Available online at www.sciencedirect.

com

Cement & Concrete Composites 30 (2008) 113–121

www.elsevier.com/locate/cemconcomp

Factors influencing chloride transport in concrete structures

exposed to marine environments
Ha-Won Song *, Chang-Hong Lee, Ki Yong Ann
School of Civil and Environmental Engineering, Yonsei University, Seoul 120-749, Republic of Korea

Received 4 May 2006; received in revised form 30 August 2007; accepted 18 September 2007
Available online 25 September 2007

Abstract

The present study concerns a literature review regarding chloride transport in terms of the diffusion coefficient (D) and surface chlo-
ride content (CS) in concrete structures exposed to a marine environment. In addition, a refined model is proposed for D and CS for time-
dependent chloride transport. It is found that both D and CS are time-dependent: D decreases and CS increases with time, due to further
cement hydration development. D and CS are significantly influenced by concrete mix proportion, air void content in concrete, curing
methods, degree of exposure to seawater and climate. Blended cement concretes are beneficial in decreasing D, but increase CS, presum-
ably due to a refinement of the concrete pore structures and chloride binding. An increase in the air void content in concrete increases D,
but decreases CS. The CS depends on curing methods such as water, membrane and aeration. The degree of exposure to seawater does
not have a clear relationship with CS: higher exposure levels did not result in a higher level of CS. Exposure in tropical areas results in
higher CS in concrete jetty structures with a similar level of D, mainly due to an increased temperature and concentration of salinity.
Based on these findings, a refined model for CS was proposed to provide more realistic prediction of the build-up of chlorides by chloride
transport in concrete structures exposed to marine environments.
 2007 Elsevier Ltd. All rights reserved.

Keywords: Chloride transport; Diffusion coefficient; Surface chloride; Concrete structures; Marine environment

1. Introduction The steel in concrete is protected by an oxide passive

film generated in a highly alkaline environment [3],
Reinforced concrete is usually durable and cost effective, although the corrosion cell is fully composed: cathodic
which has resulted in its widespread use for the construc- molecules such as moisture and oxygen, anode, electrolyte
tion of semi-permanent structures. However, it has become and electric circuit. However, a build-up of chloride at the
increasingly apparent that attack by aggressive agents such depth of the steel accompanied by a local fall in the pH of
as chloride ions, leading to corrosion of embedded steel, the pore solution depassivates the protective film prior to
may cause a structure to deteriorate. Thus, the corrosion steel corrosion [4]. To prevent the chloride permeation into
of reinforcing steel in concrete due to chloride transport concrete, the use of finely grained materials [5–7] and
in concrete structures in a marine environment has received resin-based treatment on the cover concrete [8] have been
increasing attention in recent years because of its wide- investigated and achieved some success.
spread occurrence and the high cost of repair [1,2]. Simultaneously, investigation on predicting the chloride
profile in concrete exposed to a marine environment has
been performed using the Fick’s second law. Many authors
*
Corresponding author. Tel.: +82 (0)2 2123 2806; fax: +82 (0)2 364
have used this model to predict the time to chloride-
5300. induced corrosion with a given diffusion coefficient (D)
E-mail address: [email protected] (H.-W. Song). and surface chloride content (CS), assuming that cement

0958-9465/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconcomp.2007.09.005
114 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121

