An Assessment of Climate Change Effects on Atmospheric Corrosion Rates of Steel Structures

Abstract

This paper presents a discussion on assessing the potential impacts of climate change on the atmospheric corrosion rates of exposed steel structures. The effects on atmospheric corrosion due to changes in the environmental temperature, carbon dioxide (CO2), relative humidity (RH), wind, rainfall and pollution are considered. The limitations and complexities of these assessments are discussed. To demonstrate the use and limitations of this science to evaluate effects related to climate change, a model developed in Australia to predict corrosion is combined with climate change models to project the change in the corrosion rates of steel components and protective zinc coatings in constructions. The method is applied to constructions located along the coastal areas of two Australian cities, Melbourne and Brisbane. These assessments are made using the A1FI scenario, the highest emission scenario defined by the Intergovernmental Panel on Climate Change (IPCC), applied to nine General Circulation Models (GCMs). The projected changes in corrosion rates were found to be an increase of 14% for both zinc and steel in Brisbane and a decrease of about 14% for steel and 9% for zinc in Melbourne. It was also found that the uncertainties associated with the climate change models were small compared to the uncertainties involved in modelling corrosion for engineering purposes.

Full text

An assessment of climate change effects on
atmospheric corrosion rates of steel structures
M. N. Nguyen*
1
, X. Wang
2
and R. H. Leicester
2
This paper presents a discussion on assessing the potential impacts of climate change on the
atmospheric corrosion rates of exposed steel structures. The effects on atmospheric corrosion due
to changes in the environmental temperature, carbon dioxide, relative humidity, wind, rainfall and
pollution are considered. The limitations and complexities of these assessments are discussed. To
demonstrate the use and limitations of this science to evaluate effects related to climate change, a
model developed in Australia to predict corrosion is combined with climate change models to
project the change in the corrosion rates of steel components and protective zinc coatings in
constructions. The method is applied to constructions located along the coastal areas of two
Australian cities: Melbourne and Brisbane. These assessments are made using the A1FI scenario,
the highest emission scenario defined by the Intergovernmental Panel on Climate Change, applied
to nine general circulation models. The projected changes in corrosion rates were found to be an
increase of ,14% for both zinc and steel in Brisbane and a decrease of ,14% for steel and 9% for
zinc in Melbourne. It was also found that the uncertainties associated with the climate change
models were small compared to those involved in modelling corrosion for engineering purposes.
Keywords: Galvanised steel, Zinc coating, Atmospheric corrosion, Climate change, Emission scenario, Airborne salinity
Introduction
Steel is the most commonly used material for a wide
range of infrastructure components and industrial
equipment. It is often employed in exposed conditions
in coastal areas and/or highly polluted industrial areas,
both areas where unfortunately the atmospheric corro-
sion is of high concern. Koch et al.
1
reported that
corrosion costs in the United States were about
US$137?9 billion per year for 26 industry sectors and
extrapolated to US$276 billion per year (3?14% of gross
domestic product) for the entire US industry. In
Australia, it was estimated that corrosion may have
cost up to $32 billion per annum, which means more
than $1500 for every person in Australia each year.
2
As a
mean of protection from corrosion, a zinc coating is
often used on the surface of structural steel. The zinc
coating (galvanising) industry for steel construction is
large, consuming about half of the zinc production of
the world.
Projected future climate changes will affect the extent
of atmospheric corrosion, and it is important to assess
the potential magnitude of these changes. Recently,
Cole and Paterson
3
and Roberge
4
have reviewed and
discussed possible impacts that climate change may have
on the corrosion. It was argued that it is currently not
possible or very difficult to quantify the effects of
climate change on atmospheric corrosion. The argu-
ments are based on the following: the climate change
models are not clear on predictions of aerosol produc-
tion and the movement of various airborne corrosive
agents,
3
and it is unclear as to how the world economy
will evolve in the future.
5
However, it is obvious that for
future planning purposes, there is benefit to be obtained
in assessing the magnitudes and uncertainties associated
with corrosion estimates related to the use of climate
change projection models.
Some work has been reported on the impact of
climate change on corrosion of steel reinforcement in
concrete structures (e.g. Refs. 6–8). However, this paper
is probably the first to attempt to quantify climate
change impacts on the atmospheric corrosion of steel
structures in specific locations.
The following is a review on some significant features
of atmospheric corrosion.
Review of factors related to atmospheric
corrosion
Overview
The atmospheric corrosion of steel components and zinc
coatings has been an active research field for several
decades. It encompasses a wide range of industrial and
domestic steel structural components in constructions
that are fully exposed to weather and thus subjected to
1
Land & Water Division, Climate Adaptation Flagship, Commonwealth
Scientific and Industrial Research Organisation, Melbourne, Vic., Australia
2
Ecosystem Sciences Division, Climate Adaptation Flagship, Commonwealth
Scientific and Industrial Research Organisation, Melbourne, Vic., Australia
*Corresponding author, email Minh.Nguyen@csiro.au
ß2013 Institute of Materials, Minerals and Mining
Published by Maney on behalf of the Institute
Received 23 October 2012; accepted 1 1 February 2013
DOI 10.1179/1743278213Y.0000000087 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5359

