Vous êtes sur la page 1sur 11

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

net/publication/261185637

Enhancement in Chloride Diffusivity due to Flexural Damage in Reinforced


Concrete Beams

Article  in  Journal of Materials in Civil Engineering · April 2014


DOI: 10.1061/(ASCE)MT.1943-5533.0000836

CITATIONS READS
4 178

1 author:

Walid Al-Kutti
Imam Abdul Rahman bin Faisal University
23 PUBLICATIONS   102 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Walid Al-Kutti on 27 August 2014.

The user has requested enhancement of the downloaded file.


Enhancement in Chloride Diffusivity due to Flexural
Damage in Reinforced Concrete Beams
Walid A. Al-Kutti 1; Muhammad K. Rahman 2; Mohammed A. Shazali 3; and Mohammed H. Baluch, M.ASCE 4

Abstract: A multiphysics formulation for chloride diffusion in an RC beam with stress-induced damage quantifying the enhancement in
chloride diffusivity due to damage is presented. An experimental investigation involving measurement of chloride profile was conducted on
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

RC beams damaged under applied flexural stress. Numerical simulation of the RC beam is carried out using a two-dimensional finite-element
approach incorporating the damage due to the applied stress, chloride binding, and the chloride diffusion in the model. Concrete is assumed to
be a perfectly elastoplastic (Drucker-Prager) material and the steel as an elastoplastic (von Mises) material with hardening. Drucker-Prager
parameters, cohesion c, and friction angle φ are obtained by calibrating numerical load-deflection (P-Δ) curve to an experimentally deter-
mined (P-Δ) plot for beams loaded in flexure. Defining a scalar damage index as the degradation in elastic modulus expressed in terms of total
strains, the chloride transport problem is addressed, using an effective diffusion coefficient, Deffd , expressed as a function of the damage index
and chloride binding and obtained by calibrating to data for chloride profiles as determined in flexurally damaged beams. Using the ex-
pressions for the effective diffusion coefficient, Deffd , the chloride profiles are shown to match the experimentally determined chloride profiles
in beams damaged at various stress levels. DOI: 10.1061/(ASCE)MT.1943-5533.0000836. © 2014 American Society of Civil Engineers.
Author keywords: Chloride diffusion; Reinforced concrete beam; Elastoplastic modeling; Damage index; Tension damage; Compression
damage; Chloride binding.

Introduction salt ponding tests by fitting profiles of acid-soluble chloride con-


centration versus depth using Crank’s solution to Fick’s law. A
The chloride diffusion coefficient is a key parameter for assessing numerical approach using the finite-element or finite-difference
the service life of concrete structures subjected to the deicing salt, method has also been adopted in several researches to predict the
marine environment and the hot, arid and chloride-laden ambient complex phenomenon of chloride diffusion in concrete coupled
environment in the coastal areas. In the Arabian Gulf Region, the with moisture diffusion and various factors affecting the chloride
harsh environment has resulted in significantly reduced service life ingress in concrete. Xi and Bazant (1999) developed a mathemati-
of concrete structures in a short span of time, requiring extensive cal model to predict the penetration of chloride ions in saturated
repairs. The chloride diffusion coefficient in concrete is influenced concrete incorporating the influence of chloride binding, water
by several factors including properties of concrete constituents, to cement (w/c) ratio, curing time, cement type, and aggregate con-
concrete mix design, type of binder, temperature, age at exposure, tents. Saetta et al. (1993) used coupled diffusion and convection of
concrete moisture content, and environmental conditions. The in- chloride ions and Nielsen and Geiker (2003) employed Fick’s law
fluence of these parameters has been investigated in several re- with modified chloride diffusion coefficient and finite-difference
searches over the past three decades (Gjorv and Vennesland 1979; method to investigate chloride diffusion in partially saturated con-
Page et al. 1981; Sergi et al. 1992; Tang and Nilsson 1992; Mangat crete. Ababneh et al. (2003) used a multiscale model for predicting
and Molloy 1994; Thomas and Bamforth 1999; Song et al. 2008). chloride penetration in nonsaturated concrete, and Yuan et al.
Most of the studies conducted have concentrated on undamaged (2011) used a multispecies model to describe chloride transport
concrete specimens in which the experimentally determined chlo- in saturated concrete.
ride profile is obtained across the depth and an analytical approach In concrete structures in service, the reinforced concrete mem-
is adopted to predict the chloride diffusion coefficient. The apparent bers are subjected to environmental and mechanical loading, which
chloride diffusion coefficient is computed from bulk diffusion or results in compressive and tensile stresses and associated cracking
in members. In a stressed concrete element, the chloride diffusion
1
Dept. of Construction Engineering, Univ. of Dammam, P.O. Box 1985, coefficient is expected to increase significantly due to microstruc-
Dammam 31451, Saudi Arabia. tural damage induced under applied stress. In the undamaged con-
2
Center for Engineering Research, Research Institute, King Fahd Univ. crete specimen, the chloride diffusion coefficient depends on pore
of Petroleum & Minerals, Dhahran 31261, Saudi Arabia (corresponding size, pore distribution, and interconnectivity of the pore system and
author). E-mail: mkrahman@kfupm.edu.sa is also significantly influenced by physical adsorption and chemical
3
INCO Precast Engineering, Industrial Contractors Co. Ltd., P.O. Box reaction between the chloride ions and solid skeleton in the con-
437, Al-Khobar 31952, Saudi Arabia. crete. While in a concrete specimen damaged under applied stress,
4
Dept. of Civil Engineering, King Fahd Univ. of Petroleum & Minerals, the chloride diffusion coefficient is additionally influenced by the
P.O. Box 5058, Dhahran 31261, Saudi Arabia.
presence of microcracks and macrocracks. The rate of penetration
Note. This manuscript was submitted on September 15, 2012; approved
on April 11, 2013; published online on March 14, 2014. Discussion period of chloride is strongly influenced by the extent of stress-induced
open until September 1, 2014; separate discussions must be submitted for damage. This is analogous to the fact that the permeability of dam-
individual papers. This paper is part of the Journal of Materials in Civil aged concrete is increased by several orders in magnitude.
Engineering, Vol. 26, No. 4, April 1, 2014. © ASCE, ISSN 0899-1561/ The effect of cracks on concrete permeability and chloride pen-
2014/4-658-667/$25.00. etration in concrete by generating artificial crack in the concrete

