See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/250328441

Influence of Cracks on the Service Life of Concrete Structures in a Marine

Environment

Article in Key Engineering Materials · January 2009

DOI: 10.4028/www.scientific.net/KEM.399.153

CITATIONS READS

26 1,308

3 authors, including:

Liviu Marsavina Geert De Schutter

Polytechnic University of Timisoara Ghent University
324 PUBLICATIONS 4,238 CITATIONS 619 PUBLICATIONS 15,265 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Multiscale thermo-chemo-physical simulation of ASR View project

MTKD-CT-2004-014058: Modern Composite Materials Applied in Aerospace, Civil and Mechanical Engineering: Theoretical Modelling and Experimental Verification View
project

All content following this page was uploaded by Geert De Schutter on 06 January 2014.

The user has requested enhancement of the downloaded file.

Key Engineering Materials Vol. 399 (2009) pp 153-160
online at http://www.scientific.net
© (2009) Trans Tech Publications, Switzerland

Influence of Cracks on the Service Life of

Concrete Structures in a Marine Environment

Katrien Audenaert1, a, Liviu Marsavina2,b and Geert De Schutter1,c

1
Magnel Laboratory for Concrete Research, Department of Structural Engineering,
Ghent University, Technologiepark-Zwijnaarde 904, 9052 Ghent , Belgium
2
Department of Strenght of Materials, ‘Politehnica’ University of Timisoara,
Blvd. M. Viteazu No. 1, Timisoara 3000222, Romania
a b c
Katrien.Audenaert@UGent.be, msvina@mec.upt.ro, Geert.DeSchutter@UGent.be

Keywords: chloride penetration, concrete, cracks, durability, service life

Abstract. Chloride initiated reinforcement corrosion is the main durability problem for concrete
structures in a marine environment. If the chlorides reach the reinforcement steel, it will depassivate
and start to corrode in presence of air and water. Since the corrosion products have a larger volume
than the initial products, concrete stresses are induced, leading to spalling and degradation of the
concrete structures.
If cracks, caused by early drying, thermal effects, shrinkage movements or overstress, are present in
the concrete, the penetration of chlorides is much faster compared to uncracked concrete. In this
way, the corrosion process is initiated earlier and the service life is decreasing drastically.
In order to study the influence of existing cracks in concrete structures on the penetration of
chlorides a test program was set up at the Magnel Laboratory for Concrete Research of Ghent
University, Belgium in cooperation with the “Politehnica” University of Timisoara, Romania.
The first part of the test program consists of concrete specimens with artificial cracks. The chloride
penetration into the concrete was realised with a non-steady state migration test and modelled with
the finite element method COSMOS/FFE Thermal software.
Based on the experimental and numerical results, a crack influencing factor was determined. With
this factor, the resulting service life of the cracked concrete construction is determined and
compared with the original service life.

Introduction
Concrete is a widely used material for structures in a marine environment. The main durability
problem for concrete structures in a marine environment is chloride initiated reinforcement
corrosion. The chlorides, which are present in the sea water, penetrate into the concrete by the
action of diffusion, capillarity or permeation or a combined action of these transport mechanisms. If
these chlorides reach the steel reinforcement, the corrosion process will start in the presence of
oxygen and water. Since the corrosion products have a larger volume than the initial products,
concrete stresses are induced, leading to spalling and degradation of the concrete structures. In
designing structures for a marine environment, certain choices have to be made in order to guarantee
the required service life: concrete composition, thickness of the concrete cover,…. By modelling
and analysing the service life, the transport mechanisms of chloride ions into the concrete cover and
the degradation processes should be related. However these models assume that the concrete cover
is a homogenous material. In practice, the concrete cover is mostly of lesser quality in comparison
to the quality of the concrete mass. This is caused on the one hand by early drying, creating a more
porous structure and on the other hand by thermal effects (hydration shrinkage), shrinkage
movements and stresses, creating (small) cracks. This leads to significantly different properties in
the concrete cover, which is of major importance because this cover should be the first protection to
aggressive substances.

All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the
publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 157.193.62.188-16/09/08,10:17:38)
154 Advances in Strength of Materials

This paper describes the results of an experimental and numerical study of the influence of cracks
on the chloride penetration in concrete, carried out at the Magnel Laboratory for Concrete Research
of Ghent University, Belgium in cooperation with the “Politehnica” University of Timisoara,
Romania. Firstly, the experimental results will be presented and commented, followed by the
numerical modelling. Based on these results, a crack influencing factor will be determined. In the
last part of the paper, the influence of cracks on the service life will be discussed based on this crack
influencing factor.

