effect of climate change on corrosion rates of structures in australia

14
Effect of climate change on corrosion rates of structures in Australia Nayruti Siddharth Trivedi & Murali Sankar Venkatraman & Clement Chu & Ivan S. Cole Received: 24 June 2013 /Accepted: 27 February 2014 /Published online: 15 March 2014 # Springer Science+Business Media Dordrecht 2014 Abstract As structures built now will be expected to last well past 2064 (50 years) it is vital that the effect of climate change be considered in their design and material selection. In particular changes in the rate of corrosion of metal components must be considered. To this end this study estimates the maximum likely change in the corrosion rate for the year 2070 so it can be included in current design. Changes in corrosion are estimated for 11 coastal and inland locations in Australia. For each station the climatic data (3-hourly) in 2070 is estimated by modifying current data with probable changes based on two climate change models (CSIRO: CSIRO-Mk 3.5 and MRI: MRI-CGCM 3.2.2). The former is for high global warming rate and the later the A1FI scenario. This climatic data is then run the Corrosion predictor(a multi-scale process model) to predict corrosion at each location. It is found that significant changes occur with corrosion in coastal locations increasing substantially, in contrast the corrosion at inland locations will decrease moderately. The increase in coastal locations is associated with a greater build up of salt due to less frequent rain evens while the reduction in inland locations is associated with a reduction in RH and thus surface wetness. 1 Introduction Significant changes in climate are expected by 2030 with average temperature estimated to rise by 1.0 °C and by 2.9 °C by 2070 (Alexander and Arblaster 2009; IPCC 2000, 2007; Meehl et al. 2007; CSIRO 2007). Most of the studies on climate change (Doney et al. 2012; Lindner et al. 2010; Jacob and Winner 2009) feature the direct effect of weather patterns and ecological status of the organisms. However, climate change greatly affects the building materials and reduces the sustainability of infrastructure and compromises the safety of the people. Corrosion in particular, involves severe risks associated with the safety and integrity of physical assets, risk to the environment, and financial risk from various decisions and also risks from poor corrosion mitigation procedures (Roberge 2010). Despite this fact, the studies on the impact of climate change on the corrosion and sustainability of the building materials are limited (Cole and Paterson 2010). Studies focus mainly on the effect of climate change on concrete structures and suggest that increase in Climatic Change (2014) 124:133146 DOI 10.1007/s10584-014-1099-y N. S. Trivedi : M. S. Venkatraman : C. Chu : I. S. Cole (*) CSIRO Materials Science and Engineering, CSIRO, Clayton South, Australia e-mail: [email protected]

Upload: ivan-s

Post on 20-Jan-2017

212 views

Category:

Documents


0 download

TRANSCRIPT

Effect of climate change on corrosion ratesof structures in Australia

Nayruti Siddharth Trivedi & Murali Sankar Venkatraman &

Clement Chu & Ivan S. Cole

Received: 24 June 2013 /Accepted: 27 February 2014 /Published online: 15 March 2014# Springer Science+Business Media Dordrecht 2014

Abstract As structures built now will be expected to last well past 2064 (50 years) it is vitalthat the effect of climate change be considered in their design and material selection. Inparticular changes in the rate of corrosion of metal components must be considered. To thisend this study estimates the maximum likely change in the corrosion rate for the year 2070 soit can be included in current design. Changes in corrosion are estimated for 11 coastal andinland locations in Australia. For each station the climatic data (3-hourly) in 2070 is estimatedby modifying current data with probable changes based on two climate change models(CSIRO: CSIRO-Mk 3.5 and MRI: MRI-CGCM 3.2.2). The former is for high globalwarming rate and the later the A1FI scenario. This climatic data is then run the Corrosion“predictor” (a multi-scale process model) to predict corrosion at each location. It is found thatsignificant changes occur with corrosion in coastal locations increasing substantially, incontrast the corrosion at inland locations will decrease moderately. The increase in coastallocations is associated with a greater build up of salt due to less frequent rain evens while thereduction in inland locations is associated with a reduction in RH and thus surface wetness.

1 Introduction

Significant changes in climate are expected by 2030 with average temperature estimated to riseby 1.0 °C and by 2.9 °C by 2070 (Alexander and Arblaster 2009; IPCC 2000, 2007; Meehlet al. 2007; CSIRO 2007). Most of the studies on climate change (Doney et al. 2012; Lindneret al. 2010; Jacob and Winner 2009) feature the direct effect of weather patterns and ecologicalstatus of the organisms. However, climate change greatly affects the building materials andreduces the sustainability of infrastructure and compromises the safety of the people.Corrosion in particular, involves severe risks associated with the safety and integrity ofphysical assets, risk to the environment, and financial risk from various decisions and alsorisks from poor corrosion mitigation procedures (Roberge 2010).

