Documente Academic
Documente Profesional
Documente Cultură
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/281750966
CITATIONS READS
0 81
2 authors, including:
Mark G. Stewart
University of Newcastle
238 PUBLICATIONS 3,852 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Climate change effects and adaptation of civil infrastructure and buildings View project
All content following this page was uploaded by Mark G. Stewart on 15 September 2015.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
CONCRETE INSTITUTE
CONCRETE IN AUSTRALIA
of AUSTRALIA
www.concreteinstitute.com.au
SPECIAL FEATURE
■ DETERIORATION OF
REINFORCED CONCRETE
VOL 41, NO 3
CONCRETE 2015
SEPTEMBER 2015
An increase of CO2 concentration in the atmosphere and changes in temperature and humidity may cause an acceleration of
deterioration processes of reinforced concrete (RC) structures as a result of climate change. This paper will describe cost-benefit
analysis of climate adaptation strategies for new RC structures in Australia. Spatial time-dependent reliability analysis is used
to predict likelihood and extent of carbonation-induced corrosion damage and corresponding damage costs. The cost-benefit
analysis includes the estimation of costs of adaptation strategies and corrosion damage, as well as cost effectiveness of adaptation
strategies for RC buildings in three Australian cities (Sydney, Canberra and Brisbane) under two climate emission scenarios.
Adaptation strategies include increasing concrete cover and upgrading concrete strength. The results show that upgrading concrete
strength for beams and small slabs may be more cost effective than the current durability design provisions. Break-even analysis
is also applied to estimate the maximum adaptation cost to make an adaptation strategy cost effective.
Figure 2: Projected annual average temperatures and relative humidity for the six GCM projections for RCP 8.5, RCP 4.5 and Year 2015 emission scenarios.
Figure 3: fT (t) and fRH(t) value of three cities under RCP 4.5 emission scenario.
of changes in temperature and relative humidity in three cities. where t is defined in calendar years starting from 2015,
The average trend of temperature increases in the three cities assuming the buildings go into service in the year 2015;
following the global warming projection, while the mean trend CCO2(t) is the time-dependent increase in atmospheric CO2
of relative humidity decreases. All the projections based on the concentration (10-3 kg/m3); ksite is a factor to account for
six climate models and two emission scenarios will be used in increased CO2 levels in non-remote environments; fT(t) is
the simulation of concrete deterioration under climate change. time-dependent change in diffusion coefficient due to changes
in temperature; fRH(t) is time-dependent change in diffusion
2.2 Deterioration and spatial time-dependent coefficient due to changes in RH; the age factor nm for
modelling microclimatic conditions related to the frequency of wetting
Corrosion-induced concrete cover cracking and damage can be and drying cycles and equals to zero for sheltered outdoor, and
found along the steel rebar. When the concrete crack reaches a nm = 0.12 for unsheltered outdoor exposures. The functions
limit crack width, the concrete structure is severely damaged. for ksite, fT(t) and fRH(t) are detailed in Peng and Stewart5, 6.
The various stages of carbonation-induced corrosion process Generally, fT(t) will increase when temperature rises, however,
can be described as: (i) Ti – time to corrosion initiation; (ii) fRH(t) decreases when RH becomes higher. The fT(t) and fRH(t)
t1st – time from corrosion initiation to crack initiation (visible value of three cities under RCP 4.5 are shown in Figure 3.
crack widths of 0.05 mm); (iii) tsev – time for a crack to It can be found that temperature increase caused by climate
propagate from crack initiation to a limit crack width. The change will have an adverse impact on carbonation induced
time to severe corrosion damage is thus Tsev= Ti +t1st +tsev. corrosion. The site correction parameter ksite of suburban
areas will be applied to CO2 concentrations projections in
2.2.1 Corrosion initiation Canberra and outer suburbs of Sydney and Brisbane (see
Table 3). DCO2(t) is time-dependent CO2 diffusion coefficient
The preliminary models of carbonation depth developed by
in concrete; D1 is CO2 diffusion coefficient at t = 2015; nd
Duracrete14, Yoon et al15 and Stewart et al3 are improved by
is the age factor for the CO2 diffusion coefficient; t0 is one
four aspects: (i) CO2 levels, temperature and RH are time-
year; a is binding capacity relies on the degree of hydration of
dependent variables; (ii) a correction factor ksite is introduced
the cement; Ce is cement content in concrete (kg/m3); CaO is
to take into account increased CO2 levels in non-remote
CaO content in cement (0.65); αH is a degree of hydration for
environments; (iii) time-dependent change in the diffusion
ordinary Portland cement (OPC) after more than 400 days16;
coefficient due to changes in temperature; and (iv) time-
MCaO is molar mass of CaO and equal to 56 g/mol; and MCO2
dependent change in the diffusion coefficient due to changes
is molar mass of CO2 equal to 44 g/mol.
