Documente Academic
Documente Profesional
Documente Cultură
STRUCTURES
Marine concrete structures
H. T. CAO and V. SIRIVIVATNANON
CSIRO Division of Building, Construction and Engineering, North Ryde, NSW,
Australia.
Abstract
This paper presents a simple approach to predicting service life of marine reinforced
concrete structures. This approach is based on the solution derived from Ficks 2nd
law of diffusion with modifications. The modifications required are the corrections
for chloride surface concentration CS and chloride diffusion coefficient D. Acceptable
steel corrosion rate is suggested for use in the prediction of service life.
Keywords: Service life, chloride penetration, corrosion
Introduction
Time from construction until when the chloride content at the reinforcement is
high enough to initiate steel corrosion (corrosion-free models);
Time from construction to time at which signs of distress are visible/observable
(acceptable corrosion models);
Time from construction to time at which the member/component/structure is
considered to be functionally unacceptable or unsafe (end-of-function models).
These items will be discussed in this paper. The objective of this paper is to
present the rationale behind the process of prediction of service life of marine RC
concrete structures. The key to predicting service life of marine RC structures is to
evaluate the life of each component using an appropriate model/process. For a given
structure, different models may be required for different components. The
components with the shortest predicted service life would dictate the overall service
life, especially when the failure of such a component has a serious structural/safety
consequence. The discussion of service life prediction in this paper is therefore
limited in scopes.
Propagation period
Reinforcement corrosion
3
Cracking
Time (y)
(1)
x
C ( x, t ) = Ci + (CS Ci )erfc
(2)
4tD
where C ( x, t ) is the chloride content at depth x at time t; Ci is the initial chloride
content of the concrete (usually negligible); CS is the chloride content at the exposed
surface; D is the chloride diffusion coefficient and erfc is the error function
complement.
Based on this equation, the concrete is characterised by CS and D ( Ci ~0).
Curve fitting programs are widely used to obtain values of D and CS from an
experimentally determined chloride profile at a known time. The C(x,t) is predicted
with relevant figures of concrete cover depth and exposure time. This procedure has
been adopted in many models for predicting chloride penetration into concrete
Recently, it has been pointed out that theoretically the use of Fick law in the
prediction of chloride penetration is not appropriate [Chatterji, 1994 a,b; 1995]. The
main reason for this inapplicability is that Ficks law is valid for non-ionic
diffusants through inert homogeneous medium. Chloride penetration into concrete is
clearly an ionic transportation process involving other ions in the pore solution
[Chatterji 1994c; Sergi 1992]. Concrete is an inhomogeneous and dynamic medium
chemically and physically. Practically, analyses of concrete subjected to real marine
environments indicate that CS and D are not constants with exposure time [Johansen
et. al 1995; Maage et. al 1996; Weyers et. al 1998]. From both theoretical and
practical viewpoints, one must concede that the application of the solution derived
from Ficks second law in the prediction of chloride penetration into concrete is not
appropriate. The Nernst/Nernst-Planck equations are more appropriate for movement
of ions [Chatterji 1994 a]. Even with these equations, catering specially for ionic
transportation, the authors are not convinced that they can cope adequately with the
chemical and physical variations of concrete with time. At this stage, it appears that
the solution to Ficks 2nd law is still the most convenient and practical means of
predicting chloride penetration into concrete, provided that adequate corrections are
made.
With considerations to concrete behaviour and the nature of ionic movement,
the parameters defining concrete in equation (2), CS and D, can not be constants in the
process of prediction of chloride penetration. CS should be a time-dependent
parameter and D should be dependent on time and depth of concrete. The time
dependent corrections for CS and D are necessary for the variations in pore structure
and hydration/reaction between concrete and the environment, particularly in the
presence of chloride ions. The depth dependence of D is required for the
concentration effect on ionic movement, which is related to both time and depth.
x
x
x2
C ( x, t ) = kt (1 +
)erfc(
)(
)e x / 4 Dt
2 Dt
2 Dt
Dt
(3)
2 Dt
(4)
Assume that concrete porosity at the surface is saturated with chloride ion (~35%
NaCl equivalent). This is about 190 kg of Cl per m3 of saturated salt solution.
Some chloride ions are bound in hardened cement pastes, which should be
related to the binder content of the concrete (B). A binding capacity of 2.5% by
weight of binder is proposed [Nagataki et. al 1993]. The bound chloride is
estimated as 2.5 x B/100. For a concrete containing 400 kg of binder, the amount
of bound chloride is about 10 kg.
For concrete with surface porosity P (eg. 10% or 0.1) and density (eg. 2400
kg/m3), the surface chloride concentration CS (total Cl concentration by weight of
concrete) can be estimated as (bound chloride + 190 x P) /. With the values
given as example; CS is estimated as (10 + 19)/2400 = 1.2% by weight of
concrete. (The connection between CS to porosity can be used to cope with
cracked concrete).
Rough estimates of limiting CS values (by weight of concrete) are about 1.5%
for 32 MPa; 1% for 50 MPa and about 0.8% for 60 MPa (uncracked) concretes.