matrix in concrete is chemically inert and evenly porous [9– the convenience by acid soluble extraction in a nitric acid
12]. However, this hypothesis and models have been solution followed by potentiometric titration against silver
challenged, since Bamforth and Price [13] addressed the nitrite. The chloride concentration of each solution sample
time-dependent characteristic of chloride transport in is often given in ppm, which is then converted to the per-
terms of D and CS, and Bentz et al. [14] showed a change centage by weight of concrete and finally the percentage
of D with time. of chloride by weight of cement, assuming a uniform distri-
The present study surveys published data on chloride bution of cement through the cover concrete. This gives the
transport of concrete specimens or structures exposed to chloride profile at the time when the specimens are
a marine environment (i.e. aerated, splash and tidal zones) removed from concrete structures in seawater or saline
for the duration from 1 to 64 years. Time-dependent char- environment.
acteristics of D were extensively investigated altogether The D can be determined by fitting the error function
with other factors influencing chloride transport such as solution to Fick’s second law, for non-steady state diffusion
concrete mix proportion, curing method, degree of expo- in a semi-infinite medium, given by Eq. (1). In the analysis
sure to seawater and climate, while a change in CS has been of the chloride profile, the sum of the squared differences
rarely dealt with. The present study suggests a refined between the fitted (theoretical) and the actual data for
model for chloride transport considering time-dependency the chloride content of each sample must be minimised
of the CS along with extensive investigation on CS. by adjusting the regressor variable. Once a best fit is
achieved, the D can be calculated. The initial chloride
2. Chloride transport model content (i.e. from the mix ingredients) is usually very low
and thus is neglected in analyzing the chloride profiles.
2.1. Diffusion model D can be calculated from a measurement of the chloride
concentrations at each depth of concrete, whereas a direct
There are a number of mechanisms by which chloride measurement of the CS is unreliable, because the CS is the
transport can take place in concrete. These include diffusion chloride content at the exposed surface of concrete. The
under the influence of a concentration gradient, absorption approach to obtain the CS is to fit a curve to the chloride
due to a capillary action, migration in an electrical field, a profile with the CS and the D as independent variables.
pressure-induced flow and wick action when water absorp- The difficulty in determining the CS usually arises from
tion and water vapor diffusion are combined [15]. the reduction of chloride content at the concrete surface
Diffusion is the process by which ions or molecules move of cover concrete as shown in Fig. 1. The chloride content
from an area of higher concentrations to lower concentra- within millimeters inside the outer surface of cover concrete
tions. Diffusion is the primary mechanism of chloride is decreased, because the concrete skin has a different com-
transport in concrete where there is no applied electric field position, compared to the internal concrete, due to phe-
and the moisture condition of the concrete pore structure is nomena such as a contact with the moulds, segregation
stable. For one dimensional into a semi-infinite medium, an of aggregates or dielectric reaction between the concrete
error function solution to the Fick’s second law [16] is surface and chloride environment [18]. Moreover, chlorides
usually used as follows: at the surface of cover concrete can often be washed out in
  preparation of sampling and thus lowers the CS.
x
Cðx; tÞ ¼ C S 1  erf pffiffiffiffiffi ; ð1Þ
2 Dt
5
where D = diffusion coefficient (m2/s); t = time of exposure
Surface chloride fitted for model
(s); C(x, t) = chloride concentration at depth x after time t
(%/m3); Cs = surface chloride concentration (%/m3); and 4 Duration: 400 days
Chloride content (%, cement) 1

erf is the error function. A true D can be achieved using Concentration of Cl-: 4M NaCl
Eq. (1), if concrete is homogeneous as being an ideal media,
D and CS are constant with time, and the chlorides either 3

do not react with cement matrix or if the ratio of chloride

binding is constant. This model is commonly used in deter-
mining chloride transport in concrete [17]. 2
Surface chloride measured

2.2. Measurement
1

Measurements of a chloride profile to calculate D and

CS usually follow the following procedures. The chloride 0
profile is obtained from samples collected by grinding the 0 3 6 9 12 15
concrete in a certain increment of depth from concrete Depth (mm)

surface to the steel. The chloride content of each sample Fig. 1. An example of the chloride profile with fitted curves for surface
can be usually determined in total chloride content for chloride content.
 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121 115

3. Reported values of D and CS published data does not provide clear relationship with
the time of exposure, due to other various factors to chlo-
A summary of D and CS reported from published data is ride transport including binder type, free water/binder
given in Table 1, which includes information on the type of ratio (W/B), curing condition, location of sampling in
sample and its exposure condition from which the data was structure and weathering.
obtained. The information includes reinforced structures,
laboratory concrete specimens exposed to an outdoor envi- 4. Influencing factors to chloride transport
ronment and concrete specimens exposed in a controlled
laboratory environment. In this study, total chloride For chloride transport, aggregates usually have very low
content is expressed as a percentage by weight of cement/ permeability and therefore have little impact on the perme-
binder. For the cases that the chloride content is only ation of chloride. Although the interface between coarse
reported as a percentage by weight of concrete specimen/ aggregate and cement paste often provides an easier path
structure but its mix design is not reported, a cement con- for chloride ions to move through concrete so that the size
tent of 350 kg/m3 and a concrete density of 2300 kg/m3 have of aggregate affects D [27], the rate of transport in a con-
been assumed. If D and CS were not provided, D in all the crete is largely dependant on the characteristics of cement
works have been estimated using the Fick’s second law and paste. In this study, concrete mix proportion, curing condi-
CS was obtained by fitting the curve to the chloride profile. tion, exposure to seawater and other physical properties of
The degree of contact with chlorides in an environment concrete are mainly taken as influencing factors to chloride
is a key factor for each different level of the surface chloride transport.
content. The CS for a concrete exposed to seawater may
not be changed with time due to the chemical equilibrium, 4.1. Concrete mix
but at the tidal zone the chloride content on concrete sur-
face can be increased by a number of wet and dry cycles. It is well known that binder type and W/B significantly
Also it is seen that degree of contact to seawater much affects chloride transport because formation of pore struc-
influences CS. For example, CS obtained from submerged ture is different, depending on concrete mix proportion.
condition in seawater was higher than that for tidal/splash, For example, pulverized fuel ash (PFA) or ground granu-
and specimens/structures exposed to airborne chlorides lated blast-furnace slag (GGBS) in concrete produces a
produces the lowest level of CS. However, D or CS from dense cement matrix with fewer capillary pores due to the