in RH. The carbonation penetration process is simulated as
a steady state diffusion process, and the carbonation depth at
time t is5: 2.2.2 Corrosion propagation
When carbonation depth reaches the reinforcing bar, corrosion
2 DCO2 (t ) t
t0nm initiates (Ti) where Ti is the year when the carbonation depth
xc (t ) = ∫ fT (t ) f RH (t )ksiteCCO 2 (t )dt
a 2015
t − 2014 exceeds the concrete cover. The corrosion rate is variable and
(1) highly dependent on exposure conditions and atmospheric
t ≥ 2015
situations. Because the corrosion reactions consume oxygen
DCO2 (t ) = D1 (t − 2014)
− nd
(2) and water, the reactions will be slowed down when RH is
either too high or too low. A very high humidity in concrete
M CO2 −3.38 w/ c
may reduce oxygen diffusion through the concrete while
a = 0.75CeCaO αH αH ≈ 1 − e (3) water shortage in concrete also declines the corrosion rate.
M CaO The corrosion rate will become negligible when RH is less
Class w/c ratio Ce a (kg/m3) Mean D1b (× 10-4 cm2/s) Dbar (mm) Cover (mm) F’c (MPa)
A2 0.56 320 2.22 12 30 25
Note – a: from AS 5100.527. b: from Yoon et al.15.
than about 50%17, 18. The peak RH for corrosion is 70-80%, of identical elements of size Δ and the random variables
and then the corrosion rate decreases as RH increases18. An within the random field are statistically correlated based on the
empirical model that includes both RH and temperature corresponding correlation function24. The midpoint method is
effects on the corrosion rate is proposed by Breysse et al19. used that an element is represented by the value at the centroid
The corrosion rate at a reference state is assumed to be of that element, and this value is assumed to be constant within
log-normally distributed with statistical parameters shown the element. The validation and calibration of correlation
in Table 3 for moderate humidity sheltered exposure, and functions demand large amounts of spatial data that is hard to
the temperature and RH effects on the corrosion rate are achieve. Therefore, engineering judgement or expert experience
modelled as Breysse et al19 model. For more details see Peng is required to match a correlation function to a particular
and Stewart6. random field. The scale of fluctuation (θ) defines the distance
over which correlation persists in a random field. Once the
2.2.3 Crack initiation and propagation stochastic random field is defined, Monte Carlo simulation
The time to crack initiation is estimated by a model proposed methods are used to generate random variables for each
by El Maaddawy and Souki20. Note that the accuracy of the element, and the spatially variable parameters of each element
time to corrosion damage is dominated by the accuracy of are then defined using the correlation function. This allows the
time to corrosion initiation and the time to crack propagation, proportion of the concrete surface to experience severe corrosion
and so damage risk predictions are relatively insensitive to the damage at time t (dcrack(t)) to be calculated.
crack initiation model21. The definition of how ‘excessive’ the
crack width depends on individual conditions and asset owner 2.4 Durability design specifications and statistical
policies; in this paper, concrete structures will be regarded as parameters for RC structures
‘severely cracked’ when crack width reaches a crack width of The exposure classifications for RC structures in the three
1.0 mm22. The crack initiation and propagation models used Australian cities are near-coastal (B1: 1-50 km from coast)
in the simulation analysis are influenced mainly by concrete and coastal excluding tidal and splash zones (B2: <1 km
strength, concrete cover and corrosion rate. from coast) for the most parts of Sydney and Brisbane, and
A2 (>50 km from coast) for Canberra and outer suburbs of
2.3 Spatial time-dependent reliability model Sydney and Brisbane. Design cover for B1 is 40 mm, which
The inherent variability of RC structures due to material is higher than 30 mm cover for A2. Note that RC structures
properties and quality control in construction can result in close to the shore (B1 and B2) have higher concrete cover
corrosion damage being spatially variable. Moreover, modelling and are not susceptible to carbonation induced corrosion.