The above estimated values of CS can be used in Equation (2) for long term
prediction of chloride penetration and in Equations 3 and 4 as limiting values for kt or
k t . It is expected that in practice, due to salt recrystallisation, much higher value of
CS may be observed. However, salt crystals eg. NaCl do not participate in the inward
movement of saturated Cl solution. Hence the use of these apparently high CS values
in the prediction of chloride penetration is not justified.
D is assumed to be constant in the above discussion. As noted previously, D is a
time and depth dependent parameter. Provided that other physical and chemical
processes do not cause damages to concrete, the measured chloride diffusion
coefficient D should decrease with exposure time. This is partly due to the
modification of the pore structure and partly due to the reactions between cement
hydrates and advancing chloride ions. Chloride binding capacities of cementitious
materials and pore refining/degree of hydration of the hardened cement pastes are
time dependent. In practice, the trend of reduction of D with exposure time is well
published [Maage et. al 1996; Mangat and Molley 1995; Tumidajski et. al 1995].
An empirical form of correction for the time dependence of D is Data = K
[Maage et. al 1996], where Da is the chloride diffusion coefficient at time ta and K is a
constant. Chloride diffusion coefficient Db at time tb > ta can be estimated from Da,
and Db is smaller than Da. is a parameter depending on the material, mix design and
environment. In most if not all applications, is not known. The reduction of D with
time, therefore should be determined by obtaining two pairs of D and t and by plotting
D vs t on a double log system. A straight line is expected and can be used in the
extrapolation of D to the required exposure time as shown in Figure 2.
high w/b
Different binders
low w/b
Different binders
In the previous discussion, the diffusion coefficient D is that obtained from the
chloride concentration profile. When D is calculated on each measured point, it has
been demonstrated that it is depth dependent and increases with depth. Theoretically,
this can be partly explained by the increase in diffusivity with decreased concentration
of diffusant solution (Cl-) at higher depth. The chloride binding capacity has also
been observed to decrease with depth [Nagataski et. al 1993; Mangat and Molley
1995]. Boltzmann-Mantano analysis has been used to treat both time and depth
dependent characteristics of D [Shewmon 1989; Tumidajski et. al 1995]. The issue of
depth dependence of D is still needed to be sorted out in a practical manner. At the
moment, it appears that there is no practical need for a correction with a concrete
cover of 50 mm or less. For a concrete cover depth greater than 50 mm, a factor of
(cover depth/50) should be applied to D for long term prediction of chloride
penetration using solutions derived from Fick 2nd law.
It can be summarised that long term prediction of chloride penetration into
concrete, which is the most critical module in any service life model, is not a precise
process. It is performed with the use of empirical relationships (some with limited
application). For practical purposes, it appears that the Ficks 2nd law solutions with
suitable corrections can be used to provide a reasonable estimation of chloride
penetration.
Chloride threshold level for steel corrosion and acceptable corrosion rate
Molley 1995; Hussain et. al 1996]. Furthermore the pore solution chemistry is
expected to change with exposure time due to OH- movement and the effect of SO42and carbonation. In the laboratory, the authors observed that the chloride threshold
level varies significantly with binder type, concrete mix design, initial curing and
concrete cover. It is also depend on the measurement techniques (eg. halfcell
potential, anodic polarization, linear polarization, steel loss, etc.) and adopted
definitions.
The sometimes used figures of chloride threshold level of less than about 0.4%
by mass of binder or about 0.1% by mass of concrete (total chloride) can be either
over optimistic or pessimistic for applications in service life prediction. Due to this
additional uncertainty/arbitrary, the authors opted to abandon the concept of chloride
threshold level in the prediction of service life. Using laboratory-generated results,
the authors observed that when steel corrosion rate is at a reasonable level, the
variation of chloride level with respect to binder types, concrete mix design and
concrete cover become much less than that observed at low corrosion rate (nearer to
chloride threshold level).
Similar observation has been given in the literature
[Mangat and Molley 1995]. For this reason, it is recommended that acceptable
corrosion rate be used in the service life prediction.
It has been suggested that cracking of concrete cover depends on bar size,
corrosion rate and cover depth [Weyers 1998; Lopez et. al 1993; Andrade et. al 1993;
Schiel and Raupach 1997]. A corrosion rate of about 2A.cm-2 appears to be a
reasonable value for acceptable corrosion rate. Available data [Weyers 1998; Lopez
et. al 1993; Andrade et. al 1993] suggest that at a corrosion rate of 2A.cm-2, the
expected time require to crack a concrete cover greater than 25 mm is greater than 2
years. It would take about 10 years for the crack to reach 0.3mm in width or a 5%
section reduction of the bar. Structural performance of the reinforced concrete is not
expected to be seriously damaged [Cabrera 1996].
The issue is to determine the chloride concentration at which the steel corrosion
rate is 2A.cm-2. Based on electrochemical experimental data generated using a wide
range of binders including portland cements and blended cements containing different
fly ashes, slag and silica fume at different percentages and water-to-binder ratio, the
authors suggest an interim value of 0.2% Cl (total) by weight of concrete. This value
appears to be in reasonable agreement with actual site data [Liam et. al 1992]. Work
is still in progress to validate this figure and to establish a list of values for different
binders and W/B. This chloride concentration is then used in conjunction with the
predicted chloride penetration for service life prediction.