Table 1
Reported values of the diffusion coefficients D and surface chloride contents CS of concrete structures and specimens exposed to marine environments
CS (%, cem.) D (·1012 m2/s) Time (years) Exposure Concrete mix References
Binder W/B
4.08 Unknown 2 Laboratory (submerged) OPC 0.4 Gjørv and Vennesland [19]
2.55 0.5
1.33 0.6
3.74–5.54 3.52–4.66 16 Structure (tidal/splash) OPC Unknown Funahashi [10]
0.43–2.22 Unknown 23–58 Structure (aerated) OPC Unknown Uji et al. [20]
0.83–5.23 Structure (splash)
2.39–6.41 Structure (tidal)
1.50–3.10 2.13–3.39 24 Structure (splash/tidal) OPC 0.5 Liam et al. [21]
1.97 3.50 20 Structure (splash/tidal) OPC Unknown Kudoh et al. [22]
0.20–2.43 Unknown 30 Structure (aerated) OPC Unknown Morinaga [23]
0.016–0.143 4.41–4.91 1 Specimen (aerated) OPC 0.5 Mustafa and Yusof [24]
2.93, 4.72 12, 20 1, 4 Specimen (submerged) OPC 0.4 Bentz et al. [14]
2.10, 4.80 1.3, 5.0 30% PFA
0.48, 3.05 6.53, 7.85 3, 8 Specimen (tidal) OPC 0.66 Bamforth and Price [13,31]
0.63, 4.54 0.89, 0.78 30% PFA 0.54
0.64, 3.25 0.76, 0.56 70% GGBS 0.48
0.69, 3.66 3.98, 3.25 8% SF 0.72
3.53 2.142 15 Specimen (submerged) OPC 0.45 Mohammed et al. [25]
4.12 0.486 GGBS
5.80 0.552 PFA
10.51 0.042 33 Structure (splash/tidal) OPC 0.44–0.6 Troconis de Rincón et al. [26]
10.51 0.648 38
9.26 0.027 60
11.96 0.136 64
116 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121

pozzolanic reaction [28]. Moreover, the particle size distri- in D for PFA, GGBS or SF concretes occurs due to mainly
bution in the finest range, which accounts for 100–1000 Å, a refinement of pore structure [29], and the increased bind-
along with filler effect is more likely to reduce a chloride ing capacity of GGBS [31,32].
transport [29]. The comparison of the results by Bentz et al. [14] and
This study collected published data on D and CS of con- Bamforth [33] shows that the D for the concrete with
crete containing OPC, PFA, GGBS and silica fume (SF) W/B = 0.4 was about one order greater than the D with
exposed to a marine environment. Fig. 2 shows the rela- W/B = 0.66, even with similar exposure duration and
tionship between D and W/B, and CS and W/B. A curve external environment, which may be explained by the effect
was fitted to the values of D for OPC concretes, in the form of binder content. Buenfeld and Okundi [34] also showed,
of D = aeb(W/B), indicating an increase in D, by an increase from chloride solution-exposure testing, that the higher
in W/B by neglecting exposure durations. In contrast, the binder content, at a given W/B, could increase chloride
curves by recent data from JSCE [30] were fitted in the ion ingress in concrete.
form of D = a(W/B)b to the values for OPC and SF con- A reduction of CS with higher W/B was reported as
cretes. It is seen that an increase in W/B results in an shown in Fig. 2b. In Fig. 2b, a single study with same con-
increase in D for both OPC and SF concretes, and D for ditions from Gjørv and Vennesland [19] showed that the CS
SF concrete was lower than that for OPC concrete. PFA, at a given cement content was reduced by an increase in
GGBS or SF in concrete was, as expected, effective in W/B for a two-year submerge in chloride solution. This
enhancing the resistance to chloride transport. Decrease can be explained by the skin effect that reduces CS by a for-
mation of cement paste layer on the outer surface of cover
concrete of the specimen. Higher W/B is more likely to
a 100
form a thicker cement paste layer on the concrete surface
Liam et al 1992: 24 years [21] Bentz et al 1996: 4 years [14]
(i.e. concrete skin), so that the layer would enhance the
Bamforth 1999: 8 years [13, 31] Mohammed et al 2002: 15 years [25]
JSCE 1998 (OPC): 7-10 years [37] JSCE 1998 (SF): 7-10 years [37]
dielectric activity between hydroxyl ions in the concrete
skin and chloride ions in solution, and thus the increased
10
JSCE fit to OPC repulsive force removes chloride ions from the surface of
2.0586
D = 4.5342(W/B)
2
concrete. The CS was also much influenced by binder type
R = 0.83
m /s)