Despite this fact, the studies on the impact of climate change on the corrosion andsustainability of the building materials are limited (Cole and Paterson 2010). Studies focusmainly on the effect of climate change on concrete structures and suggest that increase in

Climatic Change (2014) 124:133–146DOI 10.1007/s10584-014-1099-y

N. S. Trivedi :M. S. Venkatraman : C. Chu : I. S. Cole (*)CSIRO Materials Science and Engineering, CSIRO, Clayton South, Australiae-mail: [email protected]

temperature and CO2 concentrations will lead to increase in carbonation rate of reinforcedconcrete (Talukdar et al. 2012; Stewart et al. 2011; Stewart and Peng 2010; Yoon et al. 2007).For example, Stewart et al. (2011) predict that by 2100, the damage due to carbonation forconcrete structures would rise by 400% for inland arid or temperate climates in Australia. Theyalso predict the chloride induced corrosion to be approximately 15% and the corrosion loss dueto reinforcement to be approximately 9.5 %. A summary of possible effects of climate changeon building materials (Nijland et al. 2009) for Netherlands concluded that the damage throughsalt, rising humidity and bio-deterioration, will intensify with changing climatic conditions.

Most studies on the effect of climate change on corrosion are based on the dose–responsefunction equation (Grøntoft 2011; Nijland et al. 2009), employing annual climatic data therebynot accounting for seasonal variability in the area. For example, a recent Norwegian study(Grøntoft 2011) shows that, increased (decreased) precipitation in a region in synchronizationwith increased (decreased) temperatures will lead to increase (decrease) in corrosion. Grøntoft(2011) used a dose-function equation with average weather data and took only the wetnessfrom rain as the major parameter for calculating the corrosion rate and did not take cleaning byrain into account. Although global studies comparing the observed and modeled trends forclimate change show reasonably good agreement with the temperature trends, they show pooragreement or multi-model disagreement with the observed precipitation patterns (Kharin et al.2007; Kiktev et al. 2007). Kiktev et al. (2007) also observed that the multi-model scenariosgive better synchronization with the precipitation pattern than any single model scenario.

To assess the impact of climate change on various weather parameters accurately, it is veryimportant to choose the appropriate climate model (Clarke et al. 2011). A common approach isto select a small number of ‘best’ climate models based on their ability to represent the currentclimatic conditions of the region (Smith and Chandler 2010; Pitman and Perkins 2008), usingRepresentative Climate Futures (RCF) framework, which has classified a number of climatemodels falls based on two major climate variables (usually annual mean temperature andrainfall) (Whetton et al. 2012). Detailed methodology for the application of the RCF frame-work and selection of a suitable model was demonstrated by Clarke et al. (2011).

Many of the effects of climate change, including change in temperature, increase in pollutantconcentrations, changes in RH, precipitation, wind patterns, and frequency of severe events(meteorological phenomena with the potential to cause significant damage, serious social disrup-tion, or loss of human life), could have significant impacts on infrastructure lifespan (Cole andPaterson 2010). Assessing the comprehensive impact of all the factors accurately is difficult, asthe relationship between infrastructure degradation and climate change is complex. Criticalclimate change parameters that affect the corrosion rates must be identified to develop suitableheuristics for constructing corrosion resistant infrastructure and preventing degradation of metals.

In this paper, we study the effect of climate change on corrosion rates of galvanized steelused in buildings (predominantly as roofing but also as structural members and fixings) fordifferent projected climate change scenarios using real meteorological data procured fromBureau of Meteorology, Australia (BoM) for the years 1995 to 2005. The future climatescenarios are predicted using climate models from CSIRO Marine and Atmospheric Research(CMAR), which are then coupled with our in-house corrosion estimator software (Cole et al.2011) to predict the corrosion rates (reported as annual mass-loss of metal in g.m−2.year−1) ofgalvanized roof sheeting. Our methodology is general and our software can estimate corrosionrates for other scenarios in Australia and elsewhere.

Here we predict the corrosion rates of galvanized steel for the year 2070 using suitableclimate models. Many studies (Grøntoft 2011; Nijland et al. 2009) have employed yearly ormonthly averages of weather data to predict future climate scenarios. While models based on“average” changes in climatic parameters are useful measure in many domains (e.g. global

134 Climatic Change (2014) 124:133–146

warming and glacial melting), they cannot readily incorporate some of the effects that controlatmospheric corrosion. For example atmospheric corrosion is highly influenced by the effect ofrain events in cleaning a surface of pollutants. This and other processes that control corrosionneed a knowledge of the possible changes to the full temporal variation in weather data (at asfine a resolution as possible and no less than three hourly climate data). Hence our study usesthe 3-hourly climate data from 11 different weather stations across Australia over a 12 monthperiod. By coupling the corrosion model to our weather maps prepared using GIS platform, wepredict the corrosion rates accounting for the variability in the weather patterns. For example,our study considers the intensity and frequency of rainfall, which are critical factors indetermining the rate of corrosion. Thus our study envisages the distribution and impact ofclimate change on corrosion rate of infrastructure across Australia.