Experimental program
Concrete composition. Two types of concrete with the same W/C ratio and based on ordinary
Portland cement (CEM I 52.5 N) were investigated. T1 and T2 have a cement content of 300 kg/m³
and 400 kg/m³ respectively. The water content was respectively 150 kg/m³ and 200 kg/m³. Sand 0/4
and river gravel 2/8 and 8/16 were added. The compressive strength, measured on cubes with side
length 150 mm, at the age of 28 days was respectively 57.4 MPa and 49.6 MPa. For each concrete
composition, eight concrete batches have been produced, one for each crack configuration (see next
section). For each mix, 20 specimens for the determination of the chloride penetration were cast.
The concrete specimens were stored in a climate room at 20 °C ± 2 °C and more than 90 % R.H.
until the testing age.
Crack configuration. As shown in Figure 1, concrete specimens were made with artificial cracks
by means of the positioning and removal after approximately 4 hours of thin copper sheets inside
the specimen. These copper sheets had a thickness of 0.2 mm or 0.3 mm. The copper plates were
placed at a depth in the concrete of 5 mm, 10 mm, 15 mm or 20 mm. As a result, the surface of the
artificial cracks or notches contained more cement than the surface of a real crack in a concrete
element. After hardening of the concrete specimens, cores were drilled from the concrete surface
with a diameter of 100 mm and a thickness of 50 mm (Fig. 1b).
Chloride penetration. The ingress of chloride ions into concrete is realised with the non-steady
state migration test, following the method of [1] as described in [2] and shown in Fig. 1c, leading to
results within a few hours. Chloride diffusion tests are also started, but it will take much longer time
period to obtain experimental results.
For the un-notched specimens, the test duration was 24 hours, as prescribed by [2]. For the notched
specimens the test duration was shortened in order to prevent complete penetration. The test
duration was varied from 2 to 4, 6, 8 and 10 hours. Three specimens were tested for each test
duration. In this way, the evolution of the penetration front can be visualized. After the test, the
specimens were axially split, perpendicular to the notch (Fig. 1d). On each of the freshly split
sections, a 0.1 M AgNO3 solution was sprayed and the chloride penetration depth is measured at 15
points, visually checking the presence of white silver chloride precipitation (Fig. 1e). This
colorimetric method is described in [3]. For each concrete core, two penetration profiles were
obtained. In this way, 6 penetration profiles are obtained for each composition, notch configuration
and test duration. The mean penetration profile for crack width 0.3 mm and crack depth 20 mm for
T2 is given in Fig. 2. On the horizontal axis, the distance to the crack (or notch) is given. Distance 0
indicates the crack location. As the CTH tests were not carried out on a constant concrete age, a
correction is needed in order to take into account the time dependency of the chloride migration
coefficient and to compare the penetration depths. A reference concrete age of 28 days was chosen
[4, 5].
Test results. In order to analyse the test results from the chloride penetration profiles, the
information of each profile was reduced to 3 figures: the maximum and mean penetration depth and
the penetration depth in the end zones. The maximum penetration depth of the crack zone is
calculated as the mean value of the penetration depths at the notch and at 3 mm distance from the
notch. These values are given in Table 1. The mean value in the end zones (from -50 mm to -30 mm
and from 30 mm to 50 mm), is supposed not to be influenced by the notch.
 Key Engineering Materials Vol. 399 155

a. Concrete blocks b. Notched specimen

e. Penetration d. Splitting specimen c. CTH test

Fig. 1: Experimental procedure

50
Penetration
[mm] 2h 4h
40
6h 8h
30 10h

20

10

0
-50 -40 -30 -20 -10 0 10 20 30 40 50
Distance from crack [mm]

Fig. 2: Chloride penetration depth for crack width 0.3 mm and crack depth 20 mm for T2