The suggested procedure for prediction of service life is that based on Ficks 2nd
law with modifications as discussed in the previous sections. For example, if a
service life of 50 years is required for a marine RC component with a 50 mm cover,
this means that the expected chloride concentration at 50 mm depth after 50 years
should be about 0.2% by mass of concrete. Using equation (1) with suggested CS of
1%, the required chloride diffusion coefficient is about 0.5 x 10-12 m2.sec-1. This is
the long term chloride diffusion coefficient. Data are then required to translate this
requirement back to short term chloride diffusion coefficient needed for specification
purpose.
An example of the required data is shown in Figure 3. This figure illustrates the
diffusion coefficient data for a 25% fly ash blended cement 50 MPa concrete cured
for 7 days. Based on this figure, it is likely that this fly ash concrete will achieve the
service life of 50 years. It can be seen from this figure that the measured chloride
diffusion for this concrete after 1 and 6 month of immersion in ocean water are about
1.0 x 10-10 m2.sec-1 and 1.9 x 10-12 m2.sec-1respectively. The D value of 1.0 x 10-11
m2.sec-1 can be used in the specification for fly ash concrete of similar replacement
percentage if the test method is based on 28-day immersion. However, this value
should not be used when concretes are made with other binder types. This is due to
the different response of D vs t.
From this example, the case-by-case feature of service life prediction is
illustrated. A 50 MPa/25% fly ash concrete cured for 7 days may satisfy the
Australian Standard AS 3600 expectation of 40-60 years service life in marine
exposure. Other 50 MPa concrete made with different binders and cured for 7 days
may not.
Chloride diffusion coefficient, m 2 sec -1
2E-11
1E-11
5E-12
2E-12
1E-12
"Limiting value" ?
5E-13
0.03
0.1
0.3
10
30
100
Fig. 3: D vs t for a 25% fly ash blended cement concrete, 50 MPa, 7 days cure
Concluding remarks
References
Liam, K.C., Roy, S.K. and Northwood, D.O. (1992), Chloride ingress measurements
and corrosion potential mapping study of a 24 year old reinforced concrete
jetty structure in a tropical marine environment, Mag. of Concr. Res., V. 44,
No.160, 205-215.
Lopez, W., Gonzalez, J., Andrade, C. (1993), Influence of temperature on the service
life of rebars, Cem. and Concr. Res., V.23, 1130-1140.
Maage, M., Helland, S., Poulsen, E., Vennesland, O and Carlsen, J.E. (1996),
Service life prediction of existing concrete structures exposed to marine
environment, ACI Mat. J., V. 93, No.6, 602-608.
Mangat P.S. and Molloy, B.T. (1995), Chloride binding in concrete containing PFA,
GBS or silica fume under sea water exposure, Mag. of Concrete Res., 47,
No.171, 129-141.
Nagataki, S., Otsuki, N., Wee, T-H and Nakashita, K. (1993), Condensation of
chloride ion in hardened cement matrix materials and on embedded steel
bars, ACI Mat. J., V.90, No.4, 323-332.
Neville, A. (1995), Chloride attack of reinforced concrete: an overview, Materials
and Structures, 28, 63-70.
Pommersheim, J. and Clifton, J. (1985), Prediction of concrete service-life,
Materiaux et Constructions, Vol. 18, No. 103, 21-30.
Schiel, P. and Raupach, M. (1997), Laboratory studies and calculations on the
influence of crack width on chloride-induced corrosion of steel in concrete,
ACI Mat. J., V.94, No.1, 56-62.
Sergi, G., Yu, S.W. and Page, C.L. (1992), Diffusion of chloride and hydroxyl ions
in cementitious materials exposed to a saline environment, Mag. of Concr.
Res., 44, No. 158, 63-69.
Shewmon, P. (1989), Diffusion in solids, 2nd Ed., the Minerals, Metals and Materials
Society, Warrendale, PA.
Tumidajski, P.J., Chan, G.W., Feldman, R.F. and Strathdee, G. (1995), A
Boltzmann-Matano analysis of chloride diffusion, Cem. and Concr. Res.
V.25, No.7, 1995, 1556-1566.
Uli, K., Matsuoka, Y and Maruyat (1990), Formulation of an equation for surface
chloride content of concrete due to permeation of chloride, Proc. 3rd Int.
Symp. Corrosion of Reinforcement in Concrete, Wishaw, 1990, Soc. Chem
Indus., Elsevier Applied Science.
Weyers, R. et al, Concrete bridge protection and rehabilitation: chemical physical
techniques, service life estimates, SHRP-S-668, National Research Council,
Washington, DC.
Weyers, R. (1998), Service life model for concrete structures in chloride laden
environments, ACI Mat. J., V.95, No.4, 445-453.
Zemajtis, J. and Weyers, R. (1996), Concrete bridge service life extension using
sealers in chloride ladden environments, Transportation Research Record No.
1561, Nov. 1996, 1-5.