2 Methodology

The study involves three stages to predict the corrosion rate in terms of annual mass-loss of themetals. It involves two major in-house softwares and an interactive web-based model forprediction of corrosion rate. Flowchart of the methodology is given in Figs. 1 and 2.

2.1 Weather stations and climate models

Eleven weather stations selected for this study were found distributed along the coast andinland locations (see Table 2 and Fig. 3). The data procured was 3-hourly and for the period of12 months. We processed these data using different climate change models and arrived atpossible projected weather patterns for future years.

All stations fall under different weather pattern zones (Fig. 3) characterized by annual rainfalland temperature-humidity across the country (Source: BoM). Representatives from all majorzones were studied, Summer dominant rainfall: Cairns and Darwin; Summer rainfall: Brisbane;

Fig. 1 Flow chart of the methodology

Climatic Change (2014) 124:133–146 135

Uniform rainfall: Sydney and Melbourne; Winter rainfall: Canberra and Adelaide; WinterDominant rainfall: Perth; Arid: Port Headland, Alice Springs and Broken Hill. The focus wasalso given to the eight stations (Port Headland, Perth, Adelaide, Melbourne, Sydney, Brisbane,Cairns, and Darwin) located near the coast due to high salinity, humidity, and marine aerosols.Although, the major effect of various climatological parameters is seen predominantly near shore,the inland effects are also important to compare and comprehend the resulting change. Hence,three inland stations (Canberra, Broken Hill and Alice Springs) were selected based on theirdistance from coast and the climatic conditions. Moreover many of these stations are major citiesand contain extensive infrastructures, which need extra attention for corrosion rate estimation.

To produce the projected weather pattern, a suitable climate scenario was chosen based onthe RCF framework (Clarke et al. 2011) and appropriate climate models were chosen (Rafterand Abbs 2009; Suppiah et al. 2007) based on the significant factors affecting the corrosion ofmetal surface (Rafter and Abbs 2009; Perkins et al. 2007). Data for our climate models werecollected from CSIRO’s interactive web-based model on Ozclim website (http://www.csiro.au/ozclim/home.do). Ozclim provides many options to predict the desired conditions for the givenclimate data. For every 5 years interval (from 2020 to 2100), annual, seasonal or monthlyaverage data for 15 different climate variables can be calculated. This calculation can beperformed for six different emission scenarios using 29 different climate models with threedifferent global warming rates. Out of those 29 climate models, three climate models wereselected based on the parameters required for the study. The models selected are: 1) MRI-CGCM 3.2.2—most likely model, 2) CSIRO-Mk3.5—dry model, 3) CCCAM- CGCM 3.1(T63)—wetter model. Based on these three models nine different possibilities (Table 1) of theclimate conditions were arrived at, and two (cases 1 and 4 in Table 1) were used in the study.The models MRI –CGCM 3.2.2 and CSIRO-Mk3.5, generate results for high global warming

Fig. 2 Global average warming (relative to 1980–99) for the scenarios A2, A1B and B1, shown as continuationsof the 20th century simulations. Shading denotes the plus/minus one standard deviation range of individualmodel annual averages. The orange line is for the experiment where greenhouse gas concentrations were heldconstant at year 2000 values. The grey bars (right) indicate the multi-model mean warming (solid line withineach bar) and the likely range of warming by the year 2100 for the six SRES marker scenarios, based on the A2,A1B and B1 simulations, plus results from independent models and observational constraints. (From IPCC(2007) Fig SPM-5.)

136 Climatic Change (2014) 124:133–146

Fig. 3 Percentage change in mass-loss in Australia in 2070 predicted by models a CSIRO-Mk 3.5 and b MRI-CGCM 3.2.2

Climatic Change (2014) 124:133–146 137

rate and A1FI emission scenario. They were chosen to highlight the difference between the dryand the most likely weather conditions, which in turn correspond to most likely and maximumcorrosion rates respectively.

2.2 Climate variables and parameters

The Ozclim data provides percent change of a climate variable from the base data for a givenmonth from a specified year. This percent change was used to change the RH and temperatureuniformly in the corrosion data base. For example Fig. 4(a) shows the original RH data forSydney in the month of December compared to the predicted data stream.

Such a simply multiplication by a climate change factor is not appropriate for rainfall as it isknown that although the total rainfall may increase or decrease, the frequency of rain may tendto decrease (even when total rainfall increases). Thus more intense rain will happen lessfrequently. One measure that accounts for rainfall intensity and frequency is the—SimpleDaily Intensity Index (SDII). SDII is defined as the total annual rainfall divided by number ofrain-days when total daily rainfall is more than 1 mm (Frich et al. 2002). Tebaldi et al. (2006)generated the SDII values for Australia and for rest of the world as a result of climate change.High resolution data for SDII based on Tebaldi et al. 2006, exclusive to Australia wereprovided by Alexander and Arblaster (2009) which have been used in this study.