corrosion. Various studies have been conducted, mostly
by laboratory experiments and/or field monitoring.
There is generally a consensus
9–11
that long term
atmospheric corrosion of metal is described by the
following equation
catm~c0tn(1)
where c
atm
is the atmospheric corrosion depth (in mm)
after tyears in service, c
0
(mm) is the corrosion depth at
the end of the first year, tis the duration in service (in
years) and nis a power factor ,1?0, giving the generic
corrosion curve with time as shown in Fig. 1. Since the
corrosion rate at any given time is proportional to c
0
, for
descriptive convenience, we will refer to c
0
as the
‘corrosion rate parameter’.
It should be noted that an alternative and rather more
sophisticated concept of atmospheric corrosion in marine
environments has been investigated by Melchers.
12,13
For
this model, he proposes several phases for the progression
of corrosion, each one with its own mechanism. There is
no doubt that this model is an improvement on existing
models in terms of both the rationality of the corrosion
mechanisms proposed and accordingly the accuracy in
predicting the progress of corrosion. However, for
reasons related to data availability and practical feasi-
bility, the corrosion models considered in this paper have
been chosen to relate to equation (1) above.
Many studies have been based on the use of short term
measurements of corrosion to estimate the corrosion rate
parameter c
0
in equation (1) (e.g. Ref. 14) and on long term
tests to estimate the power factor n(e.g.Refs. 10,11and15).
There has been general agreement that the corrosion rate
parameter c
0
is mainly governed by the following factors:
(i) time of surface wetness, which is approximated
as the percentage of time in a year that the metal
surface is wet by a moisture layer; for Australian
coastal areas, an atmospheric corrosion model
16
has been fitted to the assumption that the
moisture layer is formed when the relative
humidity (RH) is above 80% and the tempera-
ture above 0uC
(ii) airborne salinity, in terms of chloride concentra-
tions in the air
(iii) airborne pollution, mainly in terms of sulphide
concentrations in the air.
For steel components with zinc coatings, it is assumed
that the corrosion will progress successively, i.e. corrosion
will first occur on the zinc coating, and then on the steel
substrate after there is no zinc coating left. Corrosion of
both zinc and steel in the atmosphere is a complex
discontinuous electrochemical process, which happens
with the presence of water on the metal surface and
airborne pollutants, including sulphur dioxide in indus-
trial areas and airborne chlorides in marine areas.
A review of the atmospheric corrosion process of zinc
was provided by de la Fuente et al.
11
In general, when
exposed to the atmosphere with the presence of water,
the reaction of zinc with atmospheric oxygen produces
various corrosion products, where the most important
stable carbonate is hydrozincite [Zn
5
(CO
3
)
2
(OH)
6
],
which has a protective effect against further corrosion.
Since the corrosion process of zinc is also remarkably
slower than that of steel, the zinc coating acts as a
protection layer that delays the corrosion progressing
into a steel substrate, which is the part that bears the
structural loads. The thickness of a zinc coating for
structural steel components is usually in the range of 25–
85 mm.
17
For practical purposes, galvanised steel com-
ponents are considered to reach the end of their service
life when the zinc coating layer is fully corroded.
De la Fuente et al.
15
have also provided a detailed
review of the atmospheric corrosion process of steel.
Oxides, hydroxides and salts (chloride or sulphate) of
iron are commonly found in the corrosion products that
form corrosion layers. The corrosion layers exhibit a
large number of pores and microcracks that make them
highly defective and permeable, and may eventually
become detached when they become too thick. The
corrosion layers of steel therefore practically provide
little protection against further corrosion.
Effect of changes in rainfall patterns
There is a well known beneficial effect of rain in washing
out the atmospheric corrosive pollutants that have
settled on exposed surfaces, thus reducing the corrosion
rate.
18
The projection for future climate is that rainfall
will occur in higher intensity falls but with reduced
frequency, i.e. the cleansing effects may be reduced. In a
sophisticated model, Cole and Paterson
3
introduced the
concept of cutoff rainfall, which is the minimum rainfall
required to clean a surface. The extent of the washing
effect of a rain event is then estimated depending on its
intensity relative to the cutoff rainfall.
Effect of changes in RH
Apart from affecting the time of surface wetness
mentioned in section on ‘Overview’, increases in RH
would increase the size of surf produced aerosols and thus
the salt deposit, especially in marine areas.
3
However, this
effect would not be significant for Australia, where RH is
projected to decrease over most of the country.
19
Effect of temperature
The effect of temperature on atmospheric corrosion is
complex. Atmospheric corrosion is an electrochemical
process. In theory, the corrosion rate is therefore
dependent on ambient temperature. The dependence
follows the Arrhenius law
c0~ae{Ea=RK (2)
where c
0
is corrosion rate parameter, ais a frequency
factor, Kis the absolute temperature (kelvin), E
a
is the
1 Progress of atmospheric corrosion (equation (1))
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
360 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5