658 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014

J. Mater. Civ. Eng. 2014.26:658-667.


specimen has been investigated by several authors (Wang et al. influence of stress-induced damage on the transport property of
1997; François and Arliguie 1998; Aldea et al. 1999; Garces- chloride diffusion. This paper presents an approach for the simu-
Rodriguez and Hooton 2003; Ismail et al. 2008; Djerbi et al. 2008). lation of the coupled boundary value problem of chloride transport
The chloride diffusion coefficient increases with the increasing in stress-damaged reinforced concrete flexural members using
crack width for crack widths up to approximately 80 μm, beyond COMSOL multiphysics software. Damage is taken in the form
which the cracks provide free concrete surface promoting two- of a scalar related to the state of strains in the flexural member,
dimensional (2D) chloride diffusion in the specimen. Gérard and and its influence on chloride diffusivity is established by calibrating
Marchand (2000) reported an increase in concrete diffusivity by numerical results to those obtained from suitably designed experi-
a factor ranging from 2 to 10 due to cracks in the concrete. ments. Influence functions are proposed to quantify the influence of
Experimental studies have also been conducted on chloride stress-induced damage and also of cement binding on chloride dif-
penetration in the concrete specimen in which microcracks and fusivity of concrete.
macrocracks were generated by the application of external loading.
With compressive-load-induced microcracking, Samaha and Hover
Chloride Diffusion and Chloride Binding in
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

(1992) determined using a chloride migration test that chloride per-


meability increases by approximately 20% at a stress level above Saturated Concrete
75% of the compressive strength. At lower stress levels there was
no significant effect. Lim et al. (2004) conducted a salt pounding Chloride transport in concrete based on the equation of mass
test on concrete cylinders stressed in compression at 20, 35, and balance in one dimension can be written as
50% of the compressive strength and found no significant effect ∂Ct ∂J c
on the chloride penetration in concrete. þ ¼0 ð1Þ
∂t ∂x
Only limited studies have been conducted on concrete speci-
mens damaged under flexural stress. Gowripalan et al. (2000) stud- where Ct = total chloride content (grams of total chloride per gram
ied the effect of tensile steel area on the chloride diffusivity in the of concrete); and J c = chloride flux (m=s). The total chloride con-
tension and compression zones of concrete cracked in flexure. The tent Ct can be expressed as the sum of the bound chloride Cb and
apparent chloride diffusion coefficient in the tension zone was free chloride Cf , which can be expressed in term of the chloride
found to be higher than in the compression zone. Sahmaran (2007) isotherm or the chloride binding capacity ∂Cb =∂Cf as
found that in flexurally loaded reinforced mortar specimens, an in-
∂Ct ∂C
crease in crack width resulted in an increase in the effective diffu- Ct ¼ Cf þ Cb ¼1þ b ð2Þ
sion coefficient. Guoping et al. (2011) investigated the effect of ∂Cf ∂Cf
stress on chloride ion penetration in concrete. The content of chlo-
ride ions is higher in concrete stressed in tension than in unstressed Taking the diffusive flux of free chloride ions in saturated con-
concrete. For concrete stressed in compression, chloride content crete and combining it with Eqs. (1) and (2), the diffusion of free
depends on stress level. Wang et al. (2011) investigated the chloride chloride in concrete can be written as (Tang and Nilsson 1992;
diffusivity of concrete under three levels of sustained compressive Xi and Bazant 1999)
and flexural stresses and reported an increase in apparent chloride ∂Cf
diffusion under increasing flexural stress and a decrease under in- J c ¼ −De ð3Þ
∂x
creasing compressive stress.
Based on a continuum damage mechanics concept, the effect of  
∂Cf ∂ ∂Cf 1
damage on concrete chloride diffusivity and permeability has been ¼ Fb De where Fb ¼ ∂Cb
ð4Þ
studied by researchers in recent years (Xi and Nakhi 2005). The ∂t ∂x ∂x 1 þ ∂Cf
material degradation is described by the evolution of the material
stiffness, or compliance, in a continuum setting (damage models). where De = intrinsic chloride diffusivity of concrete (m2 =s); Cf =
Xing et al. (2005) developed a parallel diffusion model to deter- free chloride (grams of free chloride per gram of concrete); and
mine the effect of damage on the chloride diffusivity of the Fb = chloride binding influence function. Following Martin-Perez
distressed concrete material. Chatzigeorgiou et al. (2005) reported et al. (2000), of the three chloride binding isotherm models (linear,
a significant increase of the permeability of concrete upon micro- Langmuir, and Freundlich) the Langmuir chloride binding isotherm
cracking and observed a good correlation between the evolution is adopted in which the associated binding influencing function Fb
of damage (material stiffness) and permeability experimentally. can be written as follows:
Pijaudier-Cabot et al. (2009) stipulated that a strong interaction ex- αCf ∂C α 1
ists between material damage and transport properties of concrete. Cb ¼ → b¼ → Fb ¼   ð5Þ
1 þ βCf ∂Cf ð1 þ βCf Þ2 α
1 þ ð1þβC Þ2
In a concrete element with diffused cracking, the material per- f

meability is controlled by damage (decrease of average stiffness


due to microcracking). In the case of localized microcracking
and after a discrete macrocrack has formed, permeability is con- Influencing Function F d for Effect of Damage on
trolled by a power function of the crack opening. Numerical sim- Chloride Diffusion
ulation of chloride migration in concrete cylinders damaged under
compression was carried out recently by Rahman et al. (2012) To consider the effect of mechanical damage and the chloride bind-
adopting a phenomenological damage model and a linear chloride ing capacity on the diffusivity of chloride in concrete, a form of
binding isotherm coupled with a Nernst-Planck equation. The ef- multivariate law for undamaged concrete is adopted in which the
fective chloride diffusion coefficient was found to increase three effective chloride diffusion coefficient Deffd is defined as
times in concrete samples damaged at 90% of the concrete ultimate Deffd ¼ De · Fb · Fd ð6Þ
strength.
A realistic prediction of service life of stressed concrete struc- where De may be regarded as the reference or nominal diffusion
tures in a corrosive environment requires incorporation of the coefficient when the influencing factors (Fb and Fd ) assume values