The predicted rainfall sequence was required to meet the monthly change as predicted byOzclim and the change in SDII as predicted by Alexander and Arblaster (2009). The currentrainfall sequence was adapted by introducing a cut-off value and a percent change value. Thecut-off data was applied first and all rainfall events below cut-off were deleted from thesequence and then the percent change value was applied to the remaining events. The predictedrainfall sequence was deemed acceptable when its SDII value fitted within the range given by

Table 1 Possible combinations of climate change scenarios with three climate models, three global warmingrates and three emission scenarios

No. Climate change model Globalwarmingrate

Emissionscenario

Details

1 MRI: MRI-CGCM 3.2.2 High A1FI Median Model with intensiveFossil fuel technology

Median Model

2 MRI: MRI-CGCM 3.2.2 Medium A1B Median Model with BalancedEnergy Source technology

3 MRI: MRI-CGCM 3.2.2 Low A1T Median Model with non fossilfuel technology

4 CSIRO: CSIRO-Mk 3.5 High A1FI Dry Model with intensive Fossilfuel technology

Dry Model

5 CSIRO: CSIRO-Mk 3.5 Medium A1B Dry Model with Balanced EnergySource technology

6 CSIRO: CSIRO-Mk 3.5 Low A1T Dry Model with non fossil fueltechnology

7 CCCMA: CGCM-M 3.1 High A1FI Wet Model with intensive Fossilfuel technology

Wet Model

8 CCCMA: CGCM-M 3.1 Medium A1B Wet Model with Balanced EnergySource technology

9 CCCMA: CGCM-M 3.1 Low A1T Wet Model with non Fossil fueltechnology

138 Climatic Change (2014) 124:133–146

Alexander and Arblaster and the % change in total rainfall matched that of Ozclim. The cut-offvalue and SDII range for each station are given in Table 2.

Figure 4(b) shows the original and predicted rainfall for Sydney in the month of December.The cut-off value and intensification factor are 0.9 and 1.242 respectively. Ozclim predicts adecrease in rainfall of 31 %. It is evident that the predicted rainfall shows significantly fewerrainfall events, however the intensity of the remaining events remains very similar to theiroriginal values.

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30

Rel

ativ

e H

um

idit

y in

%

Dates of month

CurrentRelativeHumidity

ProjectedRelativeHumidity

0

1

2

3

4

5

6

7

8

9

10

0 5 10 15 20 25 30

Rai

nfa

ll in

mm

Dates of month

CurrentRainfall

ProjectedRainfall

a

b

Fig. 4 Distribution of the current and future projected data for a Relative humidity and b Rainfall

Climatic Change (2014) 124:133–146 139

Tebaldi et al. (2006) reported the projected change in the SDII from 1980–1999 to 2080–2099 data for A1B scenario. However, our study aims to project the corrosion rate for 2070based on A1FI scenario. Hence if the SDII value of Tebaldi et al. is to be used, the climatepredicted by A1B scenario for 2090 must be close to that predicted by A1FI Scenario in 2070.This is ensured by referring to Figure 4.5 of CSIRO Technical Report- Climate Change inAustralia, (2007). In that the predicted temperature rise for A1B for year 2090 is reported as2.8 °C, using the same graph the predicted temperature rise for A1FI scenario for year 2070can be estimated to be 2.72 °C. This substantiates the use of SDII from Tebaldi et al. for thisstudy (Fig. 2). [Figure 4.5 of CSIRO Technical Report- Climate Change in Australia, (2007) isoriginally taken from IPCC Fig. SPM-5, 2007].

2.3 Corrosion model and software

The corrosion model used in this study was developed by Cole et al. (2011). The model uses asinput three hourly climate data (RH, rainfall, wind speed, wind directions), ocean wave activity,coastal land forms and topography. It then calculates for every 3 h, the amount of salt depositedon a surface, any surface cleaning of salt (by rain) the surface temperature and RH of theexposedmetal. From these parameters the “state of the surface” is defined, that is the surface dry(state 1), wet from rain (state 2) or wet from the hygroscopic absorption of moisture (state 3)along with the quantity of retained salt. These calculations are made by linking a series ofprocess models including Computational Fluid Dynamics models (for aerosol transport anddeposition) and heat and mass transfer models for temperature and wetting. The mass-loss for a3 h period is then calculated from empirical equations with different equations being used forthe three different states. The total mass-loss for a year is then the sum of all the 3 h mass-losses(taking into account of course that the “state” of the metal may change every 3 h). Theseempirical equations have been derived from a series of controlled environment chamber tests.For example the corrosion rate of a surface wetted by hygroscopic salts is estimated as.