activation energy and Ris the Boltzmann gas constant
(R58?31 J K
21
mol
21
).
Based on the theoretical Arrhenius equation, it has
been suggested that temperature could be an important
factor in atmospheric corrosion.
20
This has been
reflected in a general ‘rule of thumb’ that ‘a 10uC
increase in temperature will double the corrosion
rate’
18,21
within a normal range of ambient temperature
of 20–30uC (293–303 K). With this rule, the activation
energy E
a
is ,50 kJ mol
21
. Pacheco and Ferreira
20
reported similar values of activation energy from a series
of lab tests. However, it is important to note that this
extent of the temperature effect on the corrosion is only
valid at a constant condition
22
of a very high humidity
level.
23
In reality, the atmospheric corrosion is a complex
discontinuous electrochemical process subjected to highly
variable ambient conditions, where the corrosion reaction
only occurs when there is a moisture layer formed on the
metal surface. For Australian conditions, metal surfaces
commonly undergo daily wetting cycles.
3
At night, as the
ambient RH increases and temperature decreases, a
moisture layer forms and then facilitates the corrosion
reactions on the metal surface. At sunrise, when
temperature increases and RH decreases, the moisture
layer evaporates and thus the corrosion reactions halt.
With such a complex condition, the effect of temperature
on atmospheric corrosion in ambient condition was
perceived to be secondary.
3,24
This was also observed in
the results of some field corrosion tests.
11,20
Lindstrom
et al.
25
reported an interesting experimental result that in
the presence of CO
2
, the zinc corrosion rate did not show
a dependence on temperature. Cole and Paterson
3
have
suggested that for practical applications ‘an increase of
2 K (from 293 to 295 K) will promote a 0?6% change in
corrosion rate’.
Effect of pollution
Airborne pollution in terms of SO
2
is produced from
industrial activities. Global projections for SO
2
have
been provided by the Intergovernmental Panel on
Climate Change (IPCC)
26
with large uncertainties. On
average, the IPCC projected that the global increases of
SO
2
‘are generally modest, and numerous scenarios
even depict a long term decline in emissions’. Graedel
and Leygraf
5
established two scenarios for the projec-
tion of SO
2
emission at various regions and the whole
world for consideration of atmospheric corrosion. One
is for a ‘no control’ scenario where it is assumed that
there are no new control measures on SO
2
emission in
any region after 1990. While SO
2
emission levels of
developing countries and the whole world are projected
to increase, the SO
2
emission levels in developed
countries were projected to decrease, as cleaner and
more efficient technologies were employed. In particu-
lar, the levels of SO
2
emission in Australia were
projected to reduce by half in 2100, even in the ‘no
control’ scenario.
The pollution effects are usually only a minor
component of corrosion in marine locations.
3
Further-
more, pollution is essentially a local problem, where a
simple formula for estimating the level of SO
2
based on
distance to the industry zones and type of the indus-
try proposed by Nguyen et al.
16
is deemed to be
appropriate for practical design and is not dependent
on climate. Accordingly, the effects of airborne pollution
on atmospheric corrosion in Australia need not be
considered in the context of examining the impacts of
climate change.
Effects of CO
2
concentration
The CO
2
concentration increase may have effects on zinc
corrosion rates. A series of laboratory experiments
reported by Chung et al.
27
indicated that the formation
of hydrozincite, which is the stable protective zinc
corrosion product, is accelerated with increasing RH
and CO
2
concentration, resulting in an increase in the
resistance against corrosion and thereby a reduction in
the corrosion progress of zinc. These experiments
indicated that at high RH (95–100%), after 72 h
exposure, the corrosion resistance of zinc specimens in
a chamber of 1000 ppm CO
2
concentration is 20%
higher than the resistance of specimens in a chamber of
350 ppm CO
2
concentration. These concentrations are
quite similar to the condition of the climate change
scenario A1FI, which has the highest CO
2
emission
among all scenarios defined by IPCC. In this case, the
CO
2
concentration is projected to increase from
350 ppm in 1990 to ,1000 ppm in 2100.
28
Increases of
CO
2
concentration would reduce the corrosion rate, but
it is important to note that this was an effect observed
only at a very early stage of corrosion through
experiments in laboratory conditions only, i.e. within a
chamber at a constant high RH condition. Consider-
ing that the corrosion is a long term discontinuous
process, this effect is likely to be secondary in reality
and hence assumed to be negligible herein. Further
research and validation to clarify this effect on steel
component construction at real service conditions are
required.
Method for assessment of climate
change effects
In this study, an attempt was made to go as far as
possible in using the available information on corrosion
and climate change models to assess corrosion rates in
the years up to 2100, relative to the values in 1990.
Most of the assessment was made using an Aus-
tralian model for corrosion developed for engineering
purposes;
16
the model predicts the atmospheric corro-
sion rates of structural steel fasteners and related zinc
coatings. It was developed using available scientific
information and available field data to provide a set of
simple parameters that were suitable for engineering
applications.
Atmospheric corrosion is a complex process that
depends on the interactions of many environmental
factors. To account for these factors, the application
of the Australian model requires the input of nu-
merous parameters such as the geographical location
of the structure, the distance from the coast, the
coastal geometry, topographical influences, the terrain
roughness, building shelter and parameters related to
pollution.
In the following assessment, for clarity in presentation,
pollution effects will not be considered, and only changes
in the corrosion of structures located within the coastal
zone, say within a distance of 1?0 km from the coast, will
be evaluated. Furthermore, since these changes in
corrosion rates will be evaluated as relative percentages,
most of the engineering parameters mentioned in the
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5361