Discussion of test results. From the experimental results, the influence of crack depth, crack width
and the amount of cement is studied. More information on these aspects could be found in [6]. Both
for T1 and T2 and for both crack widths, a higher chloride penetration depth is noted for a higher
crack depth. This is more pronounced for longer test durations. The influence of the crack width is
less pronounced. For T2 and in earlier research [7, 8], studying the chloride penetration by
immersion in mortar prisms with notches, smaller crack widths lead to smaller penetration depths.
In the experimental tests on real cracks with a crack width up to 0.2mm, it is found that for an
increasing crack width, an increasing penetration depth is obtained, certainly for crack widths
between 0 and 0.1mm. For crack widths between 0.1mm and 0.2mm, this increase is less clear and
seems to be independent from the crack width [9].
By increasing the amount of cement from 300 kg/m³ for T1 to 400 kg/m³ for T2 at a constant
W/C ratio, the maximum and mean penetration depth are decreasing. This could be explained by the
higher amount of chlorides that is bound to the cement matrix.
156 Advances in Strength of Materials

Table 1: Maximum penetration depth (mm)

Crack Crack Duration of test (h)
width depth
2 4 6 8 10
(mm) (mm)
T1 0.3 5 10.4 13.4 17.2 16.6 16.7
0.3 10 10.0 12.2* 20.8 30.6
0.3 15 12.4 25.0 25.1 27.6 35.3
0.3 20 26.0 29.8 35.1 31.8 33.5
0.2 5 10.4 15.2 18.6 19.3 29.2
0.2 10 14.0 24.4 29.8 27.3 26.7
0.2 15 19.2 22.1 22.5 30.1 34.2
0.2 20 26.4 32.7 36.3 31.1 36.4
T2 0.3 5 5.4 8.5 12.3 16.3 17.7
0.3 10 6.0 10.8 14.5 22.1 22.9
0.3 15 17.3 21.2 25.5 26.8 33.2
0.3 20 24.1 30.9 33.3 38.6 39.7
0.2 5 7.1 9.8 12.1 12.4 15.8
0.2 10 8.2 14.4 16.7 15.9 22.6
0.2 15 14.5 20.2 24.7 21.8 26.3
0.2 20 9.5 22.7 25.7 28.4 33.4
*: test duration 5 hours

umerical simulation
Simulation method. The transport of chloride ions in concrete by diffusion is mostly modeled by
Fick’s second law, a linear diffusion equation. For one dimension, this equation is written as:
∂C ∂ ∂C (1)
= D
∂t ∂x ∂x
where C is the total chloride content, t time and D the diffusion coefficient. If the following
boundary conditions are considered:
• a single spatial dimension x, ranging from 0 to ∞ for the semi – infinite case,
• C = C0 at x = 0 and t > 0 (boundary condition),
• C = 0 at x >0 and t = 0 (initial condition).
and the diffusion coefficient is supposed to be constant, the analytical solution of the equation (1)
has the form:
  x 
C = C 0 1 − erf   (2)
  2 D t 
 

in which C0 is a constant, erf( ) represents the error function. This solution is criticised for its use in
cement based materials because it is unable to take into account a non-constant diffusion
coefficient. This diffusion coefficient is depending on the depth in the concrete, on the chloride
concentration, time,… It should also be mentioned that analytically many problems arise when
Fick’s second law is integrated for finite geometries with discontinuities like notches and cracks.
For the modelling, the COSMOS/FFE Thermal software was used to solve numerically the diffusion
equation (1). A transient analysis was used, which implies that the total concentration at any given
point in the medium varies with time. In the transient analysis, nonlinearity was produced, when a
 Key Engineering Materials Vol. 399 157

diffusion coefficient dependent on the total concentration is considered. The Newton–Raphson

method is used for the iterative solution of the resulting set of non-linear equations.
Simulation parameters. Based on symmetry and in order to reduce the size of the model a quarter
of the cylindrical specimen was modelled. The 3D model geometry with a notch of 20 mm depth
and 0.2 mm width is shown in Fig. 3. Solid 3D - 20 node linear elements were used for meshing the
model.

Fig. 3: Geometries and applied boundary conditions for notched specimen 20 x 0.2 mm