Mass loss over 3hð Þ ¼ λþ β � Sψ

Table 2 Cut-off values for Rainfall for 11 weather stations for different climate models (Station IDs are providedby the Bureau of Meteorology, Australia)

Station name(station ID)

MRI-CGCM 3.2.2 (case 1) CSIRO-Mk 3.5 (case 4) Expectedchangein SDIICut-off

value forrainfall

CurrentSDII

ProjectedSDII

Changein SDII

Cut-offvalue forrainfall

CurrentSDII

ProjectedSDII

Changein SDII

Port Hedland (004032) 0.48 3.7 3.8 0.12 0.65 3.5 3.8 0.28 0.2−0.4Perth (009021) 0.6 4.2 4.5 0.29 0.2 4.0 4.2 0.11 0−0.2Darwin (014015) 1.5 8.0 8.5 0.47 0.2 8.0 8.0 −0.059 0.4−0.6Alice Springs (015590) 0.4 3.0 2.9 −0.24 0.5 3.1 2.9 −0.2 −0.4−−0.2Adelaide (023034) 0.5 3.0 3.6 0.57 0.62 3.0 3.4 0.32 0.4−0.6Cairns (031011) 1.85 10.0 10.8 0.79 1.25 10.0 9.8 0.46 0.4−0.6Brisbane (040223) 1.04 4.7 4.8 0.051 0.7 4.7 4.8 0.097 0−0.2Broken Hill (047007) 0.75 3.7 4.1 0.42 0.7 3.8 4.2 0.43 0.4−0.6Sydney (066062) 0.45 5.2 5.7 0.57 0.9 5.2 5.7 0.59 0.4−0.6Canberra (070014) 0.58 3.1 3.9 0.80 0.62 3.1 3.9 0.79 0.6−0.8Melbourne (086071) 0.4 3.1 3.7 0.54 0.62 3.1 3.6 0.47 0.4−0.6

140 Climatic Change (2014) 124:133–146

Where mass loss is in g/m2, S is the salinity level (mg/m2) on the surface, λ, β and ψ areconstants with values 8.3×10−4, 11.3×10−4 and 0.5 respectively.

The newly generated weather data files (for temperature, RH, and rainfall) were replaced inthe corrosion prediction software and the results for annual mass-loss were generated for all 11stations. The corrosion predictor also allowed for alteration of the wind speed, wind direction,salinity, and aerosol production parameters. However, for this study we have only includedmajor climatic parameters like dry bulb temperature, RH, and rainfall. The predicted mass-lossis plotted on the map of Australia to understand the distribution of corrosion rates in the future.

3 Results and discussions

The study shows the effect of climate change on the corrosion rate of the structures inAustralia. The wide ranging results presented are based on the data from the various weatherstations based on their location and climatic factors prevailing in that area.

For the purpose of this study we initially classified with the station into two groups (Coleet al. 2003a, b).

1) Coastal Stations: Weather stations within 50 km from the coastline are considered coastal.These include: Port Headland, Perth, Adelaide, Melbourne, Sydney, Brisbane, Cairns, andDarwin. Due to their near coast location, corrosion is mainly controlled by the saltdeposition on the surfaces, which in turn is correlated to salt levels in the atmosphere(surf produced aerosols) and cleaning of the surfaces (for cleaning to occur a minimumlevel of rainfall is required, Cole and Paterson 2007). As a result of climate change infuture years the frequency of rain events and thus rain will decrease, leading to higherretained salt on the surface and consequently higher mass-loss.

2) Inland Stations: Weather stations which are further than 100 km from the coast areconsidered inland. These include: Canberra, Broken Hill, and Alice Springs. As thesestations are substantially far from coast the effect of the surf produced aerosols is minimal.The main factor affecting the corrosion rate in these regions is the diurnal cycle. The RHat night-time which promotes condensation. The climate change in future years will leadto higher temperatures and thus to a reduced RH in these areas. This in turn reduces thenight time condensation, and therefore will lead to less mass-loss.

Our results co-inside with this reasoning and show the actual amount of change (increase ordecrease), in terms of percentage mass-loss, expected in 2070 for two different climate changemodels with high global warming rate in A1FI scenario.

3.1 Effect of rainfall

Rainfall is a discrete variable and the variability in the rainfall arises from the frequency andintensity of the rainfall in the future years, both of which affect the rate of corrosion. Thecorrosion rate depends both on the deposition and retention of the salts on the metal. If thesurface is cleaned often the resulting cleaning effect can reduce the corrosion. Many climatechange prediction studies show that in the future years the increased temperatures will causethe change in the rainfall patterns (Kharin et al. 2007; Alexander and Arblaster 2009; Rafterand Abbs 2009). Results also indicate the overall increase in the intensity of rainfall withreduced frequency of rainfall events. Hence, the reduced frequency of cleaning will lead toincreased corrosion of the metal surfaces. This effect will vary every month due to variation in

Climatic Change (2014) 124:133–146 141

the seasons and the climatic condition of the respective areas. Therefore this study focuses onstudying the 3-hourly weather patterns for the entire year rather than just the yearly averages.