previous paragraph remain constant, including the exact
distance from coastline, and hence need not be consid-
ered. In addition, for reasons of clarity in presentation,
corrosion rates at only two locations situated on the east
coast of Australia will be examined; these are the cities of
Melbourne and Brisbane.
Climate change science has projected a different
climate in the future with a mean global temperature
increase of 4–6uC by the end of the century, leading to
changes in other climate parameters and changes in
intensity and frequency of weather events.
19,28
For this
paper, the assessments of climate change will be made
using the A1FI emission scenario, the highest emission
scenario defined by the IPCC, applied to the nine
general circulation models (GCMs) listed in Table 1.
Models used for atmospheric corrosion
Model for corrosion due to effects of airborne
salt
For engineering applications in Australia, Nguyen
et al.
16
proposed that in the absence of pollution, the
mean atmospheric corrosion rate of zinc may be
estimated from
c0,z~0:025 t0:6
wetS0:5
air (3)
and for steel
c0,s~0:5t0:8
wetS0:5
air (4)
where c
0,z
and c
0,s
are corrosion rate parameters (mm),
which are the corrosion depths (mm) of zinc and steel for
the first year respectively, t
wet
is the time of surface
wetness (%) and S
air
is the airborne salinity (mg/m
2
/day)
(measured as a deposit on a salt candle).
29
The
parameter t
wet
is computed as the percentage of time
that the RH is .80% and the temperature is above 0uC.
The general shape of the corrosion curve corresponds to
equation (1) and is shown in Fig. 1.
Nguyen et al.
16
presented models for estimating the
parameters of equations (3) and (4) for Australia. These
models were calibrated to various sources of corrosion
data, including the Australian Standard AS 4312,
30
various field test data
31–33
and corrosion test data in
South East Asia including North Australia.
34
The model
calibration is presented in a report by Nguyen et al.
35
The effect of climate change on the parameters t
wet
and S
air
cannot be obtained directly from the climate
change models, and so indirect methods must be used.
Although climate change models do not provide
projected values of t
wet
, they do provide projections of
RH, the annual average value of RH. Historical weather
data from past years have indicated that for the two
target cities of Melbourne and Brisbane, these para-
meters have been related by
twet~a|RH zb(5)
where the parameters aand bare given in Table 2, and
the data on the parameter relationships are plotted in
Fig. 2.
For the projection of airborne salinity for the future,
it will be assumed that most of this salinity is generated
primarily from the action of surf and ocean waves.
These sea state activities are governed by wind and fetch
characteristics for a specific location. A relationship of
volumetric airborne salinity S
air,vol
(mgm
23
) with mean
wind speed U(m s
21
) was given by McKay et al.
36
based
on measurements as follows
ln Sair,vol
ðÞ
~0:23Uz3:05 (6)
Cole et al.
37
used this relationship in the establishment
of a geographic information system for the prediction of
airborne salinity based on wind and fetch characteristics
at any location in Australia, and produced an Australia
wide map of corrosivity.
38
The simplified model for
estimating airborne salinity in five Australian coastal
zones, presented by Nguyen et al.,
16
was developed
based on this prediction framework.
The climate change projections provide estimates of
likely changes in mean wind speed,
19
and using
equation (6), these changes can be used to estimate
changes of volumetric airborne salinity S
air,vol
.
Model for corrosion due to effect of temperature
As noted previously in the section on ‘Effect of
temperature’, Cole and Paterson
3
have suggested that
‘an increase of 2 K (from 293 to 295 K) will promote a
Table 2 Coefficients for linear relationship between t
wet
and RH for two Australian cities
City
Length of historical
records/years aB R
2
Melbourne 37 0.0183 20.9127 0.838
Brisbane 58 0.0197 21.0315 0.793
Table 1 General circulation models used in this study
Models Developers
CCCMA Canadian Centre for Climate Modelling and Analysis, Canada
CNRM National Center for Meteorological Research, France
CSIRO-Mk3.5 Commonwealth Scientific and Industrial Research Organisation, Australia
GISS-AOM NASA Goddard Institute for Space Studies, USA
GISS-EH NASA Goddard Institute for Space Studies, USA
IAP-FGOALS Institute of Atmospheric Physics, Chinese Academy of Sciences, China
IPSL-CM4 Institute Pierre Simon Laplace, France
MIROC-M Centre for Climate System Research of University of Tokyo, National Institute for Environmental
Studies, and Frontier Research Centre for Global Change of Japan Agency for
Marine-Earth Science and Technology, Japan
MRI-CGCM2.3.2 Meteorological Research Institute, Japan Meteorological Agency, Japan
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
362 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5

0?6% change in corrosion rate’. Fitting this assumption
to equation (2), the increase in corrosion rate with
temperature can be expressed as
c0,T2~c0,T1e260 1=T1
ðÞ{1=T2
ðÞ½ (7)
where c0,T1and c0,T2are the corrosion rates at absolute
temperature T
1
and T
2
respectively. Although the effect
of changes in temperature is expected to be only a minor
secondary effect on corrosion rates, the temperature
effect predicted by equation (7) is simple to evaluate and
so will be considered in this study.
Climate change models used
As mentioned previously, the climate change impact
assessment is carried out considering the A1FI emission
scenario, the highest emission scenario defined by IPCC
(2007). The A1FI scenario assumes a very rapid economic
growth, a global population that peaks in mid-century and
declines thereafter, and a rapid introduction of new and
more efficient technologies with intensive fossil energy
consumption.
This emission scenario is applied to the nine GCMs
listed in Table 1. Each of the GCMs has different
strengths and weaknesses, and IPCC
28
therefore sug-
gests the use of multiple models to take into account the
uncertainties in impact assessments. This set of GCMs
and emissions scenario has also been used previously for
assessing the climate change impacts and adaptations
for housing thermal performance by Wang et al.
39
and
for reinforced concrete structures by Stewart et al.
7
The projected local climate variability in Australia with
the different GCMs is simulated using OzClim, climate
change prediction software specifically developed for
Australia by CSIRO.
40
For the purpose of this study, the
projected local climate includes monthly average tem-
perature, RH and wind speed for the two Australian
cities, Melbourne and Brisbane, which are generated
every 5 years from 1990 to 2100. Projected yearly
averages of the climate parameters are computed from
the projected monthly averages and shown in Fig. 3.
Projections of changes in corrosion rates
In the previous section, it was indicated that there are
three useful parameters related to atmospheric corrosion
that are projected by existing climate change models.
These are the temperature, RH and wind speed. In the
following, these parameters will be integrated into the
atmospheric corrosion model developed by Nguyen
et al.
16
Taking 1990 as the reference year, the climate change
factor due to temperature change, C
temp
, can be defined
using equation (7)
Ctemp~c0,Tprojected
c0,T1990
~e260 1=T1990
ðÞ{1=Tprojected
ðÞ½
(8)
where T
1990
is the absolute yearly average temperature at
year 1990, T
projected
is the projected absolute yearly
average temperature due to climate change in the future
and c0,T1990 and c0,Tprojected are corrosion rate parameters at
T
1990
and T
projected
respectively.
Similarly, by equation (5), the climate change factor
for the time of surface wetness due to the change in RH
Ctwet is defined as
Ctwet ~twet,projected
twet,1990
~a|RHprojectedzb
a|RH1990zb(9)
where RH
1990
is the RH at the reference year 1990.
RH
projected
is the projected RH due to climate change.
t
wet,1990
and t
wet,projected
are the times of surface wetness
at RH
1990
and RH
projected
respectively.
The airborne salinity S
air
in equations (3) and (4) is
stated in terms of deposit on a salt candle
29
and needs to
be related to the volumetric airborne salinity S
air,vol
.
Cole and Corrigan
41
expressed the relationship by
Sair~gSair,vol UA
s(10)
where gis a deposition efficiency factor, and A
s
is the
area of the salt candle surface that the salinity aerosol
impacts on. Using equations (6) and (10), the climate
change factor CSair can be expressed as
CSair ~Sair,projected
Sair,1990
~P
m~1...12
Uprojected,m exp 0:23Uprojected,m