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014 / 659

J. Mater. Civ. Eng. 2014.26:658-667.


of unity. Fb denotes the influence of the chloride binding and The beam was reinforced with carbon steel having a yield strength
Fd represents the influence of the mechanical damage on chloride of 560 MPa.
transport into concrete. The 2D form of diffusion equation of chlo- After the beams were cured for 28 days under dry laboratory
ride ion for predicting the time and spatial concentration of chloride conditions, two RC beams were loaded by four-point flexural load-
in concrete considering the influence of chloride binding and ing test up to the failure to find the ultimate bending moment capac-
mechanical damage is given as follows: ity in an Instron machine of 300 kN capacity. The strain in the
reinforcement bars at the bottom of the RC beams and the strain
∂Cf ∂ 2 Cf ∂ 2 Cf in concrete in the compressive zone and the deflection at the mid-
¼ Deffd 2
þ Deffd ð7Þ
∂t ∂x ∂y2 span of the beams was measured using a linear variable differential
transformer (LVDT). To generate mechanically induced damage,
Material constitutive models characterized by a loss of stiffness eight RC beams (a pair for each level of load) were loaded at
or a reduction of the elastic moduli introduced by Kachanov (1958) 40, 60, 75, and 90% of the ultimate loading capacity of the beam
have been used to describe the strain-softening behavior of concrete and they were subsequently unloaded. Each pair of beams was sub-
both in tension and in compression. Concrete damage models of sequently loaded back to back in steel frame and the nuts were
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

varying degrees of sophistication have been proposed (Mazars tightened up while monitoring the strain in the reinforcing steel
and Pijaudier-Cabot 1989; Taher et al. 1994; Khan et al. 2007; at the midspan using a data logger. The nuts were tightened to
Voyiadjis et al. 2009). The phenomenological approach suggested achieve strain in reinforcing steel of approximately the same mag-
by Mazars and Pijaudier-Cabot (1989) is used to evaluate a scalar nitude as recorded in the damaged state from the four-point flexural
damage. The influence of damage on the response of the material load tests. Fig. 1 shows the arrangement for stressing the RC beams
based on this approach is based on the degradation of its elastic to the desired stress levels.
stiffness. The uniaxial stress-strain relationship is given by The damaged and undamaged beams were then exposed to 8%
σ ¼ Eε and E ¼ ð1 − dÞEo ð8Þ NaCl solution to simulate marine exposure condition for a period of
3 months. The left and right surfaces of the beams were sealed with
where Eo and E = secant undamaged and damaged moduli, respec- epoxy so that chloride could penetrate only from tension and com-
tively; and d = scalar damage variable. Taher et al. (1994) devel- pression surfaces. After 90 days of exposure, the specimens were
oped an elastodamage model for concrete using a constitutive law cleaned and dried to remove the surface moisture and drilled to
proposed by Popovics (1973) for stress-total strain relation of plain depths of 5, 15, 35, 50, and 75 mm at a distance of 120 mm from
concrete subjected to uniaxial compressive stress given as follows: the support and at the midspan of the RC beam as shown in Fig. 2
 to obtain the powder samples for determining the water-soluble
σ mðεεu and acid-soluble chloride concentration in concrete. The rapid chlo-
¼   ð9Þ
σu m − 1 þ εε m ride test (RCT) and rapid chloride test water (RCTW) systems by
u
Germann Instruments (2006) were used to determine the amount of
where σu and εu = peak stress and strain, respectively; and m = acid-soluble and water-soluble chlorides (free chloride), respec-
parameter dependent on σu as reported by Popovics (1973). Using tively. The powder sample was mixed into a distinct amount of ex-
Eq. (9), a relationship between the moduli and damage variable traction liquid and shaken for 5 min. The extracted liquid removes
could be obtained and the scalar damage parameter d in compres- disturbing ions, such as sulfide ions, and extracts the chloride ions
sion and tension can be obtained as in the sample. A calibrated electrode is submerged in the solution to
determine the amount of chloride ion, which is expressed as per-
E mc − 1 centage of the concrete mass.
d¼1− ¼1−   for εx < 0ðcompressionÞ ð10Þ
Ec mc − 1 þ εεx mc
u