material and dimensional parameters as homogeneous can lead Therefore, only RC structures of exposure A2 are studied here,
to non-conservative predictions of failure for RC structures in as shown in Table 1.The geometry of RC building slabs is
corrosive environments23. Also, a spatial analysis can provide assumed to be 6 m × 6 m two-dimensional random field, with
more detailed information on the extent of damage at a given element size Δ=0.5 m. A RC beam with size of 0.5 m × 0.5 m
point in time to be estimated, allowing detailed analysis of × 6 m can be discretised into a one-dimensional random field.
maintenance strategies and damage cost estimation. As such, For review details about discretisation methods, the correlation
incorporating spatial variability into predictive models is function, scale of fluctuation and element sizes, etc, see
important, particularly for deterioration processes. The spatial Vanmarcke24 and Peng 25. The spatial and spatially dependent
variability of concrete cover, concrete strength, diffusion random variables, standard random variables and deterministic
coefficient and binding capacity are included in the analysis6. variables, as well as the model errors used in the spatial time-
Random field theory is applied to model the spatial variability dependent reliability analysis, are summarised in Tables 2 and
of these parameters that a surface is discretised into a number 3. The results for RC slabs can be applied to floors, panels
Table 3: Statistical parameters for corrosion parameters, material properties and dimensions.
40 MPa 1.07F’c f
0.151 Truncated normal b 29
Notes – a: truncated at 8 mm. b: truncated at 0 MPa. c: truncated at 1.0. d:1 μA/cm2 = 0.0116 mm/year. e: Cnom is the nominal or
design cover. f: F’c is the nominal design concrete compressive strength.
and walls, etc. Note that RC structures are assumed to use be carefully considered when defining the inspection interval.
OPC concrete, and standard formwork and compaction. For an RC building, such as a residential apartment which
can be easily inspected, any corrosion damage can be noticed
3.0 MAINTENANCE STRATEGIES quickly, therefore the routine inspection interval Δt = 1 year34.
A patch repair maintenance strategy will be used for The damage limit state, Xrepair (also referred to as the repair
comparison purposes to study the impact of climate threshold), is usually over a range of practical values from
adaptation strategies. Patch repairs are the most common 0.5% to 5% for patch repairs, and 12% and 20% for complete
type of repair for corrosion damaged RC structures31, 32. It rehabilitative overlay35. In this study, the minimum repair area is
is a corrective repair strategy in which repair takes place 0.5 m × 0.5 m for RC slabs and beams. Therefore, Xrepair is 0.7%
after severe concrete cracking, but the loss of the cross- and 8.3% for RC slabs and beams, respectively. Patch repair is
sectional area of rebars is not significant. The patch repair assumed to be carried out immediately if the extent of corrosion
process involves the mechanical removal of the damaged damage has been discovered to exceed the repair threshold Xrepair
concrete (typically to approximately 25 mm beyond the at the time of ith inspection iΔt. In other words, all observed
rebar), the cleaning and treatment of the corroded steel, and corrosion damaged areas will be repaired immediately after
reinstatement of the concrete cover with a suitable material. inspection.
The use of patch repairs in a maintenance strategy can be It is assumed that (i) severe corrosion induced cracking is
modelled very effectively using a spatial time-dependent always detected when the structure is visually inspected; (ii)
reliability analysis as both the likelihood and extent of damage only the damaged area will be repaired; the remaining RC
can be predicted33. Therefore, the patch repair maintenance surface area will continue to deteriorate; (iii) repair will not
technique defined herein will be integrated into the cost- improve durability performance of the repaired structures; ie
benefit analysis of an adaptation strategy. A maintenance cover and concrete quality are the same as the original design
strategy includes: (i) inspection interval; (ii) damage limit specification; and (iv) damage will not re-occur for the repaired
state; (iii) maintenance technique. area during the remaining service life of the structure. This last
The inspection interval Δt is usually based on the controlling assumption is slightly non-conservative for estimating corrosion
authority’s policy or the prior experience of the asset owner/ damage losses; however, Peng25 shows less than 0.2% probability
operator. A longer inspection interval will save on inspection that damage may reoccur once repaired.
costs but can mean that when detected, the damage to the
structure has exceeded the pre-defined repair threshold and 4.0 ADAPTATION STRATEGIES
hence results in a more costly repair area. Further, at longer A wide range of methods can be used to enhance the
inspection intervals, there is a risk of the corrosion damage durability of RC structures, and these can be applied to
reducing the structural safety of the RC structure, and this must reduce the adverse effects of a changing climate. The design
Table 4: Durability design requirements of Australian RC buildings for BAU and four adaptation strategies.