P
m~1...12
U1990,m exp 0:23U1990,m
ðÞ
(11)
where Sair,projected is the projected airborne salinity due to
climate change and Sair,1990 is the airborne salinity at the
reference year 1990. U
1990,m
is the mean wind speed for
month mof the reference year 1990; U
projected,m
is the
projected mean wind speed for month mof the projected
year under climate change. The monthly mean wind
2 Relationships between time of wetness and yearly average of RH in Melbourne and Brisbane based on historical
weather data
(11)
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5363

speeds were used to capture the effect of seasonal
changes of wind speed on the airborne salinity. This is
consistent with the airborne salinity model used by
Nguyen et al.
16
The values of U
projected,m
have been
computed by applying the values shown in Fig. 3 as
modification factors to the values of U
1990,m
. It is also to
be noted that an implicit assumption in all of this is that
the relative change in wind speed will be the same for all
directions, i.e. the changes will not be dominant in a
specific direction.
Using equations (3), (4), (8), (9) and (11) into
equations (3) and (4) leads to C
rate,zinc
and C
rate,steel
,
the projected relative corrosion rates of zinc and steel
Crate,zinc~Ctemp C0:6
twet C0:5
Sair
Crate,steel~Ctemp C0:8
twet C0:5
Sair
(12)
3 Projected changes of temperature, RH and wind speed for Melbourne and Brisbane under A1FI scenario: in each plot,
thin curves are of nine GCMs and thick curve denotes average
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
364 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5

The terms C
rate,zinc
and C
rate,steel
indicate increases in the
corrosion rate when they are .1?0 and indicate
decreases in the corrosion rate when they are smaller
than 1?0, in comparison with the rate in 1990. The
difference between the factors and 1?0 gives the
percentage of the changes in corrosion rates caused by
climate change.
Using equations (8), (9), (11) and (12), the three
climate change factors C
temp
,Ctwet and CSair and the
relative corrosion rates C
rate,zinc
and C
rate, steel
were
estimated every 5 years from 1990 to 2100 using the
climate change effects projected from each of the nine
GCMs under the A1FI emission scenario. Because the
climate projection data, i.e. T
projected
,RH
projected
and
U
projected,m
in particular, are changing with time (Fig. 3),
the climate change factors and the relative corrosion
rates also change with time.
The climate change factors C
temp
,C
wet
and CSair from
1990 to 2100 are shown in Fig. 4. Comparative plots of
these three parameters are shown in Fig. 5; these plots
show the relative influence of the three climate change
factors. Finally, the relative corrosion rates C
rate,zinc
and
4 Climate change factors C
temp
,Ctwet and CSair for Melbourne and Brisbane under A1FI scenario: in each plot, thin curves
are of nine GCMs and thick curve denotes average
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5365

C
rate,steel
from 1990 to 2100 are shown in Fig. 6 and
given in tabulated form in Table 3.
Discussion
Computed changes in corrosion rates
As can be seen in Fig. 5, the most influential parameters
affecting changes in corrosion rates were Ctwet in the case
of Melbourne and CSair in the case of Brisbane. As
shown by equations (9) and (10), these changes relate to
changes in the RH
projected
(projected annual RH) and
U
projected,m
(projected monthly wind speed) respectively.
The effect of temperature on the corrosion rate,
represented by C
temp
, is negligible.
It is of interest to note that for the period 1990–2100,
the projected relative corrosion rates decreased in the
case of Melbourne and increased for Brisbane. On
average, in 2100, the average projected climate change
5 Comparison of average of three climate change factors C
temp
,Ctwet and CSair for Melbourne and Brisbane under A1FI
scenario
6 Effects of climate change on relative corrosion rates C
rate,zinc
and C
rate,steel
for two cities under A1FI scenario: in each
plot, thin curves are of nine GCMs and thick curve denotes average
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
366 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5