E mt − 1 Simulation of Coupled Diffusion and Mechanical


d¼1− ¼1−   for εx > 0 ðtensionÞ ð11Þ Damage Problem
Et mt − 1 þ εεx mt cr
Based on the model accounting for chloride binding (Shazali et al.
where Ec and Et = undamaged secant modulus in compression 2012), COMSOL finite-element software was used to simulate the
and tension in megapascals, respectively; mc and mt = concrete chloride diffusion in concrete taking into consideration the effect of
material parameters for compressive and tensile-induced damage, flexural-stress-induced damage. The coupled problem of mechan-
respectively; and εx = total strain in millimeters per millimeter. ically induced damage and chloride binding capacity on chloride
Based on the damage model and the ratio of the experimentally diffusivity in concrete was solved using two multiphysics problems
obtained diffusion coefficient of damaged and undamaged samples involving:
Deffd =ðDe · Fb Þ, the damage influence function Fd for tensile- or • The structural damage mechanics partially embodied in
compressive-induced damage can be obtained. Eqs. (8)–(11); and
• The chloride diffusion with binding as described by
Eqs. (2)–(5).
Experimental Program The 2D plane stress reinforced concrete beam was modeled as a
perfect elastoplastic Drucker-Prager (D-P) material with tension
In order to determine experimentally the effect of mechanical dam- cutoff. The D-P yield criterion can be written as
age in tension and compression on the chloride diffusivity, eight RC
pffiffiffi pffiffiffiffiffi
beams (150 × 150 × 1,200 mm) were cast. The concrete mix for fðI; J 2 Þ ¼ J 2 þ αI ¼ k ð12Þ
the RC beam included Type I portland cement at 480 kg=m3 , total
aggregate at 1,725 kg=m3 , and admixture Conplast SP-440 at where I ¼ σkk is the hydrostatic component of the stress tensor;
3.25 kg=m3 . The fine and coarse aggregates were combined such J 2 ¼ ð1=2Þsij sji is the deviatoric stress tensor invariant; and α and
that the coarse aggregate constituted 62% of the total aggregates. k are material constants that can be related to the friction angle φ
The compressive strength of concrete at 28 days was 50 MPa. and cohesion c of the modified Mohr-Coulomb (M-C) by matching

660 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014

J. Mater. Civ. Eng. 2014.26:658-667.


500 mm 200 mm 500 mm

150 mm

150 mm
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Arrangement for the pair of the RC beams loaded back to back using steel frames

Station 1 Station 2 Fig. 2 shows the finite-element modeling including dimensions


and boundary conditions of the RC beams. The steel reinforcement
was modeled as a plate representing the area of reinforcement in the
beam with the same width as the concrete section. A flow chart of
150 mm steps required for the simulation of chloride diffusion in RC beams
is presented in Fig. 3.
120 mm
550 mm
(a)
Results and Discussion
500 mm 100 mm
Cf = 0.3% Mechanical Behavior
Fig. 4 shows the experimentally determined and the numerically
150 mm simulated results for the load-deflection response of the beam
up to failure load. The experimental result shows that the first crack
Cf = 0.3%
developed in the beam at a load of 7.5 kN and the ultimate load
(b) at failure of the beam is 95 kN. The cracking load computed
using American Concrete Institute code equations 9.9 and 9.10
Fig. 2. (a) Stations for experimental determination of chloride profile; (American Concrete Institute 2002) was 7.5 kN and the finite-
(b) finite-element modeling of the RC beams element simulation in COMSOL was 7.25 kN, which match well
with the experimental results. While the load-deflection response
by both ACI and COMSOL simulation show more stiffened behav-
the outer apices of the M–C hexagon with the D–P surface. For ior than the experimental results after the cracking load, the overall
plane stress, the relation between the parameters is as follows: response appears to match well with the behavior of reinforced con-
sin φ 2 crete beam when approaching the ultimate load using cohesion c ¼
α ¼ pffiffiffi ; k ¼ pffiffiffi c cos φ ð13Þ 4.8 Mpa and friction angle φ of 53°.
3 3 Fig. 5 shows the experimental and numerical results of
In this study the value cohesion c ¼ 4.8 MPa and the friction reinforcement strain versus midspan deflection up to failure load.
angle φ ¼ 53° were adopted. These material parameters were found It can be seen that the maximum strain in tensile reinforcement
by calibrating the elastoplastic solution for load-deflection P-Δ is 3,500 με at the maximum midspan deflection of approxi-
diagram for the reinforced concrete beam in flexure to match mately 8 mm.
the experimentally determined P-Δ curve for the specimen. The Table 1 shows the parameters used in the COMSOL model for
von Mises material model was used for the steel reinforcement in simulation of chloride diffusion in damaged concrete. The coeffi-
which Es and fy were 190 GPa and 560 MPa, respectively. The cient of chloride diffusion De , determined experimentally, was
chloride transport problem was consequently solved to advance taken as 3.35 × 10−6 mm2 =s, the free chloride concentration Cf
the solution of the coupled problem of diffusion of chloride, bind- at the boundary was 0.30% per weight of concrete. Initial chloride
ing, and damage. content in the sample, Ci , was assumed to be zero.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014 / 661

J. Mater. Civ. Eng. 2014.26:658-667.


Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Flowchart for the simulation of effect of mechanical damage on chloride diffusivity in RC beams

100 4000

90
Reinforcement Strains at Midspan x 10 -6

Yield of Reinforcement 3500


80
3000
70 A100
60 2500
COMSOL
Load (kN)

100% Loading
50 ACI 2000 75% Loading
40 60% Loading
1500 40% Loading
30
COMSOL
1000
20

10 500
First cracking load at P=8 kN
0
0 1 2 3 4 5 6 7 8 9 10 0
0 1 2 3 4 5 6 7 8 9
Deflection at Midspan (mm)
Deflection at Midspan (mm)
Fig. 4. Experimental load-midspan deflection curve for beam loaded to Fig. 5. Experimental reinforcement strain-midspan deflection curve
failure versus COMSOL simulation for 40, 60, 75, and 100% loaded beams

Mechanical Damage in the beams under flexural loading at first crack load and 40,
Fig. 6 shows the damage index distribution [using the scalar 60, and 100% of ultimate loading is shown. It can be seen from
damage definition as given by Eqs. (10) and (11)] in half of the Fig. 6 that in the constant moment zone at first cracking load,
reinforced beam span (L=2 ¼ 600 mm). Development of the the scalar-tension-induced damage ranges from 0.08 to 0.1 along
stress-induced damage index for compressive and tensile stress a depth of 70 mm from the bottom of the beam. After cracking the