The boundary conditions were applied as follows: the insulation boundary condition (0 flux) for
the lateral edges and the applied total chloride concentration on the notch edges and on the exposed
surface to the chloride solution, Fig. 3. The values for the diffusivity constants were determined
experimentally from migration tests using un-notched specimens [10-12]. The non-steady-state
migration coefficient was then corrected in order to take into account the age of the concrete
according with [4, 12, 13], and with the binding [14].
The simulation time was calculated in function of the diffusion coefficient and the mean
penetration obtained experimentally with eq. (2). For the notched models the mean penetration
depth was calculated in the un-notched part of the specimen, and corresponding to the 10 hours
experiment. The steps of the transient analysis were chosen according with the time intervals from
experimental tests (2, 4, 6, 8, and 10 hours).
Results and discussion. A polynomial interpolation was applied in order to determine the
penetration corresponding to the same concentration that was obtained experimentally by using the
silver nitrate solution (1.183 kgCl-/m³). Fig. 4 presents the chloride profile for the 20 x 0.2 mm
notch model after 2, 4, 6, 8, and 10 hours. The chloride penetration was interrogated in a series of
position points with 1 mm step up to 10 mm distance from notch and with 5 mm step far from
notch. The penetration corresponding to the silver nitrate solution used in the experimental
determination was determined by polynomial interpolation. A good agreement of the experimental
and numerical values for the chloride penetration was obtained. Both for experiments and
simulation the maximum penetration depth is increasing with the increasing of the test duration and
with an increasing notch depth. The increased penetration depth for an increasing notch depth is
more pronounced for longer test durations.
Crack influencing factor. The influence of the cracks on chloride penetration was quantified by
comparing the cracked specimens with uncracked specimens. The ratio of the chloride migration
coefficient of the cracked concrete zone to the migration coefficient of the uncracked specimens was
called the crack influence factor η, depending on the crack depth. The crack width was not taken
into account because for crack widths of 0.2 mm and 0.3 mm no influence was found of the crack
width. Based on other research [7], the following equation is proposed:
158 Advances in Strength of Materials

  d b 
η = expa    (3)
  d 0  
With d = crack depth (mm) and d0 = 1mm. The constants a and b are determined with a numerical
regression analysis, based on the least square analysis method. For T1 a and b are respectively
0.1409 and 0.4507 with a R² of 0.96, for T2 respectively 0.0082 and 1.3534 with a R² of 0.99. In
earlier research on chloride penetration in mortar specimens with notches, 0.5202 was found for b
[7]. Introducing this value, the constant a is respectively 0.1163 and 0.0846 and R² respectively 0.95
and 0.98 for T1 and T2. If the amount of cement C (kg/m³) is taken into account, the constant a is
written as:
C
a=k (4)
C0
With C0 = 250 kg/m³, k is determined as 0.075 and R² is respectively for T1 and T2 0.96 and 0.98.
More research is carried out in order to verify equation (4) and the numerical values of b and k.

Fig. 4: Evolution of chloride penetration for the notched specimen T1 - 20 x 0.2 mm

Service life prediction. For service life prediction it is crucial to take into account the time
dependency of the chloride diffusion coefficient. This diffusion coefficient decreases in time due to
ongoing hydration and refining of the pore structure. This leads to a slower penetration of chlorides
into the concrete. This time dependency is taken into account by substituting D in equation (2) by
the apparent diffusion coefficient Da [15]:
D0  t ex   t ex   t 0 
1− n 1− n n

Da = 1 +  −     (5)
1 − n  ∆t   ∆t   t 
With D0 and t0 a pair of known diffusion coefficient and age of the concrete, tex the age of the
concrete at the start of exposure to chlorides, ∆t the exposure duration and n the age factor. As
illustration of the influence of cracks in concrete, Fig. 5 is given. In this figure, T2 is considered in a
marine environment with a chloride concentration C0 of 0.35%. The critical chloride concentration
is considered Ccr = 0.15% (in mass% of the concrete). The age factor n is set at 0.23 [4]. The
 Key Engineering Materials Vol. 399 159

concrete has a W/C ratio of 0.5, a cement content of 400 kg/m³ and a diffusion coefficient of 6 10-12
m²/s at 28 days. The concrete is exposed to chlorides at 28 days. In Fig. 5, the critical chloride
penetration depth is given in function of time for different crack depths from 0 to 20 mm. From Fig.
5 it is clear that in an uncracked concrete with a concrete cover thickness of 40 mm, it takes 44
years for the chlorides to reach the critical concentration at the reinforcement, which will start to
corrode. In cracked concrete, with a crack depth of 20 mm, the corrosion process will start after only
19 years.
80
Critical chloride concentration depth (mm)

70

60

50

40

30 uncracked
crack depth = 5mm
20 crack depth = 10mm
crack depth = 15mm
10 crack depth = 20mm

0
0 10 20 30 40 50 60 70 80 90 100
Time (year)

Fig. 5: Influence of crack depth on critical chloride penetration depth in function of time

In Fig. 6, the time needed for the reinforcement to start to corrode, with a concrete cover thickness
of 40 mm and a diffusion coefficient of 2.5, 5 or 7.5 10-12 m²/s is given for different crack depths.
From this figure it is very clear that the influence of cracks is enormous.