impact in Brisbane is increases by ,14% in the relative
corrosion rates of steel and zinc, while the projected
climate change impact on relative corrosion rates in
Melbourne is a reduction of ,14% for steel and 9% for
zinc, as shown in Table 3. However, it is to be noted that
if the worst GCM projections are used, then the
projected relative corrosion rates for the year 2100
increase for both cities, i.e. by ,20% for Melbourne and
40% for Brisbane.
Rainfall patterns
Unfortunately, the various climate change models do
not provide adequate information on the effect of
changes in rainfall patterns to make use of the cutoff
concept
3
mentioned in the section on ‘Effect of changes
in rainfall patterns’ to predict changes in salinity factors.
This change relates to the effect of rain in washing off
accumulated salt.
A lower limit of this effect can be found by comparing
corrosion rates for exposed and sheltered environments.
From the Australian model,
16
the corrosion rate of
exposed elements, relative to that of sheltered element, is
derived to be ,0?7; from the field data by King and
O’Brien,
42
a typical measured ratio would be ,0?6.
These are extreme cases, giving 30 or 40% increase in the
corrosion rate when changing from elements that are
fully exposed compared to those that are fully sheltered
from rain. Hence, it would not be unreasonable to
expect that the increase in corrosion due to changes in
rainfall patterns would be of the order of 10% for
exposed structures. This is comparable with the pro-
jected changes due to other climate parameters discussed
in the section on ‘Projections of changes in corrosion
rates’.
Moreover, as the reduction in corrosion for exposed
constructions is related to the washing effects of rainfall,
it is unlikely that engineers will make use of this
‘exposure factor’ in the design of significant structures
since it would be unreliable to assume that total
exposure to rainfall will be maintained indefinitely.
Uncertainties
In any estimate of climate change effects, it is also
important to estimate the uncertainties associated with
the estimate. Some idea of the errors due to climate
change estimates can be obtained by examining the plots
shown in Fig. 6. With some allowance for the variability
associated with each GCM model, the uncertainty in
projection of the relative corrosion rates for 2100 would
correspond to a coefficient of variation of ,30%.
This degree of uncertainty should be placed in the
context of the uncertainty of corrosion modelling that
currently exists when climate change effects are not
considered. In the application of the Australian model to
practical design situations, a value of 200% is recom-
mended to be used for the coefficient of variation in
estimating corrosion depths.
16
This is an uncertainty
considerably larger than the 30% estimated to account
for in the application of climate change models.
Corrosion of inland constructions
Under climate change, the atmospheric corrosion rate of
steel structures located in the inland areas could possibly
increase due to increases of airborne salinity levels, as
estimated by Cole and Paterson.
3
However, the airborne
salinity inland is dominated by ocean produced aerosol,
which at the coastline is ,25 times lower than that of the
surf produced aerosol for an open surf location.
16
Hence, the atmospheric corrosion of steel inland would
be considerably less than that near the coastline and
would not normally be a major durability concern.
Depth of corrosion
Because the climate change factors are changing with
time, and thus the corrosion rate, the corrosion depth up
to an in-service duration tis a cumulative corrosion with
a changing corrosion rate parameter during the time.
However, there is only limited available data that is
useful for developing models of corrosion depth under
changing environment conditions. In view of the
relatively small changes that are projected, it would
probably be adequate to use a constant value of
corrosion rate parameter c
0,av
, an average value to
compute the relative corrosion rate over the in-service
duration under climate change conditions. The average
corrosion rate parameter c
0,av
can be computed from
c0,av~c0
yc{y0ð
yc
y0
Crate,metal(t)dt(13)
where c
0
is the corrosion rate parameter without climate
change effects, tis a time variable, y
0
is the year of
installation of the structure, y
c
is the final year of
consideration of the corrosion and C
rate,metal
is the
relative corrosion rate for zinc or steel as defined in
equation (12). The difference between y
c
and y
0
is
therefore the in-service duration. This computation
approach is similar to that used by Stewart et al. for
estimating the carbonation process in reinforced concrete.
Conclusions
A review of corrosion processes has indicated that the
relevant science is complex and full of uncertainty. It
Table 3 Average effects of climate change on relative corrosion rates at certain years for Melbourne and Brisbane under
A1FI scenario*
City Relative corrosion rates Year 2030 Year 2050 Year 2070 Year 2100
Melbourne C
rate,zinc
(0.927–1.047) (0.853–1.096) (0.761–1.154) (0.642–1.216)
0.973 0.949 0.926 0.907
C
rate, steel
(0.903–1.032) (0.806–1.063) (0.687–1.096) (0.539–1.121)
0.963 0.929 0.893 0.858
Brisbane C
rate,zinc
(0.989–1.062) (0.978–1.135) (0.964–1.235) (0.948–1.373)
1.020 1.045 1.083 1.139
C
rate, steel
(0.980–1.065) (0.959–1.140) (0.934–1.244) (0.903–1.388)
1.019 1.045 1.082 1.138
*Values in bracket are maximum and minimum given with different GCMs.
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5367