30% PFA as shown in Fig. 2b. Irrespective of exposure condition and

2

duration, CS for blended cement concretes was greater than

-12

1 Data fit to the other OPC

that for OPC concrete, presumably due to the higher level
D (×10

-2.4679
PFA D = 0.6091(W/B)
2
GGBS R = 0.32 of chloride binding and sorptivity. With increased chloride
JSCE fit to SF
D = 63.675(W/B)
3.9118 binding capacity, total chloride contents increase nearer the
0.1
2
R = 0.98 8% SF surface of the concrete, but decrease deeper in the concrete
[35]. The increase in the CS can be induced by the binding
30% PFA effect which allows the progressive build-up of higher total
70% GGBS chloride content on the surface of cover concrete.
0.01
0.25 0.35 0.45 0.55 0.65 0.75
4.2. Curing
W/B

b 8 Curing methods and duration can control the concrete

Gjorv and Vennesland 1979: 2 years [19] quality including mainly strength and durability of con-
Mohammed et al. 2002: 15 years [25] crete. Bamforth [33], and Bamforth and Price [13]
Liam et al. 1992: 24 years [21]
Bamforth 1999: 8 years [13]
addressed the influence of curing methods on chloride
6
Bentz et al. 1996: 4 years [14] transport reporting that D for OPC, 30% PFA, 70% GGBS
PFA
and 8% SF concretes was not significantly affected by the
Cs (%, cement)

30% PFA
GGBS curing method, while the CS was varied with curing.
8% SF
4 70% GGBS Fig. 3 shows the various levels of CS for different curing
methods (water, membrane and aeration), when the con-
Cs = -4.5279Ln(W/B) - 0.4598 crete specimens were submerged in seawater for 3 and 8
2
R = 0.33
years, respectively. It is seen again that the CS was
2
increased with time, ranging from 0.44% to 0.69% by
weight of cement for the 3 year exposure, while the CS
30% PFA increased up to 1.66–6.45% after the 8 year exposure.
0 The CS shows a marginal change for the first 3 years with
0.35 0.45 0.55 0.65 0.75 curing methods, whereas the CS for the 8 years exposure
W/B was varied much with each curing method. For OPC and
Fig. 2. Apparent diffusion coefficients (a) and surface chloride content (b) 8% SF concretes, aerated curing is beneficial in reducing
of concrete structures and specimens exposed to a marine environment. CS, compared to water curing. For 30% PFA concretes,
 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121 117

Fig. 3. Time and curing-dependence of surface chloride from [13,32].

CS is much less with either water curing or membrane, 2.0

compared with aerated curing. Hence, curing method Liam et al. 1992 [21]
should be carefully chosen, depending on binder type, in TWRL 1980 [36]

order to minimise CS and thus the chloride transport in 1.0

Height from sea level (m)

concrete.
Sea level
4.3. Exposure condition 0.0

The value of D and CS may vary to a large degree by the

location of the structure in a marine environment, i.e., the -1.0
degree of exposure to chloride environment and weathering
condition with regard to prevailing winds and rainfall. For
marine concrete structures, the vicinity to seawater could -2.0
mainly govern the chloride transport behaviour. However, 0.0 1.0 2.0 3.0 4.0 5.0
Surface chloride (%, cement)
there has been little study concerning the effect of location
from sea level, except several limited data [21,36], which are 2.0
shown in Fig. 4 for D and CS for jetty structures exposed to Liam et al. 1992 [21]
tidal/splash zones for 24 years. The height from sea level TWRL 1980 [36]
Height from sea level (m)

was limited within 2 m higher or about 1 m lower from 1.0

sea level. It seems that splash zone near sea level is the eas-
iest for chlorides to accumulate on the surface of concrete,
while a build-up of CS took place the least at the sea level. 0.0
Sea level

This may be due to the wet/dry cycles, resulting in a succes-

sive supply of chlorides by wetting with seawater, and
evaporation and salt crystallization by drying. Water -1.0
absorption may lead to very rapid penetration of chloride
ions dissolved in seawater. It is, for example, well known
that concrete in the tidal zone of an offshore structure or -2.0
0 2 4 6 8
bridge decks subjected to regular applications of deicing
D(× 10-12 m2/s)
salts, may suffer from rapid chloride ion ingress due to
the water absorption. However, Uji et al. [20] showed from Fig. 4. (a) Apparent diffusion coefficient and (b) surface chloride content
a long term monitoring from 23 to 58 years that the order of a jetty concrete structure exposed to splash/tidal conditions for the 24
years.
of CS with regard to the level from seawater is tidal >
splash > atmospheric zone. The difference of CS build-up
between these data is not clearly known, but presumably Fig. 5 shows two sets of the CS build-up measured
different weathering and concrete mix may have affected from specimens and structures exposed to a chloride
to the process of chloride transport. aerated environment (i.e. coastal area) for 1 and 30 years,
118 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121

3
a 1.E-09
Mustafa and Yusof 1994: 1 year [24]
Morinaga 1992: 30 years [23] OPC (rich) OPC (lean) 30% PFA 65% GGBS
JSCE recommended 1999 [37]
1.E-10

2
Cs (%, cement)

D (m /s)
OPC

2
1.E-11

30% PFA
JSCE Guideline
1
1.E-12

65% GGBS

1.E-13
0 0.5 1 1.5 2 2.5
0
0 200 400 600 800 1000 Air void content (%)
Distance from sea (m)
6
Fig. 5. Surface chloride content in aerated conditions with the distances
b OPC (rich)
OPC (lean)
from sea for 1 and 30 years exposures. 30% PFA
65% GGB S
5