662 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014

J. Mater. Civ. Eng. 2014.26:658-667.


Table 1. COMSOL Model Parameters Chloride Binding Isotherm
COMSOL COMSOL expression The results of free and bound chloride for different RC beams are
commands and parameters Value shown in Fig. 7. As expected, it can be noticed that an increase
Boundary setting Co Cf ¼ 0.30% by weight in the free chloride leads to increase of bound chloride, tending to
of concrete an asymptotic behavior at higher values. From the experimental
Constants mt 1.46 results, a Langmuir isotherm model was found to be the best fit
mc 3.45 and the model parameters were determined to be α ¼ 2.39 and
f cr 4 MPa β ¼ 15.6. These parameters are used in the binding capacity influ-
fu 50 MPa
ence function Fb as shown in Eq. (5). The Langmuir isotherm
εcr 1.55 × 10−4
εu 2.3 × 10−3
model can then be described as
Ec 29,000 MPa 2.39Cf
Es 190,000 MPa Cb ¼ ð14Þ
Fys 560 MPa 1 þ 15.6Cf
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

α 2.39
β 15.6 1
Scalar expression Fd 7.5d2 þ 0.8d þ 1 for tension Fb ¼  2.39
 ð15Þ
1 þ ð1þ15.6C Þ2
1.15d þ 1 for compression f

Fb 1=½1 þ α=ð1 þ βCf Þ


Cb αCf =ð1 þ βCf Þ
Ct Cb þ Cf Chloride Profile in RC Beams Subjected to
Stress-Induced Damage
Figs. 8 and 9 show the experimentally determined free and total
damage increases with an increase in the loading, and the scalar chloride profile in sound and damaged concrete after 90 days of
damage is approximately 0.6 to 1 for 40% loaded beam and 0.8 exposure to 0.3% of free chloride. From Fig. 8, it can be noted that
to 1 for 60% loaded beam and the affected depth also increases with an increase in tensile stress, there is a significant increase in
to 90 mm from the bottom of the beam. the free and total chloride. It is believed that this effect manifests
The results have indicated that for loads up to 40% of ultimate itself due to the extension of microcracks in the cement paste as
flexural loading, the compressive mechanical damage index was well as in the interfacial transition zone between the coarse aggre-
zero in the constant moment zone where the maximum compressive gate and the cement paste. Fig. 8 also shows that that at a depth of
stress was found to be approximately 20 MPa. At higher loads, the approximately 12.5 mm, the free chloride content is approximately
compressive-induced damage indicating degradation of the secant 0.14% for concrete damaged under 90% of ultimate flexural load
modulus in compression was noted with an average value of 0.1, capacity, which reduces to 0.09 and 0.05% for 75 and 60% damage.
0.15, and 0.40 for 60, 75, and 90% of ultimate flexural loading, For undamaged specimens, the chloride concentration is 0.02%
with a damaged zone extending about 30 mm downwards from after 90 days of salt ponding. This is indicative of the fact that the
the top surface of the beam. free chloride concentration in damaged concrete can be expected to

Fig. 6. Tensile damage index d at (a) first cracking, (b) 40%, (c) 60%, and (d) ultimate loading

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014 / 663

J. Mater. Civ. Eng. 2014.26:658-667.


0.2
reach the threshold limit to initiate rebar corrosion in damaged
Cb = 2.39 x Cf /( 1+15.6 Cf) specimens at a much earlier age as compared with the undamaged
specimens.
0.15 Based on the flowchart shown in Fig. 3, the effective diffusion
Cb % wt. of Concrete

coefficient Deffd in the damaged concrete subjected to tensile and


compressive stresses was computed and is shown in Tables 2 and 3,
respectively. For undamaged concrete, Deffd ð¼ De Fb Fd ; Fd ¼ 1Þ
0.1
was calibrated to chloride profiles and found to be 2.1 ×
Undamaged
10−6 mm2 =s. For the tensile damaged concrete corresponding to
40%Loading stress levels of 40, 60, 75, and 90% of the ultimate flexural loading,
0.05 60%Loading the effective diffusion coefficient Deffd was determined to be 7.4 ×
75%Loading 10−6 , 1.7 × 10−5 , 2.1 × 10−5 , and 2.2 × 10−5 mm2 =s, respectively
(Table 2). This indicates a significant increase of diffusivity of up to
90%Loading
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

eight times in damaged concrete subjected to tensile stress. The


0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
tensile damage index is a reasonable estimate of the true tensile
Cf % wt. of Concrete scalar damage because it reaches its maximum value of 0.98 when
the tensile strain is 0.0013, which is close to the ultimate softening
Fig. 7. Relationship between free and bound chloride strain of concrete in tension.
For damaged concrete in the compressive zone, the chloride
diffusivity in concrete is presented in Table 3. For the concrete
damaged in compression, corresponding to stress levels of 60,

0.3
Table 2. Correlation between Damage Function Fd and Chloride
Undamaged
Diffusivity at Various Tensile Stress Levels
0.25
Level of Tensile strain Fd ¼ Deffd =
40%Loading loading εx ðmm=mmÞ d Deffd ðmm2 =sÞ ðDe · Fb Þ
Cf % wt. of Concrete

0.2
60%Loading 0 0 0 2.1 × 10−6 1.0
40 1.3 × 10−4 0.62 7.4 × 10−6 2.9
0.15
75%Loading 60 1.9 × 10−4 0.74 1.7 × 10−5 6.7
75 6.8 × 10−4 0.95 2.1 × 10−5 8.3
90%Loading 90 13 × 10−4 0.98 2.2 × 10−5 8.7
0.1

0.05
Table 3. Correlation between Damage Function Fd and Chloride
Diffusivity at Various Compressive Stress Levels
0
0 5 10 15 20 25 30
Level of Compressive strain Fd ¼ Deffd =
Depth in mm loading εx ðmm=mmÞ d Deffd ðmm2 =sÞ ðDe · Fb Þ
0 0 0 2.0 × 10−6 1.00
Fig. 8. Free chloride profile in damaged and undamaged beams 60 1.6 × 10−3 0.10 2.9 × 10−6 1.15
75 1.8 × 10−3 0.15 3.1 × 10−6 1.23
90 2.6 × 10−3 0.40 3.6 × 10−6 1.43