In Table 3, the predicted mass-loss is given for the two different Scenarios, MRI-CGCM3.2.2 and CSIROMk3.5. It will be recalled from Table 1 that the CSIRO model is a dry model.For example compare the annual rainfall changes for Port Headland, Darwin and Adelaide, forMRI-CGCM 3.2.2 they are −35 %, −40 %, and −14 % respectively while for CSIRO-Mk 3.5they are −46%, −57 % and −45 %. The reduced rainfall results in increased corrosion rates andthese rates are increased significantly more in the CSIRO dry model.

3.2 Climate models and corrosion scenario

When designing structures with the intent of long service life, it is necessary to understand themost likely climate scenario that structure would operate under. However better design can beachieved if the design tolerances could be calculated based on the most extreme condition thestructure is likely to encounter. Hence we considered two climate models which would predictthe most likely (MRI: MRI-CGCM 3.2.2) and extreme (CSIRO: CSIRO-Mk 3.5) climatescenarios for corrosion.

The CSIRO model is a considerably dry model and therefore produces results for the mostsevere climatic conditions for future years. Results generated using this model give highertemperatures and low RH and significant rainfall variations and show severe increase in thecorrosion rate in the coastal areas except for Perth and Brisbane. The MRI model gives morebalanced climatic scenario with a moderate increase in temperature and moderate decrease inRH and rainfall. Results generated using this model give the most likely climatic scenario andtherefore we may expect the predicted corrosion rates to be lower than the rates predicted byCSIRO model. While the MRI model in general shows relative increase in the corrosion ratefor the coastal stations (except for Melbourne) similar to .the CSIRO model, the predictedcorrosion rates are far higher than the latter.

Table 3 Percent change in mass-loss for the dry model (CSIRO-Mk 3.5) and most likely model (MRI CGCM3.2.2) with High Global Warming Rate and A1FI Emission Scenario

Location State MRI-CGCM 3.2.2 CSIRO-Mk 3.5

2070 annualmass-loss

Current annualmass-loss

% change inmass-loss

2070 annualmass-loss

Current annualmass-loss

% change inmass-loss

Port Hedland WA 30 11 179 34 11 215

Perth WA 18 5.3 247 4.9 5.3 −7.7Darwin NT 21 7.1 192 9.0 7.1 26

Alice Springs NT 0.1 4.3 −98 0.03 4.3 −99Adelaide SA 16 5.6 192 29 5.6 415

Cairns QLD 83 44 86 52 44 18

Brisbane QLD 12 5.5 115 4.6 5.5 −16Broken Hill NSW 0.04 0.9 −95 0.050 0.9 −94Sydney NSW 42 36 16 97 36 168

Canberra ACT 0.8 4.1 −80 0.9 4.1 −78Melbourne VIC 1.6 4.5 −65 7.1 4.5 59

Underlined values show the tolerance levels (highest corrosion rate expected) considered for the future structures

142 Climatic Change (2014) 124:133–146

Although corrosion rates generally increase for coastal locations there are three exceptionsin the data: Perth and Brisbane in the CSIRO climate model, and Melbourne in the MRIclimate model. In fact these three locations are at a considerable distance from surf beachesalthough they are close to the bay beaches. The work of Cole et al. (2003a, b) has demon-strated that salt production from bays is only a fraction of that generated by surf beaches andthus the crucial factor for determining airborne salinity is the distance to the active coast. Asillustrated in Fig. 5 all three locations are at a considerable distance from “active” coasts andthus have relatively low salinity values. Due to these low values the impact of salt accumu-lation is less dramatic on corrosion and so the impact of reduced cleaning is also lowered.

On the other hand, all the inland weather stations show a decrease in the corrosion rate forboth the models. This decrease can be attributed to two factors: a) low airborne salinity due tolarge distance from the coast b) the reduced RH and consequently less condensation duringnight times.

In addition to these major divisions between coastal and inland locations, other climaticeffects may influence corrosion rates. Some indication of this can be gained from Fig. 3 wherethe predicted corrosion rates are overlaid on the map of the Australia zoned based on therainfall and humidity. It is notable that in the MRI- CGCM 3.2.2 model, the increase incorrosion rate for Brisbane is much higher than for Sydney even though they are both coastalcities. The predicted annual reduction in rainfall for Brisbane according to this scenario will be46.2 % while that of Sydney will be 8.7 %. Note that the rain in Brisbane is predominately insummer while that in Sydney is uniform over the year. Thus reduction of rainfall events inBrisbane could lead to long periods (out of summer) when cleaning does not occur. Thus it ispossible that the seasonal variation and intensity of rainfall in the area affects the corrosionrates greatly.

The analysis presented here also reveals another interesting aspect: higher corrosion ratesare predicted for the “most-likely” climatic condition (MRI model) than for the “extreme”climatic condition (CSIRO model). For example, the predicted corrosion rate for Darwin usingthe CSIRO and MRI models are 26 % and 192 % respectively. Thus, what is perceived as anextreme climate scenario by the climate models (say from the point of view of humaninhabitation or environmental and ecosystem analysis) need not necessarily correspond tothe conditions required for extreme corrosion rates. Once again this result can be attributed tothe highly non-linear effect of climate variables on holistic corrosion prediction.