Cs (%, cement)
respectively [23,24], along with a Japanese Standard Spec-
ification for airborne chloride build-up [37]. The CS from
Mustafa and Yusof [24] was marginally changed with the
4
distance from sea, because of a relatively short-term expo-
sure to the chloride atmospheric condition. Moreover, the
values of CS were too small to set a high level of chloride
concentration gradient in concrete for corroding the 3
0 0.5 1 1.5 2 2.5
embedded steel. For a 30-year exposure period, it is clearly Air void content (%)
seen that the distance from the sea strongly influences a
build-up of CS, ranging from 0.20% to 2.43% by weight Fig. 6. Apparent diffusion coefficient (a) and surface chloride content (b)
of concrete specimens exposed to 4 M NaCl solution [43].
of cement, for the distance 50–1000 m away from sea. It
is notable that the values of CS specified by the Japanese
Standard [37] nearly equate to the CS after a 30 years expo- slightly increased as the air voids increased, except for
sure to airborne conditions. It suggests that aerated chlo- the 30% PFA concrete where D was relatively constant.
rides also may impose a risk of corrosion of steel, when It seems likely that larger air void content, which may be
structures are located in the coast close to sea and the expo- induced by various sources (e.g., poor compaction in the
sure duration is long enough for CS to accumulate. How- casting of the concrete) results in the formation of pore
ever, structures about 1 km away from coast may be free structures that favor the transport of chloride ions. It is,
from a chloride-induced corrosion deterioration, because however, notable that an increase in the air void content
it would take extremely long for chloride ions, with a lower resulted in a decrease in CS, irrespective of binder as shown
level of CS, to reach the chloride threshold level [39] for in Fig. 6b.
steel corrosion at the depth of the steel. Miera et al. [44] Fig. 7 shows an influence of climate on chloride trans-
provide additional information on chloride accumulation port in terms of CS. Three different series of CS were mea-
in concrete due to natural exposure to marine aerosols. sured in the UK, Japan and Venezuela, where the latitudes
of sampling places are different. The exposed duration was
4.4. Other factors 8, 7–58 and 7–64 years from the UK, Japan and Venezuela,
respectively. The specimens from the UK were only sub-
The importance of air void content in concrete has been merged in seawater and the others were exposed in the
emphasized due to its influence on concrete strength as well tidal/splash zones. The CS from the UK (i.e. the highest
as corrosion initiation [38]. However, to date, the influence in the latitude) were the lowest, probably because of the
of air void on chloride transport has rarely been addressed. short duration of exposure as well as the submerged condi-
Ann and Song [39] indicated an increase in D and a tion, compared to the CS from Japan and Venezuela, which
decrease in CS, resulting from an increase in air void con- were measured after the even longer durations of exposure.
tent. Fig. 6 shows D and CS with the air void content for However, the CS from Venezuela was much higher than
OPC, 30% PFA and 65% GGBS concretes. As expected, that from Japan, despite a similar exposure condition and
65% GGBS and 30% PFA concrete indicated lower levels duration. This difference may be due to different climates,
of D, compared to D for OPC concrete. For each mix, D since the tropical climate fosters chloride ion movement
 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121 119

15 implies a loss of paths for chloride ions. Thus, D can be

Submerged Tidal/splash Tidal/splash
60-64 years decreased by time (i.e. degree of hydration) and CS can
12 be increased by the increase in chloride binding and the
build-up of chlorides.
Cs (%, cement)