1
0.45 0.9

0.4 Undamaged 0.8

0.35 40%Loading 0.7


Damage Index, d
Ct % wt. of Concrete

0.3 0.6
60%Loading
0.5
0.25 75%Loading Compressive
0.4 Damage
0.2 90%Loading
0.3
Tensile
0.15 Damage
0.2
0.1
0.1
0.05 0
0 0.001 0.002 0.003 0.004 0.005 0.006
0
0 5 10 15 20 25 30
Total Strain ∋x (mm/mm)
Depth in mm
Fig. 10. Relationship between the scalar damage d and total strain εx at
Fig. 9. Total chloride profile in damaged and undamaged beams the maximum moment zone

664 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014

J. Mater. Civ. Eng. 2014.26:658-667.


75, and 90% of the ultimate flexural loading, the effective diffusion Based on the damage model and the ratio of diffusion coefficient
coefficient Deffd was computed to be 2.9 × 10−6 , 3.1 × 10−6 , and of damaged and undamaged samples Deffd =ðDe · Fb Þ), the damage
3.6 × 10−6 mm2 =s, respectively. Up to 40% of ultimate flexural influence function Fd was established as follows (Fig. 11):
loading, there was no increase in the chloride diffusivity in concrete
in the compression zone, while at higher levels of compressive For tensile-induced damage∶Fd ¼ Deffd =ðDe · Fb Þ
stress, there was approximately 50% increase in the chloride dif- ¼ 7.5d2 þ 0.8d þ 1 ð16aÞ
fusivity in damaged beams as compared with the control undam-
aged beams. In order to corroborate how reasonable the damage
index linked to degradation of the elastic modulus is to the true For compression-induced damage∶Fd ¼ Deffd =ðDe · Fb Þ
scalar damage as computed in plastodamage theories in continuum ¼ 1.158d þ 1 ð16bÞ
mechanics, Fig. 10 shows the plot of the evolution of tensile
and compressive damage indexes in terms of total strain. The peak It is not surprising that the influence of damage on chloride
damage index in tension is quite close to the ultimate softening diffusivity is far more significant in tensile zones than in compres-
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

strain of concrete in tension, with a reasonable correlation between sive zones because cracks in the tensile zones form parallel to the
compressive-induced damage and strain as well. direction of chloride ingress, whereas, in compression, they are
approximately transverse to the direction of flow.
10.00
9.00
Simulation of Chloride Diffusion in Damaged Concrete
y = 7.5d2 + 0.8d + 1
8.00
R² = 0.95 Using the calibrated models obtained for effective diffusivity, Deffd ,
7.00
Fd = Deffd/(Fcb x Do)

as given by Eqs. (16a) and (16b), this next paragraph focuses on


Exp. In
6.00
Tensile simulation of chloride diffusion in a flexurally damaged reinforced
5.00 concrete beam.
4.00
Two-dimensional numerical simulation was conducted for the
Exp. In
Comp. RC beam using COMSOL software and the numerical results were
3.00
compared to the experimental results obtained after 90 days chlo-
2.00 y = 1.15d + 1 ride exposure period. Fig. 12 shows the 2D free chloride distribu-
1.00 R² = 0.94 tion in the RC beam at first cracking and 40, 60, and 90% of
0.00 ultimate flexural loading. As expected, an increase of mechanical
0 0.2 0.4 0.6 0.8 1 1.2 damage in both compressive and tensile zone will increase the
Damage d chloride penetration into concrete and the influence of tensile dam-
age is higher than the compressive damage. A comparison between
Fig. 11. Relationship between mechanical damage and Deffd
the experimental and numerical results of free chloride profiles at a

Fig. 12. Free chloride distribution in RC beams subjected to different flexural loading: (a) undamaged; (b) 40% loading; (c) 60% loading; (d) ultimate
loading

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014 / 665

J. Mater. Civ. Eng. 2014.26:658-667.


0.3
accounts for the influence of chloride binding for which the Lang-
Undamaged muir model was used. It has been shown that stress-induced dam-
0.25
40%Loading
age in flexural reinforced concrete members influences chloride
diffusivity significantly in tension zones and moderately in com-
60%Loading
Cf % wt. of Concrete

0.2 pression zones. This effect of damage on chloride diffusivity is pre-


75%Loading sented in the form of influence functions, relating the diffusivity to
90%Loading the level and type of damage.
0.15
COMSOL In undamaged concrete, the effective diffusion coefficient,
Deffd , was found to be 2.1 × 10−6 mm2 =s, while for the damaged
0.1 concrete with tensile damage levels corresponding from 40 to 90%
of the ultimate flexural load, the ratio of damaged to undamaged
0.05 concrete diffusivity was found to range from approximately 300 to
900%. For the compressive damage zone, the diffusivity showed a
moderate increase of up to 40% corresponding to 90% of ultimate
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

0
0 5 10 15 20 25 30 flexural loading. Influence of compression damage on diffusivity is
Depth in mm not as pronounced as that due to tension-induced damage, as in the
tensile zone the orientation of cracks is in the direction of chloride
Fig. 13. Chloride profiles using the model and experimental values for diffusion, whereas in the compression zone major segments of the
different tensile damage levels (120 mm from support) cracks are perpendicular to the path of chloride transport. The in-
fluence of binding phenomenon on chloride diffusion results in a
reduction of approximately 25% of chloride diffusivity in concrete.
7.0 Using the approach outlined in this paper, the impact of damage
on chloride transport may be used for predicting the reduction in
6.0
service life of stressed reinforced concrete members in a corrosive
Proposed Model
environment. This can be achieved by estimating the ratio of ser-
vice stress to ultimate strength levels in the structural components,
5.0 Xing et al.(2005) correlating it to the damage influence functions that would estimate
He and Gong (2005) the enhancement in chloride diffusivity, which in turn would serve
Deffd/(Fcb x Do)