Table 3 reports the greatest increase (underlined values) from the above two models foreach location for the year 2070. To ensure that our predictions for 2070 are keeping with thetrends over the years, we predicted the annual mass-loss results for year 2030 and 2050 forCairns and Alice Springs as representatives of coastal and inland weather stations respectively.Cairns was estimated to incur a 15 % change in mass-loss in 2030, 23 % in 2050 and 86 % in2070 while the change in Alice springs was −98 % for all years, thus confirming the trends. Itwould be conservative to take the extreme values into consideration for the sustainabledevelopment of future building structures.

This paper has followed a different methodology for the estimation of mass loss thanprevious works that looked at the effect of climate change on infrastructure [11, 26]. Both [11]and [26] used damage equations or dose functions where an aspect of or total damage wascalculated as a function of annual averages in climate variables. Thus the effect of climatechange was calculated from changes in the annual average variables. In this work three hourlyclimatic data are used to calculate damage as a function of the state of the surface in that 3 hperiod and thus to determine the effect of climate change a full-yearly climate sequence isrequired for the projected year. The present authors contend that this methodology is moreaccurate when specific temporal events and their frequency (i.e. rain washing for atmospheric

Climatic Change (2014) 124:133–146 143

Fig. 5 The geographic locationof weather stations and theirsurroundings: a Brisbaneb Melbourne, and c Perth

144 Climatic Change (2014) 124:133–146

corrosion) control the rate of corrosion. However comparative research between the twoapproaches is required to determine the relative merits of each method for different materialdegradation problems. Unfortunately these comparisons cannot be made from the currentliterature as the papers have either studied diverse geographic zones or materials. Howeverwhat this and the other papers [11, 26] have highlighted is that climate change effects can bequiet significant on materials degradation and that they can change dramatically depending ongeographic and climatic zone. More research is thus needed to resolve these methodologicalvariations and to map out the variations in climate change effects for a wide range ofconstruction materials.

4 Conclusion

In this study we have used real climate data to predict future climate scenarios using twoclimate models CSIRO: CSIRO-Mk 3.5 and MRI: MRI-CGCM 3.2.2 and used the predictedscenarios to forecast the future corrosion rates using an in-house corrosion predictor model.The study revealed complex non-linear relationships between various climate variables andother geographical factors. Coastal cities generally showed higher future corrosion rates whencompared to their inland counterparts although anomalies were detected and attributed to thelocal geographical features of the areas surrounding the weather stations. The analysis, whichused the real climate data for Australia also revealed that, what is perceived as an extremeclimate scenario by the climate models need not necessarily correspond to the conditionsrequired for extreme corrosion rates. However the methodology presented here is general andcan be applied to any geographical region in the world. The balance of factors and conse-quently the forecasted corrosion rates may however be different for that region. This method-ology could be applied to computer design tolerances for infrastructure to ensure safety andlongevity in operation thus contributing to higher sustainability.

Acknowledgments We would like to acknowledge the help of Dr. Leanne Webb from CSIRO Marine andatmospheric Research, Aspendale, Victoria, Australia, in selecting appropriate climate models for this study.

References

Alexander LV, Arblaster JM (2009) Assessing trends in observed and modelled climate extremes over Australiain relation to future projections. Int J Climatol 29:417–435

Clarke JM, Whetton PH, Hennessy KJ (2011). Providing application-specific climate projections datasets:CSIRO’s Climate Futures Framework. 19th International Congress on Modelling and Simulation, Perth,Australia, 12–16 December 2011. http://mssanz.org.au/modsim2011

Cole IS, Paterson DA (2007) Holistic model for atmospheric corrosion- Part 7- Cleaning of salt from metalsurfaces. Corros Eng Sci Technol 42(2):106–111

Cole IS, Paterson DA (2010) Possible effects of climate change on atmospheric corrosion in Australia. CorrosEng Sci Technol 45(1):19–26

Cole IS, Paterson DA, Ganther WD (2003a) Holistic model for atmospheric corrosion Part 1 – Theoreticalframework for production, transportation and deposition of marine salts. Corros Eng Sci Technol 38(2):129–134

Cole IS, Paterson DA, Ganther WD, Neufeld A, Hinton B, McAdam G, McGeachie M, Jeffery R, ChotimongkolL, Bhamornsut C, Hue HV, Purwadaria S (2003b) Holistic model for atmospheric corrosion- Part 3- Effect ofnatural and manmade landforms on deposition of marine salts in Australia and south-east Asia. Corros EngSci Technol 38(4):267–274

Climatic Change (2014) 124:133–146 145

Cole IS, Muster TH, Azmat NS, Venkatraman MS, Cook A (2011) Multiscale modelling of the corrosion ofmetals under atmospheric corrosion. Electrochim Acta 56:1856–1865