indicated that care is required in examining the context
in which research conclusions have been made. When
these facts are added to the difficulties in the use of
climate change models, it is realised that there is a
considerable complexity and scope for error in making
predictions of future corrosion rates.
Within this paper, an attempt has been made to go as
far as possible within the available science, to make
predictions on future corrosion rates. To do this,
existing corrosion and climate change models have been
used. Specifically, an existing Australian corrosion
model and nine GCMs under the climate change
scenario A1FI were used to predict changes in corrosion
rates from 1990 to the year 2100. To achieve clarity in
presentation, pollution effects were neglected and the
applications were limited to the coastal areas of only two
Australian cities: Melbourne and Brisbane.
The output parameters of the climate change models
that were found to be useful were the projections of RH,
wind speed and temperature. Of these, temperature was
found to be of very minor significance. Using these
parameters, it was found that on average the relative
corrosion rates were projected to have a decrease of ,9%
(for zinc) and 14% (for steel) for Melbourne, and an
increase of ,14% for Brisbane in 2100 for both metals.
A major deficiency in the climate projection models
was the availability of quantitative data on the rainfall
patterns. This would have provided information on the
beneficial effects of rain wash on exposed constructions.
It is anticipated that this effect will lead to an increase of
,10% in the predicted corrosion rates for exposed
constructions. A refinement of current climate predic-
tion models to assist in these predictions would be
useful.
Finally, the matter of uncertainty in climate predic-
tion forecasts was examined. The data appear to indicate
the uncertainty in the prediction of corrosion rates, i.e.
equivalent to a coefficient of variation ,30%, whereas
the corresponding uncertainty in the Australian model
to predict corrosion rates (excluding climate change
effects) would be a coefficient of variation of ,200%.
Hence, for engineering design purposes, the uncertain-
ties associated with climate change models are of
secondary importance.
Acknowledgements
The authors would like to express their appreciation for
the support by CSIRO Climate Adaptation Flagship
and the National Climate Change Adaptation Research
Facility under the project ‘Pathways to Climate Adapted
and Healthy Low Income Housing’. The authors would
also like to thank K. Hennessy and J. Clarke of CSIRO
Marine and Atmospheric Research and R. Jones of
Centre for Strategic Economic Studies at Victoria
University for their assistance and advice on climate
projections using OzClim.
References
1. G. H. Koch, J. H. Payer, M. P. H. Brongers, N. G. Thompson and
Y. P. Virmani: ‘Corrosion costs and preventive strategies in the
United States’, Technical Summary Report, Publication
No. FHWA-RD-01-156, Federal Highway Administration. http://
www.nace.org/uploadedFiles/Publications/ccsupp.pdf (accessed
August 2012).
2. S. Ratcliff: ‘Research shows corrosion costs the local economy’,
Curtin News, Curtin University. http://news.curtin.edu.au/news/
research-shows-corrosion-costs-the-local-economy/ (accessed August
2012).
3. I. S. Cole and D. A. Paterson: ‘Possible effects of climate change on
atmospheric corrosion in Australia’, Corros. Eng. Sci. Technol.,
2010, 45, (1), 19–26.
4. P. R. Roberge: ‘Impact of climate change on corrosion risks’,
Corros. Eng. Sci. Technol., 2010, 45, (1), 27–33.
5. T. E. Graedel and C. Leygraf: ‘Scenarios for atmospheric corrosion
in the 21st century’, Electrochem. Soc. Interface, 2001, 24–30.
6. E. Bastidas-Arteaga, A. Chateauneuf, M. Sanchez-Silva, Ph.
Bressolette and F. Schoefs: ‘Influence of weather and global
warming in chloride ingress into concrete: a stochastic approach’,
Struct. Saf., 2010, 32, 238–249.
7. M. Stewart, X. Wang and M. Nguyen: ‘Climate change impacts
and risks of concrete infrastructure deterioration’, Eng. Struct.,
2011, 33, 1326–1337.
8. X. Wang, M. G. Stewart and M. Nguyen: ‘Impact of climate
change on corrosion and damage to concrete infrastructure in
Australia’, Clim. Change, 2012, 110, (3–4), 941–957.
9. M. Benarie and F. L. Lipfert: ‘A general corrosion function in
terms of atmospheric pollutant concentrations and rain pH’,
Atmos. Environ., 1986, 20, (10), 1947–1958.
10. M. Feliu, M. Morcillo and S. Feliu, Jr: ‘The prediction of
atmospheric corrosion from meteorological and pollution para-
meters. II: long-term forecasts’, Corros. Sci., 1993, 34, (3), 415–422.
11. D. de la Fuente, J. G. Castan˜o and M. Morcillo: ‘Long-term
atmospheric corrosion of zinc’, Corros. Sci., 2006, 49, 1420–1436.
12. R. E. Melchers: in ‘Modeling of corrosion of steel specimens in
natural waters’, (ed. R. M. Kain and W. T. Young). Vol. 2, 20–33;
1997, Philadelphia, PA, American Society for Testing of Materials.
13. R. E. Melchers: ‘Advances in mathematical modelling of the
atmospheric corrosion of structural steel in ocean environments’,
Proc. 3rd Int. ASPRANet Colloq., Glasgow, UK: ASRANet Ltd.,
July 2006, paper 037–005. http://maritime-conferences.com/asranet
2010-conference/asranet2006/PDF/037_005_MelchersASRANet06-
full.pdf.
14. M. Feliu, M. Morcillo and S. Feliu, Jr: ‘The prediction of
atmospheric corrosion from meteorological and pollution para-
meters. I: annual corrosion’, Corros. Sci., 1993, 34, (3), 403–414.
15. D. de la Fuente, I. Dı´az, J. Simancas, B. Chico and M. Morcillo:
‘Long-term atmospheric corrosion of mild steel’, Corros. Sci., 2011,
53, (2), 604–617.
16. M. N. Nguyen, R. H. Leicester, C.-H. Wang and G. C. Foliente:
‘Corrosion effects in the structural design of metal fasteners in
timber constructions’, Struct. Infrastruct. Eng., 2011, 9, (3), 276–
285.
17. ‘Hot-dip galvanized (zinc) coatings on fabricated ferrous articles’,
Standards Australia, Sydney, NSW, Australia, 2006.
18. Corrosion Doctor: ‘Factor affecting atmospheric corrosion’. http://
www.corrosion-doctors.org/Corrosion-Atmospheric/Factors-atmo
spheric.htm (accessed August 2012).
19. CSIRO: ‘Climate change in Australia’, Technical Report 2007,
CSIRO Marine and Atmospheric Research, Aspendale, Vic.,
Australia. http://www.climatechangeinaustralia.gov.au/technical_
report.php (accessed August 2012).
20. A. M. G. Pacheco and M. G. S. Ferreira: ‘An investigation of the
dependence of atmospheric corrosion rate on temperature using
printed-circuit iron cells’, Corros. Sci., 1994, 36, (5), 797–813.
21. W. J. D. Shaw and J. I. Anderson: ‘Atmospheric corrosion of
carbon steel in the prairie regions’, Proc. Northern Area Western
Conf., Calgary, Alta., Canada, February 2010, NACE
International. http://www.nacecalgary.ca/pdfs/NorthernPapers/
Atmospheric%20Corrosion.pdf.
22. NACE: ‘Temperature influence on corrosion’, NACE Resource
Center. http://events.nace.org/library/corrosion/design/tempera
ture-effects.asp (accessed August 2012).
23. P. R. Roberge: ‘Handbook of corrosion engineering’; 1999, New
York, McGraw-Hill.
24. A. I. Almarshad and S. Syed: ‘Atmospheric corrosion of galvanized
steel and aluminium in marine and marine-industrial environments
of Saudi Arabia’, Mater. Corros., 2008, 59, (1), 46–51.
25. R. Lindstrom, J.-E. Svensson and L.-G. Johansson: ‘The atmo-
spheric corrosion of zinc in the presence of NaCl – the influence of
carbon dioxide and temperature’, J. Electrochem. Soc., 2000, 147,
(5), 1751–1757.
26. IPCC: in ‘Emission scenarios – special report of the
Intergovernmental Panel on Climate Change’, (ed. N.
Nakicenovic and R. Swart); 2000, Cambridge, Cambridge
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
368 Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5