4.0 to yield the reduced service life in terms of accelerated ingress of


Sakoi and Horiguchi (2006)
chloride to key reinforcement locations.
3.0

2.0
Acknowledgments
This work has been completed under the Deanship of Scientific
1.0 Research funded research project FT090004 “Service Life Assess-
ment of Stressed Concrete Members under Chloride Attack.” The
0.0 authors acknowledge the support of King Fahd University of
0.00 0.20 0.40 0.60 0.80 1.00
/fcr Petroleum & Minerals, the Department of Civil Engineering, and
the Center for Engineering Research at Research Institute in the
Fig. 14. Comparison between the proposed model and experimental pursuit of this research endeavor.
results from literature for tension damage

Notation
distance of 120 mm from the support (for tensile-induced damage) The following symbols are used in this paper:
is shown in Fig. 13 and it can be noted that the proposed model is Cb = bound chloride in grams of chloride per gram of
reasonably predicting the free chloride profile for sound and dam- concrete;
aged concrete. Cf = free chloride in grams of chloride per gram of concrete;
To verify the proposed model in Eq. (16), experimental results Ct = total chloride in grams of chloride per gram of concrete;
reported by Sakoi and Horiguchi (2006), He and Gong (2005), and c = cohesion (MPa);
Xing et al. (2005) were compared with the same level of stress ob- De = chloride diffusion coefficient in undamaged concrete
tained from the COMSOL model at a distance of 120 mm from the (mm2 =s);
support of the RC beams (Fig. 14). The proposed model predicts Deffd = effective chloride diffusion coefficient (mm2 =s);
reasonably well the data obtained for the effective diffusivity d = scalar damage index;
coefficient due to damage in tension as obtained by Sakoi and dc = compressive scalar damage index;
Horiguichi (2006), He and Gong (2005), and Xing et al. (2005). dt = tensile scalar damage index;
E = damage secant modulus (MPa);
Eo = undamage secant modulus (MPa);
Summary and Conclusions Es = Young’s modulus of reinforcement steel (MPa);
Fcb = chloride binding influence function;
A coupled experimental and numerical study of chloride diffusion Fd = mechanical damage influence function;
in RC beams damaged in flexure and exposed to 8% NaCl solution Fys = yield stress of reinforcement steel (MPa);
for 3 months is presented. In the proposed numerical model, a sca- I = hydrostatic component of the stress tensor;
lar damage index is defined in terms of total strains. The model also J 2 = deviatoric stress tensor invariant;

666 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014

J. Mater. Civ. Eng. 2014.26:658-667.


m = material parameter; Mazars, J., and Pijaudier-Cabot, G. (1989). “Continuum damage theory:
mc = material parameter for compression-induced damage; Application to concrete.” J. Eng. Mech., 10.1061/(ASCE)0733-9399
mt = material parameter for tensile-induced damage; (1989)115:2(345), 345–365.
α and β = chloride binding isotherm parameters; Nielsen, E. P., and Geiker, M. R. (2003). “Chloride diffusion in partially
saturated cementitious material.” Cem. Concr. Res., 33(1), 133–138.
ε = uniaxial strain (mm=mm);
Page, C. L., Short, N. R., and El Tarras, A. (1981). “Diffusion of chloride
εcr = tensile cracking strain (mm=mm); ions in hardened cement paste.” Cem. Concr. Res., 11(3), 395–406.
εu = compression peak strain (mm=mm); Pijaudier-Cabot, G., Dufour, F., and Choinska, M. (2009). “Permeability
σ = uniaxial stress (MPa); due to the increase of damage in concrete: From diffuse to localized
σu = compression peak stress (MPa); and damage distributions.” J. Eng. Mech., 10.1061/(ASCE)EM.1943-7889
φ = friction angle (degrees). .0000016, 1022–1028.
Popovics, S. (1973). “A numerical approach to the complete stress-strain
curve of concrete.” Cem. Concr. Res., 3(5), 583–599.
References Rahman, M., Al-Kutti, W., Shazali, M., and Baluch, M. (2012). “Simulation
of chloride migration in compression-induced damage in concrete.”
Downloaded from ascelibrary.org by University of Dammam on 03/30/14. Copyright ASCE. For personal use only; all rights reserved.