7 years
9 The so-called time-dependent model of D in the most
7 years 55-58 years cases was empirically obtained as follows:
8 years t m
6 0
DðtÞ ¼ D0 ; ð2Þ
t
3 where D(t) = diffusion coefficient at time t (m2/s); D0 = dif-
fusion coefficient at t0 (m2/s); t0 = standard time (1 year or
0 28 days) (s); t = time (s); and m = constant value. For CS, a
UK (latitude 50-55˚) Japan (latitude 35-40˚) Vemezuela (latitude 5-12˚)
Location of sampling
refined model was previously suggested [17,40] and is being
used commercially (eg. LIFE 365). For a linear build-up of
Fig. 7. Surface chloride contents of concrete structures measured in the CS over time, the chloride ingress with regard to the con-
UK, Japan and Venezuela [20,26,32].
crete cover depth and time can be obtained from Eq. (3),
into concrete due to the high levels of relative humidity, and for a square root build-up of CS over time, Eq. (4)
temperature and chloride concentration. High temperature can be used to determine the chloride ingress,
and high humidity will help chloride ions to transport in C S ðtÞ ¼ k 1 t; ð3Þ
concrete via absorption altogether with diffusion. More- pffi
C S ðtÞ ¼ k 2 t; ð4Þ
over, higher concentration of chloride ions in seawater will
increase the chloride ingress, which is responsible for the where CS(t) = surface chloride concentration at time
high CS. t (%/m3); and k1, k2 are constant values. However, these
models do not consider an initial build-up of chlorides on
5. Refinement of basic model the surface of cover concrete, at an early age concrete ex-
posed to a salt environment. For example, the CS would
The fixed value of D and CS have been extensively used be negligibly small for a 1-year exposure, accounting for
in predicting the chloride ingress using the Fick’s second 0.10% and 0.77% by weight of cement from Eqs. (3) and
law, when structures are exposed to a marine environment (4), respectively, whereas a measured CS accounts for
[17]. However, the D decreases with time, because develop- 2.1–2.5% for only a 28-day exposure. For 60 years of expo-
ment of cement hydration reduces the pore distribution in sure, Eqs. (3) and (4) overestimate CS build-up as 7.6–8.2%,
the fine size, and hydration products react with chloride compared to 6% obtained from exposure test data [14].
ions to form Friedel’s salt and/or to adsorb the chlorides Moreover, these models for CS do not consider the chloride
physically, which delays chloride transport in concrete. build-up for the different conditions of tidal/splash, sub-
Also, the CS has been intuitively hypothesized to be inde- merged zone and aerated zone. Hence, the models may
pendent of time, when concrete structures are exposed to not apply for predicting chloride transport of concrete
a tidal/splash environment, whereas in salt aerated condi- structures exposed to a marine environment both for differ-
tion, chlorides have been considered to be accumulated ent very short-term and long-term exposure and chloride
on the surface of concrete with time [40,44]. A literature content at an early stage of exposure in a chloride environ-
review showed that both D and CS are time-dependent ment is nullified.
and moreover the values are influenced by binder type, Hence, it is required to refine a model for the time-
W/B, exposure condition and even weathering. To predict dependent CS, considering the initial build-up of chlorides
the service life of concrete structures subjected to chloride- on the surface of concrete. The following model empirically
induced corrosion, the complexity of chemical, physical derived from published data on CS is proposed to consider
reactions between the cement hydration species and the initial build-up of chlorides:
chloride ions must be taken into account for the analysis
C S ðtÞ ¼ C 0 þ a lnðtÞ; ð5Þ
of chloride transport. However, it would be very difficult
to consider all these factors in the analysis of chloride where C0 = surface chloride at the standard time (1 year or
transport. Thus, the present study suggests a refined model, 28 days) (%/m3) and a = a constant value.
considering the time-dependent characteristics of D and Fig. 8 shows an increase of CS with time, based on Eqs.
CS. A number of works used for the model for time-depen- (3)–(5). For the best fit, k1, k2, and a were determined as
dent D in predicting the chloride profile [30,36]. Time- 0.137%/yr, 1.001%/yr0.5 and 0.977%/exp(yr), respectively.
dependency of D and CS may be mainly associated with For Eq. (5), C0 was taken as the CS for a 28 days exposure
hydration of the cement matrix. Capillary pore volume from Bentz et al. [14]. It is seen that the CS values from
and its connectivity are important in assessing chloride models of linear and square root build-up (Eqs. (3) and
transport, and the capillary pores reduce and are partially (4), respectively) are unrealistically low at an early age,
blocked as the cement matrix hydrates [41,42], which while the refined model in this study indicated the CS above
120 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121

9 (3) For the exposure to tidal and splash zones, D and CS

CS=k1t [17, 40]
8
were affected by the height from sea level (i.e. degree
CS=K2 t [17, 40]
of contact to seawater). As the height increases from
CS=Co + α ln(t) [this study]
7 sea level, both D and CS increase. For airborne chlo-
CS = t
6 ride environments, the exposed time and the distance
Cs (%, cement)