University Press. http://www.ipcc.ch/ipccreports/sres/emission/
index.php?idp542#anc1 (accessed August 2012).
27. S. C. Chung, S. L. Sung, C. C. Hsien and H. C. Shih: ‘Application
of EIS to the initial stages of atmospheric zinc corrosion’, J. Appl.
Electrochem., 2000, 30, 607–615.
28. IPCC: ‘Climate change 2007 – synthesis report – an assessment of
the Intergovernmental Panel on Climate Change, IPCC, Geneva,
Switzerland’. http://www.ipcc.ch/pdf/assessment-report/ar4/syr/
ar4_syr.pdf (accessed August 2012).
29. I. S. Cole, W. D. Ganther, D. A. Paterson, G. A. King, S. A.
Furman and D. Lau: ‘Holistic model for atmospheric corrosion.
Part 2: experimental measurement of deposition of marine salts in a
number of long range studies’, Corros. Eng. Sci. Technol., 2003, 38,
(4), 259–266.
30. ‘Atmospheric corrosivity zones in Australia’, Standards Australia,
Sydney, NSW, Australia, 2008.
31. G. A. King, K. G. Martin and J. F. Moresby: ‘A detailed
corrosivity survey of Melbourne’, CSIRO Division of Building
Research Report, Highett, Vic., Australia, 1982.
32. G. A. King, W. D. Ganther and I. S. Cole: ‘Studies at sites
progressively inland from the coast to aid development of a
geographic information system map of Australian corrosivity’,
Proc. Conf. on ‘Corrosion and prevention-99’, Sydney, NSW,
Australia, November 1999, Australasian Corrosion Association
Inc., Paper 39-041.
33. G. A. King and B. Carberry: ‘Atmospheric corrosivity of the
Greater Newcastle Region’, Technical Report no. TR92/1, CSIRO
Division of Building Construction and Engineering, Highett, Vic.,
Australia, 1992.
34. I. S. Cole: ‘Mechanism of atmospheric corrosion in tropical
environments’, in ‘Marine corrosion in tropical environments’,
(ed. S. W. Dean et al.), 33–47; 2000, West Conshohocken, PA,
American Society for Testing and Materials.
35. M. N. Nguyen, R. H. Leicester and C.-H. Wang: ‘Manual no. 5:
atmospheric corrosion of fasteners in timber structures’, 2008,
Highett, Vic., CSIRO Sustainable Ecosystems.
36. W. A. McKay, J. A. Garland, D. Livesley, C. M. Halliwell and
M. I. Walker: ‘The characteristics of the shore-line sea spray
aerosol and the landward transfer of radionuclides discharged to
coastal sea water’, Atmos. Environ., 1994, 28, (20), 3299–
3309.
37. I. S. Cole, W. Y. Chan, G. S. Trinidad and Paterson: ‘Holistic
model for atmospheric corrosion. Part 4: geographic information
system for predicting airborne salinity’, Corros. Eng. Sci. Technol.,
2004, 39, (1), 89–96.
38. I. S. Cole, G. A. King, G. S. Trinidad, W. Y. Chan and D. A.
Paterson: ‘An Australia-wide map of corrosivity: a GIS approach’,
Proc. Int. Conf. on ‘Durability of building materials and compo-
nents’, Vol. 2, 901–911; 1999, Vancouver, BC, NRC Research
Press.
39. X. Wang, D. Chen and Z. Ren: ‘Assessment of climate change
impact on residential building heating and cooling energy require-
ment in Australia’, Build. Environ., 2010, 45, 1663–1682.
40. CSIRO: ‘OzClim – exploring climate change scenario for
Australia’, http://www.csiro.au/ozclim/home.do (accessed August
2012).
41. I. Cole and P. Corrigan: ‘Service life prediction of materials’, in
‘Technology, design and process innovation in the built environ-
ment’, (ed. P. Newton et al.), 72–96; 2009, New York, Taylor &
Francis.
42. G. A. King and D. J. O’Brien: ‘The influence of marine
environments on metals and fabricated coated metal products,
freely exposed and partially sheltered’, in ‘Atmospheric corrosion’,
ASTM STP 1239, (ed. W. W. Kirk and H. H. Lawson), 167–192;
1995, Philadelphia, PA, American Society for Testing and
Materials.
Nguyen et al. An assessment of climate change effects on atmospheric corrosion rates of steel structures
Corrosion Engineering, Science and Technology 2013 VOL 48 NO 5369