Ababneh, A., Benboudjema, F., and Xi, Y. (2003). “Chloride penetration J. Mater. Civ. Eng., 10.1061/(ASCE)MT.1943-5533.0000458, 789–796.
in nonsaturated concrete.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899 Saetta, A. V., Scotta, R. V., and Vitaliani, R. V. (1993). ‘‘Analysis of chlo-
-1561(2003)15:2(183), 183–191. ride diffusion into partially saturated concrete.” ACI Mater. J., 5(90),
Aldea, C., Shah, S., and Karr, A. (1999). “Effect of cracking on water 441–451.
and chloride permeability of concrete.” J. Mater. Civ. Eng., 10.1061/ Sahmaran, M. (2007). “Effect of flexure induced transverse crack and self-
(ASCE)0899-1561(1999)11:3(181), 181–187. healing on chloride diffusivity of reinforced mortar.” J Mater. Sci.,
American Concrete Institute. (2002). “Building code requirements for 42(22), 9131–9136.
reinforced concrete.” ACI 318, Detroit. Sakoi, Y., and Horiguchi, T. (2006). “Loading effects on chloride penetra-
Chatzigeorgiou, G., Picandet, V., Khelidj, A., and Pijaudier-Cabot, G. tion of fiber reinforced concrete.” Proc., 2nd Int. Fib Congress,
(2005). “Coupling between progressive damage and permeability of (CD-ROM), Naples, Italy, 1–7.
concrete: Analysis with a discrete model.” Int. J. Numer. Anal. Methods Samaha, H. R., and Hover, K. C. (1992). “Influence of microcracking
Geomech., 29(10), 1005–1018. on the mass transport properties of concrete.” ACI Mater. J., 89(4),
COMSOL [Computer software]. COMSOL, Burlington, MA. 416–424.
Djerbi, A., Bonnet, S., Khelidj, A., and Baroghel-Bouny, V. (2008). Sergi, G., Yu, S. W., and Page, C. L. (1992). “Diffusion of chloride and
“Influence of traversing crack on chloride diffusion into concrete.” hydroxyl ions in cementitious materials exposed to a saline environ-
Cem. Concr. Res., 38(6), 877–883. ment.” Mag. Concr. Res., 44(158), 63–69.
François, R., and Arliguie, G. (1998). “Influence of service cracking on Shazali, M. A., Rahman, M. K., Al-Gadhib, A. H., and Baluch, M. H.
reinforcement steel corrosion.” J. Mater. Civ. Eng., 10.1061/(ASCE) (2012). “Transport modeling of chlorides with binding in concrete.”
0899-1561(1998)10:1(14), 14–20. Arabian J. Sci. Eng., 37(2), 469–479.
Garces-Rodriguez, O., and Hooton, R. D. (2003). “Influence of cracks Song, H. W., Lee, C. H., and Ann, K. Y. (2008). “Factors influencing chlo-
chloride ingress into concrete.” ACI Mater. J., 100(2), 120–126. ride transport in concrete structures exposed to marine environments.”
Gérard, B., and Marchand, J. (2000). “Influence of cracking on the diffu- Cem. Concr. Compos., 30(2), 113–121.
sion properties of cement-based materials, part I: Influence of continu- Taher, S., Baluch, M. H., and Al-Gadhib, A. H. (1994). “Towards canonical
ous cracks on the steady-state regime.” Cem. Concr. Res., 30(1), 37–43. elastoplatic damage model.” Eng. Fract. Mech., 48(2), 151–166.
Germann Instruments. (2006). “Instruction and maintenance manual.” Tang, L., and Nilsson, L. O. (1992). “Rapid determination of the chloride
Copenhagen, Denmark. diffusivity in concrete by applying an electrical field.” ACI Mater. J.,
Gjorv, O. E., and Vennesland, O. (1979). “Diffusion of chloride ions from 49(1), 49–53.
seawater into concrete.” Cem. Concr. Res., 9(2), 229–238. Thomas, M. D. A., and Bamforth, P. B. (1999). “Modelling chloride dif-
Gowripalan, N., Sirivivatnanon, V., and Lim, C. C. (2000). “Chloride diffu- fusion in concrete: Effect of fly ash and slag.” Cem. Concr. Res., 29(4),
sivity of concrete cracked in flexure.” Cem. Concr. Res., 30(5), 725–730. 487–495.
Guoping, L., Fangjian, H., and Yongxian, W. (2011). “Chloride ion pen- Voyiadjis, G. Z., Taqieddin, Z. N., and Kattan, P. I. (2009). “Theoretical
etration in stressed concrete.” J. Mater. Civ. Eng., 10.1061/(ASCE)MT formulation of a coupled elastic-plastic anisotropic damage model
.1943-5533.0000281, 1145–1153. for concrete using the strain energy equivalence concept.” Int. J. Dam-
He, S. Q., and Gong, J. X. (2005). “Influence of flexural loading on per- age Mech., 18(7), 603–638.
meability of chloride ion in concrete.” J. Build. Mater., 8(2), 134–138 Wang, H., Lu, C., Jin, W., and Bai, Y. (2011). “Effect of external loads on
(In Chinese). chloride transport in concrete.” J. Mater. Civ. Eng., 10.1061/(ASCE)
Ismail, M., Tuomi, A., Francois, R., and Gagne, R. (2008). “Effect of crack MT.1943-5533.0000265, 1043–1049.
opening on the local diffusion coefficient of chloride in cracked mortar Wang, K., Jansen, D. C., Shah, S. P., and Karr, A. (1997). “Permeability
samples.” Cem. Concr. Res., 38(8–9), 1106–1111. study of cracked concrete.” Cem. Concr. Res., 27(3), 381–393.
Kachanov, L. M. (1958). “Time of the rupture process under creep condi- Xi, Y., and Bazant, Z. P. (1999). “Modeling chloride penetration in satu-
tions.” Isv. Akad. Nauk., SSR. Otd. Tekh. Nauk, 8, 26–31. rated concrete.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899-1561(1999)
Khan, A. R., Al-Gadhib, A. H., and Baluch, M. H. (2007). “Elasto-damage 11:1(58), 58–65.
constitutive model for high strength concrete subjected to multiaxial Xi, Y., and Nakhi, A. (2005). “Composite damage models for diffusivity of
loading.” Int. J. Damage Mech., 16(3), 361–398. distressed materials.” J. Mater. Civ. Eng., 10.1061/(ASCE)0899-1561
Lim, C. C., Gowripalan, N., and Sirivivatnanon, V. (2004). “Microcracking (2005)17:3(286), 286–295.
and chloride ion diffusion of concrete under sustained uniaxial com- Xing, F., Lemg, F. G., and Feng, N. Q. (2005). “The influence of long-term
pression.” ACI Special Publication 221, American Concrete Institute, load on the chloride permeability in reinforced concrete.” Proc. Int.
Farmington Hills, MI, 893–910. Workshop Durability of Reinforced Concrete under Combined
Mangat, P. S., and Molloy, B. T. (1994). “Prediction of long term chloride Mechanical and Climate Loads (CMCL), T. Zhao, F. H. Wittmann,
concentration in concrete.” Mater. Struct., 27(6), 338–346. and T. Ueda, eds., Adeificatio Publishers, Freiburg, Germany, 139–147.
Martin-Perez, B., Zibara, H., Hooton, R. D., and Thomas, M. D. A. (2000). Yuan, Q., Shi, C., De Schutter, G., Deng, D., and He, F. (2011). “Numerical
“A study of the effect of chloride binding on service life prediction.” model for chloride penetration into saturated concrete.” J. Mater. Civ.
Cem. Concr. Res., 30(8), 1215–1223. Eng., 10.1061/(ASCE)MT.1943-5533.0000168, 305–311.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / APRIL 2014 / 667

View publication stats J. Mater. Civ. Eng. 2014.26:658-667.