from sea are key factors to CS, but the level of CS at a

5
CS=3.0431+0.6856 ln(t)
given D was negligibly low to impose a corrosion risk,
4 compared to tidal/splash or submerged region.
CS=0.137t
(4) Environmental conditions like climate significantly
3
affect chloride transport. Concrete structures exposed
2 to a marine environment near the Equator produced
1 a remarkably high CS, compared to the CS from the
high latitudes. The higher CS in the tropical area is
0
0 20 40 60 80 due to an enhanced ionic transport and high concen-
Time (years) tration of chlorides.
Fig. 8. Surface chloride content with time from a linear and square root
(5) Air void content in concretes influences D and CS. An
build-up models, and the proposed model in this study. increase in air voids resulted in increased D and
decreased CS. A large air void content (due to poor
2.0%. After 60 years, the rate of CS increase for the pro- compaction, for example) increases pore structure
posed model is the lowest and reaching a plateau, whereas formation, which provides more open pathways for
the CS for linear and square root build-up models are chloride transport.
increasing more rapidly. (6) A refined model for CS which considers time-depen-
dent characteristics of chloride build-up and binding
6. Conclusion capacity is proposed in this study. This model pro-
vides a more realistic build-up of CS, compared to a
This paper studied the change of the apparent diffusion few basic models for CS accumulation in an aerated
coefficient D and surface chloride content CS with time condition. However, further refinement of the model
from structures/specimens exposed to marine environ- and verification with long-term exposure data is nec-
ments, as reported from published data. It was found that essary for practical prediction of chloride profiles in
D for a concrete structure decreases with time due to a concrete structures exposed to marine environments.
subsequent hydration of cement matrix, which makes the
concrete pore structure denser, or removes chloride ions
by binding. CS is also time-dependent, when exposed to Acknowledgements
tidal and splash zones or even submerged in seawater.
Altogether with the time-dependent characteristics, there The authors would like to acknowledge financial sup-
are a number of influencing factors in chloride transport ports from a project on Standardization of Performance
in terms of D and CS: concrete mix proportion (i.e. free based Construction Specifications and Design and a Centre
water/binder ratio (W/B) and binder type), curing for Concrete Corea, Korea.
methods, the proximity to seawater, air void content in
concrete and external environments. This study provides References
the following conclusions.
[1] Federal Highway Administration. Building more durable bridges.
(1) It was found that, from a single series of data, an Washington DC: FOCUS Federal Highway Administration; Sep-
tember, 2001. p. 6.
increase in W/B resulted in an increase in D, due to
[2] BRE Centre for Concrete Construction. Guide to the maintenance,
formation of increased pore structure, and CS was repair and monitoring of reinforced concrete structures. DME report
decreased by an increase in W/B, which seems due No. 4. Watford, UK; 2001.
to the concrete skin effect. Blended cement concrete [3] Page CL. Mechanism of corrosion protection in reinforced concrete
containing PFA, GGBS or SF decreases D and marine structure. Nature 1975;256:514–5.
increases CS, due to a refinement of the concrete pore [4] Page CL, Treadaway KWJ. Aspects of the electrochemistry of steel in
concrete. Nature 1982;297:109–15.
structure and an increased chloride binding capacity, [5] Gjørv OE, Vennesland Ø. Evaluation and control of steel corrosion
respectively. in offshore structures. In: Scanlon JM, editor. Concrete durability.
(2) It was seen that chloride transport depends on curing ACI SP 100; 1987. p. 1575–1602.
methods: water, membrane and aeration. For OPC [6] Berke NS, Sundberg KM. The effect of admixtures and concrete mix
designs on long-term concrete durability in chloride environments. In:
and SF concretes, aerated curing is beneficial in
Corrosion-89, NACE conference. Texas, US; 1989. p. 386.
reducing the CS, whereas water curing is beneficial [7] Kropp J. Chlorides in concrete. In: Kropp J, Hilsdorf HK, editors.
for PFA and GGBS concretes. However, D does Performance criteria for concrete durability, E&FN SPON; 1995. p.
not seem to be much affected by curing duration. 138–64.
 H.-W. Song et al. / Cement & Concrete Composites 30 (2008) 113–121 121