140

120

100
Service life (years)

80

60

40

20 2,5
Diffusion
5 coefficient
0 -12
7,5 (10 m²/s)
0 5 10 15 20
Crack depth (mm)

Fig. 6: Relation between service life, diffusion coefficient and crack depth

Conclusion
In this paper, the influence of artificial cracks, so-called notches, with different crack widths and
crack depths on the chloride penetration is experimentally and numerically studied and related to the
service life. Some general conclusions could be made:
- A higher penetration of chlorides is obtained at the notch tip than in the ‘uncracked’ part of the test
specimens. The penetration depth is increasing with an increasing notch depth. This effect is more
pronounced for longer test durations
- The total chloride distribution is not influenced by the notch width (in the range 0.2-0.3mm).
 160 Advances in Strength of Materials

- Numerical simulation based on finite element method represents a good choice for modelling the
chloride penetration using a variable diffusion coefficient with total concentration and a transient
analysis. It can be concluded that the numerical results agree fairly well with experimental results
obtained from migration test for the notched specimens.
- A crack influencing factor was reduced, which is used to illustrate the influence of cracks on the
service life of a concrete structure in a marine environment, leading to the conclusion that the
presence of cracks easily halves the service life duration.

Acknowledgements
This work was done in the framework of Bilateral Scientific Agreement between Ghent
University, Belgium and “Politehnica” University of Timisoara, Romania. Also the financial
support of the FWO-Flanders is greatly acknowledged.

References
[1] Tang, L., Nilsson, L-O., ‘Rapid determination of chloride diffusivity of concrete by applying an
electric field’, ACI Materials Journal, 89 (1992) 49-53.
[2] NT BUILD 492, ‘Chloride migration coefficient from non-steady-state migration experiments’,
1999.
[3] Otsuki, N., Nagataki, S., Nakashita, K., ‘Evaluation of AgNO3 solution spray method for
measurement of chloride penetration into hardened cementations matrix materials’, ACI
Materials Journal, 89 (6) (1992) 587-592.
[4] Audenaert, K., ‘Transport mechanisms in self-compacting concrete in relation to carbonation
and chloride penetration’ (in Dutch), PhD Thesis, Ghent University, 2006.
[5] Audenaert, K., Boel, V., De Schutter, G., ‘Chloride migration in self-compacting concrete’,
CONSEC ’07, France, 2007.
[6] Audenaert, K., De Schutter, G., Marsavina, L., ‘Influence of crack width and depth on the
chloride penetration in concrete’, Geodim, France, 2008.
[7] De Schutter, G., ‘Quantification of the influence of cracks in concrete structures on carbonation
and chloride penetration’, Magazine of Concrete Research, 51 (6) (1999) 427-435.
[8] De Schutter, G., Audenaert, K., ‘Penetration of aggressive substances in cracked concrete’,
Proc. of the IVth Int. Symposium on Concrete for a sustainable Agriculture, Ghent, 2002.
[9] Audenaert K., De Schutter G., Marsavina L., ‘Influence of cracks and crack width on
penetration depth of chlorides in concrete’, European Journal of Civil and Environmental
Engineering, submitted.
[10] Tang, L., Nilsson, L-O. ‘A numerical method for prediction of chloride penetration into
concrete structures’. Jennings H. et al. eds. The Modelling of Microstructure and its Potential
for Studying Transport Properties and Durability, Kluwer Academic Press Publishers, 1996.
[11] Martin-Perez, B., Pantazopoulou, S.J., Thomas, M.D.A., ‘Numerical solution of mass transport
equations in concrete structures’, Computers and Structures (2001) 79 1251-1264.
[12] Marsavina, L., De Schutter, G., Audenaert, K., Marsavina, D., Faur, N., ’Experimental and
numerical determination of the chloride penetration in cracked concrete, Construction and
Building Materials, accepted for publication
[13] Audenaert, K., De Schutter, G., Marsavina, L., ‘Influence of cracks and crack width on
penetration depth of chlorides in concrete’, Medachs 08, Lisbon (2008)
[14] Tang, L., ‘Chloride transport in concrete – Measurement and prediction’, P96-6, Chalmers
University of Technology, Göteborg, 1996.
[15] Tang, L., Gullikers J., ‘On the mathematics of time dependent apparent chloride diffusion
coefficient in concrete’, Cement and Concrete Research (2007) 37 589-595.

View publication stats