceans,
events in Queensland, CSIRO, Canberra, Australia. 118(11), 6253–6273.

© ASCE 04014001-12 ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng.

ASCE-ASME J. Risk Uncertainty Eng. Syst., Part A: Civ. Eng., 2015, 1(1): 04014001

You might also like