[8] Keer JG. Surface treatments. In: Mays G, editor. Durability of [26] Troconis de Rincón OTO, Castro P, Moreno EI, Torres-Acosta AA,
concrete structures, E&FN SPON; 1992. p. 146–65. Morón de Bravo O, Arrieta I, et al. Chloride profiles in two marine
[9] Hooton RD, Geiker MR, Bentz EC. Effects of curing on chloride structures – meaning and some predictions. Build Environ
ingress and implications on service life. ACI Mater J 2002;99:201–6. 2004;39:1065–70.
[10] Funahashi M. Predicting corrosion free service life of a concrete [27] Oh BH, Cha SW, Jang BS, Jang SY. Development of high-
structure in a chloride environment. ACI Mater J 1990;87:581–7. performance concrete having high resistance to chloride penetration.
[11] Clear KC. Time to corrosion of reinforcing steel in concrete slab, vol. Nucl Eng Des 2002;212:221–31.
3. Report No. FHWA-RD-76-70. Federal Highway Administration; [28] Schiessl P, Breit W. Local repair measures at concrete structures
1986. damaged by reinforcement corrosion. In: Page CL, Bamforth PB,
[12] West RE, Hime WG. Chloride profiles in salty concrete. Mater Figg JW, editors. Corrosion of reinforcement in concrete construc-
Performance 1985;24:29–36. tion. Cambridge, UK: The Royal Society of Chemistry Publication;
[13] Bamforth PB, Price WF. Factors influencing chloride ingress into 1996. p. 525–34.
marine structures. In: Dhir RK, Jones MR, editors. Concrete 2000; [29] Hussain SE, Rasheeduzafar S. Corrosion resistance performance of
1993. p. 1105–18. fly ash blended cement concrete. ACI Mater J 1994;91:264–73.
[14] Bentz EC, Evans CM, Thomas MDA. Chloride diffusion modelling [30] Maruya T, Tangtermsirikul S, Matsuoka Y. Modeling of chloride ion
for marine exposed concretes. In: Page CL, Bamforth PB, Figg JW, movement in the surface layer of hardened concrete. Concrete
editors. Corrosion of reinforcement in concrete construction. Cam- Library of JSCE, No. 32; 1998. p. 69–84.
bridge, UK: The Royal Society of Chemistry Publication; 1996. p. [31] Dhir RK, El-Mohr MAK, Dyer TD. Chloride binding in GGBS
136–45. concrete. Cement Concrete Res 1996;26:1767–73.
[15] Buenfeld NR, Shurafa-Daoudi MT, McLoughlin IM. Chloride [32] Luo R, Cai Y, Wang C, Huang X. Study of chloride binding and
transport due to wick action in concrete. In: Nilsson LO, Oliver JP, diffusion in GGBS concrete. Cement Concrete Res 2003;33:1–7.
editors. Chloride penetration into concrete. Paris, France: RILEM [33] Bamforth PB. The derivation of input data for modeling chloride
Publication; 1997. p. 302–24. ingress from eight-years UK coastal exposure trials. Mag Concrete
[16] Crank J. The mathematic of diffusion. 2nd ed. Oxford Science Res 1999;51:87–96.
Publication; 1983. [34] Buenfeld NR, Okundi E. Effect of cement content on transport in
[17] ACI Committee 365. Service-life prediction: state of the art report. concrete. Mag Concrete Res 1998;50:339–51.
Manual of concrete practice, Part 5. American Concrete Institute; [35] Glass GK, Buenfeld NR. The influence of chloride binding on the
2000. chloride induced corrosion risk in reinforced concrete. Corros Sci
[18] Andrade C, Diez JM, Alonso C. Mathematical modelling of a 2000;42:329–44.
concrete surface skin effect on diffusion in chloride contaminated [36] TWRL. Marine durability survey of the Tongue Sand Tower.
media. Adv Cement Based Mater 1997;6:36–44. Concrete in the oceans. Technical Report No. 5. Cement and
[19] Gjørv OE, Vennesland Ø. Diffusion of chloride ions from seawater Concrete Association; 1980. p. 84–92.
into concrete. Cement Concrete Res 1979;9:229–38. [37] Standard Specification for Durability of Concrete. Japan Society of
[20] Uji K, Matsuoka Y, Maruya T. Formation of an equation for surface Civil Engineers; 1999. p. 11–3.
chloride content of concrete due to permeation of chloride. In: Page [38] Glass GK, Yang R, Dickhaus T, Buenfeld NR. Backscattered
CL, Treadaway KWJ, Bamforth PB, editors. Corrosion of reinforce- electron imaging of the steel-concrete interface. Corros Sci
ment in concrete. London, UK: Elsevier Applied Science; 1990. p. 2001;43:605–10.
258–67. [39] Ann KY, Song HW. Chloride threshold level for corrosion of steel in
[21] Liam KC, Roy SK, Northwood DO. Chloride ingress measurements concrete. Corros Sci, in press.
and corrosion potential mapping study of a 24-year-old reinforced [40] Amey SL, Johnson DA, Miltenberger MA, Farzam H. Predicting the
concrete jetty structure in a tropical marine environment. Mag service life of concrete marine structures: an environmental method-
Concrete Res 1992;44:205–15. ology. ACI Struct J 1998;95:205–14.
[22] Kudoh K, Hirotani A, Moriwake A, Yasuda M. Study on durability [41] Garboczi EJ. Computational materials science of cement-based
of a repaired concrete structure in Tokyo bay. In: Malhotra VM, materials. Mater Struct 1993;26:191–5.
editor. Evaluation and rehabilitation of concrete structures and [42] Winslow DN, Cohen MD, Bentz DP, Snyder KA, Garboczi EJ.
innovations in design. ACI 128, Hong Kong; 1991. p. 1293–307. Percolation and pore structure in mortars and concrete. Cement
[23] Morinaga S. Life prediction of reinforced concrete structures in hot Concrete Res 1994;24:25–37.
and salt-laden environments. In: Walker MJ, editor. Concrete in hot [43] Ann KY. Enhancing the chloride threshold level for steel corrosion in
climates, E&FN SPON; 1992. p. 155–64. concrete. PhD thesis. Imperial College of Science Technology and
[24] Mustafa MA, Yusof KM. Atmospheric chloride penetration into Medicine, University of London; 2005.
concrete in semitropical marine environment. Cement Concrete Res [44] Meira GR, Andrade C, Padaratz IJ, Alonso C, Borba Jr JC. Chloride
1994;24:661–70. penetration into concrete structures in the marine atmospheric zone –
[25] Mohammed TU, Yamaji T, Hamada H. Chloride diffusion, micro- relationship between deposition of chlorides on the wet candle and
structure and mineralory of concrete after 15 years of exposure in chlorides accumulated into concrete. Cement Concrete Compos
tidal environment. ACI Mater 2002;99:256–63. 2007;29(9):667–76.