Table 5: Costs of four adaptation strategies and damage for RC structural elements in Australia.
Costs Structural element D (mm) C1: + 5 mm C2: + 10 mm S1: + 1 grade S2: + 2 grades
Slabs – small 100 8.7 17.3 0.5 1.1
Cadapt ($/m2) Slabs – large 250 5.2 10.3 1.3 2.8
Beams 500 7.9 15.9 2.5 5.5
Cdamage ($/m )2
1000
Slabs – small 100 0.0087 0.0173 0.0005 0.0011
Cadapt / Cdamage Slabs – large 250 0.0052 0.0103 0.0013 0.0028
Beams 500 0.0079 0.0159 0.0025 0.0055
2070:
2080:
2090:
2100:
Corrosion not initiated. Corrosion initiated. Crack initiated. Severe corrosion damage.
Figure 4. Simulation of spatially distributed corrosion process showing three typical Monte Carlo realisations for
sheltered cast in-situ RC slab in Sydney under RCP 8.5 emission scenario.
Figure 6: Expected damage costs ($/m2) of BAU and four adaptation strategies for RC buildings in Sydney and Brisbane under RCP 8.5 emission scenario.
for those RC structural components having a small extent of S2 reduces the effective depth of RC beams by 1% to 3%,
corrosion damage. On the other hand, even though increased or the amount of tensile steel by 2% to 6%. For RC beams,
concrete cover (C1 and C2) reduces extent of damage and Table 5 shows that a 1% change of depth (5 mm) costs
damage cost more, Cadapt values are too high to make it cost $7.90/m2. Hence, the minimum co-benefit of reduced RC
effective. beam depth is $4/m2 and $8/m2 for S1 and S2, respectively.
The likelihood and extent of corrosion damage, cost of To be sure, there may be design outcomes that do not change
damage, and adaptation costs are subject to considerable if concrete grade is increased; however, if only 50% of designs
uncertainty and are country, site and structure specific. For result in a more efficient use of materials due to a higher
this reason, calculations of corrosion risks, costs and benefits concrete grade then co-benefit will still easily exceed $2/m2
will be imprecise. Hence, a break-even analysis may be useful and $4/m2 for S1 and S2, respectively. If a lower bound co-
to estimate the maximum adaptation costs required to make benefit of ΔB=$2/m2 for S1 is added to the mean NPV results
the benefit of an adaptation strategy equal to its cost. In other in Table 6 then mean NPV for RC beams in Brisbane would
words, if adaptation costs exceed the predicted break-even value, equal +$0.2/m2 and +$0.1/m2 under RCP 8.5 and RCP 4.5
then adaptation is not cost-effective. In this case, adaptation emission scenarios, respectively. Hence, it is highly likely that
adaptation strategy S1 is cost-effective for Brisbane. However,
costs are normalised by the ratio Cadapt/Cdamage to reflect
adding a co-benefit of ΔB=$4/m2 for S2 to the mean NPV
uncertainty in Cadapt and Cdamage. Table 7 shows the mean break-
results in Table 6 would still result in a net loss for all cities.
even value for Cadapt/Cdamage for RC structures under all emission
For RC slabs, an increase in concrete grade to 32 MPa
scenarios.
(S1) can reduce the effective depth or tensile steel by 0.3%
to 1.1%. Adaptation strategy S2 reduces the effective depth
6.2 Consideration of co-benefit
or tensile steel of RC slabs by 0.5% to 1.9%. For RC slabs,
The co-benefit (ΔB) for adaptation measures S1 and S2 can Table 5 shows that a 1% change of depth costs $1.74/m2 and
be considerable. An increase in concrete grade to 32 MPa $3.12/m2 for small and large slabs, respectively. Hence, the
(S1) can reduce the effective depth of RC beams by 0.5% to minimum co-benefits for small and large RC slabs designed
2%, or the amount of tensile steel by 1% to 4% whilst still to S1 are $0.50/m2 and $0.95/m2, respectively. The minimum
maintaining flexure and shear capacities. Adaptation strategy co-benefits for small and large RC slabs designed to S2 are