CSIRO (2007) Climate change in Australia. CSIRO, AustraliaDoney SC, Ruckelshaus M, Duffy JE, Barry JP, Chan F, English CA, Galindo HM, Grebmeier JM, Hollowed

AB, Knowlton N, Polovina J, Rabalais NN, Sydeman WJ, Talley LD (2012) Climate change impacts onmarine ecosystems. Ann Rev Mar Sci 4:11–37

Frich P, Alexander LV, Della-Marta P, Gleason B, Haylock M, Klein Tank AMG, Peterson T (2002) Observedcoherent changes in climatic extremes during the second half of the twentieth century. Clim Res 19:193–212

Grøntoft T (2011) Climate change impact on building surfaces and façades. Int J Clim Chang Strateg Manag3(4):374–385

IPCC (2000) In: Nakicenovic N, Swart R (eds) ‘Emissions scenarios’, Special report of the IntergovernmentalPanel on Climate Change, vol 570. Cambridge University Press, Cambridge

IPCC (2007) In: Solomon S et al (eds) Climate change 2007: the physical science basis. Contribution of WorkingGroup I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, vol 996.Cambridge University Press, Cambridge

Jacob DJ, Winner DA (2009) Effect of climate change on air quality. Atmos Environ 43:51–63Kharin VV, Zwiers F, Zhang X, Hegerl GC (2007) Changes in temperature and precipitation extremes in the

IPCC Ensemble of Global Coupled Model Simulations. J Clim 20:1419–1444Kiktev D, Caesar J, Alexander LV, Shiogama H, Collier M (2007) Comparison of observed and multimodeled

trends in annual extremes of temperature and precipitation. Geophys Res Lett 34(10), L10702. doi:10.1029/2007GL029539

Lindner M, Maroschek M, Netherer S, Kremer A, Barbati A, Garcia-Gonzalo J, Seidl R, Delzon S, Corona P,Kolström M, Lexer MJ, Marchetti M (2010) Climate change impacts, adaptive capacity, and vulnerability ofEuropean forest ecosystems. For Ecol Manag 259:698–709

Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye AT, Gregory JM, Kitoh A, Knutti R, Murphy JM,Noda A, Raper SCB, Watterson IG, Weaver AJ, Zhao ZC (2007) In: Solomon S et al (eds) ‘Climate change2007: the physical science basis’, Contribution of Working Group I to the Fourth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, pp 747–846

Nijland TG, Adan OCG, van Hees RPJ, van Etten BD (2009) Evaluation of the effects of expected climatechange on the durability of building materials with suggestions for adaption. Heron 54(1):37–48

Perkins SE, Pitman AJ, Holbrook NJ, McAneney J (2007) Evaluation of the AR4 climate models’ simulateddaily maximum temperature, minimum temperature, and precipitation over Australia using probabilitydensity functions. J Clim 20(17):4356–4376

Pitman A, Perkins S (2008) Regional projections of future seasonal and annual changes in rainfall andtemperature over Australia based on skill-selected AR(4) models. Earth Interact 12:1–50

Rafter T, Abbs D (2009) An analysis of future changes in extreme rainfall over Australian regions based on GCMsimulations and Extreme Value Analysis. CAWCR Res Lett 3:44–55

Roberge PR (2010) Impact of climate change on corrosion risk. Corros Eng Sci Technol 45(1):27–33Smith I, Chandler E (2010) Refining rainfall projections for the Murray Darling Basin of southeast Australia—

the effect of sampling model results based on performance. Clim Chang 102(3–4):377–393Stewart MG, Peng J (2010) Life-cycle cost assessment of climate change adaption measures to minimise

carbonation-induced corrosion risks. Int J Eng Uncertain: Hazards Assess Mitig 2(1–2):35–45Stewart MG, Wang X, Nguyen MN (2011) Climate change impact and risks of concrete infrastructure

deterioration. Eng Struct 33:1326–1337Suppiah R, Hennessy K, Whetton PH, McInnes K, Macadam I, Bathols J, Ricketts J, Page CM (2007) Australian

climate change projections derived from simulations performed for the IPCC 4th Assessment Report. AustMeteorol Mag 56(3):131–152

Talukdar S, Banthia N, Grace JR (2012) Carbonation in Concrete infrastructure in the context of global climatechange- part 1: experimental results and model development. Cem Concr Compos 34:924–930

Tebaldi C, Hayhoe K, Arblaster JM, Meehl GA (2006) Going to the extremes: an intercomparison of model-simulated historical and future changes in extreme events. Clim Chang 79:185–211

Whetton P, Hennessy K, Clarke J, McInnes K, Kent D (2012) Use of representative climate futures in impact andadaptation assessment. Climate Change 115:433–442. doi:10.1007/s10584-012-0471-z

Yoon IS, Çopuroğlu O, Park KB (2007) Effect of global climatic change on carbonation progress of concrete.Atmos Environ 41(34):7274–7285

146 Climatic Change (2014) 124:133–146